Dominant-Negative Mutations of the Tumor Suppressor p53 Relating to Early Onset of Glioblastoma Multiforme¹

More than half of all human cancers are associated with alterations of the p53 tumor suppressor gene (1, 2, 3). Most p53 alterations are missense mutations localized in the DNA-binding domain, mostly abolishing the sequence-specific transcriptional activity of the product (4, 5, 6). Previous experiments have suggested that some mutant forms of p53 are able to inactivate the endogenous wild-type p53 protein in a dominant-negative fashion by forming a heterotetramer complex (1, 7, 8, 9). These findings suggest that dominant-negative mutations in one p53 allele should be sufficient to inactivate p53 function in a cell. Development of a tumor endures a long period of coexistence of wild-type and mutant protein in the same cell, either following an initial somatic mutation or from birth of the host, in the case of a germ-line mutation (10, 11, 12). If a certain mutant p53 exerts a dominant-negative effect in this situation, it should result within a short period in cancer development, compared with usual recessive mutants. This hypothesis, however, is grounded on the limited analyses of only a few p53 mutants, and their general relevance remains controversial. In addition, the extent to which p53 mutants abrogate wild-type p53 function in the heterozygous state depends on the particular assay used (6, 8, 13, 14, 15, 16, 17). For these reasons, contribution of dominant-negative p53 mutants to clinical outcome is yet to be analyzed.

The methods using yeast have received much attention recently as a useful tool for determining the dominant-negative property of p53 mutants in sequence-specific transactivation (18, 19). These yeast-based functional assays have the advantage in their equimolar expression of both wild-type and mutant p53, because of the centromere/autonomous replication sequence (CEN/ARS) in the expression vector, which assures single-copy-maintenance of the plasmid in yeast. These experiments made it clearer that dominant-negative mutants have alterations in the amino acids that are essential for stabilization of the DNA-binding surface in the p53 core domain or for direct interaction of p53 with its DNA-binding sites. Because the most common p53 mutations found in human tumors are located in the DNA-binding sites, called “hot spots,” transdominant p53 mutations should not be uncommon in human tumors. The intriguing question of whether tumors with such dominant-negative mutants actually have biological and/or clinical features different from those with recessive p53 mutants remains unanswered.

Using a yeast-based functional assay (20), we have identified various p53 mutations in brain tumors (21, 22), breast cancers, and SCCs³ in oral epithelium (23). To investigate whether the dominant-negative p53 mutant has clinical significance, we developed a yeast-based assay for assessment of dominant-negative potential of p53 mutants and examined the dominant-negative potential of the various p53 mutants that have been identified in our previous studies. On the basis of the results of the transdominance test, we demonstrate that dominant-negative p53 mutants are related to earlier onset of brain tumors.

With the use of yeast p53 functional assay, somatic p53 mutations were detected in 29 patients with malignant brain tumors (17 with GBM, 11 with anaplastic astrocytoma, and 1 with lymphoma; Refs. 21, 22, and 24), 29 patients with breast cancer,⁴ 14 patients with oral SCC (10 cases) or precancerous leukoplakia (4 cases; Ref. 23), and 5 cultured cell lines derived from GBM. Seventy-five p53 mutants recombined in plasmid pSS16 were isolated from the red colonies given by the yeast functional assay and then subjected to transdominance test described below.

In addition to the mutations identified in the tumors or cell lines, 31 p53 mutants artificially introduced by Taq polymerase amplification were selected by yeast functional assay. Identifications of mutant p53 in the germ line in the literature were performed by the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA).

The yeast strain yIG397 was used in this study (genotype: MATa ade2-1 leu2-3, 112 trp1-1 his3-11, 15 can1-100 ura3-1::[URA3 3xRGC-pCYC1-ADE2]). The strain yIG397 contains an integrated plasmid with the ADE2 open reading frame under the control of a p53-responsive promoter. When the strain is transformed with an expression vector encoding wild-type p53, the cells express ADE2 and form white colonies; otherwise, the cells fail to express ADE2 and form red colonies because of the accumulation of an intermediate in the adenine metabolism (25).

The wild-type p53 expression vector pLS72 (26) was digested with EcoRI-HpaI to release a 1.1-kb fragment containing LEU2 marker, end-blunted, and ligated with 1.7-kb NaeI-PvuII fragment of pGBT9 (Clontech, Palo Alto, CA) containing TRP1 marker to give pTSHP53.

Mutant p53 expression vectors were rescued from red colonies given by yeast functional assay, as described previously (21, 23). Because these mutant plasmids were reconstructed in yeast by homologous recombination between reverse transcription-PCR product and linearized pSS16, which was derived from pLS72, they are identical with pTSHP53, except for the p53 mutation and nutrient marker. To confirm mutations, we then transfected the pSS16 containing a mutant p53 into XL-1 blue Escherichia coli by electroporation. The plasmids were recovered, purified, and sequenced with Dye-Terminator sequencing kit on an ABI 373A automated sequencer (Applied Biosystems, Urayasu, Japan).

Yeast were cultured in YPD medium supplemented with an excessive amount of adenine (200 μg/ml) until A_{600 nm} reached 0.8. Then the yeast were washed in LiOAc solution containing 0.1m lithium acetate, 10 mm Tris-HCl (pH 8.0), and 1 mm Na₂-EDTA, pelleted again, and resuspended in an equal volume of LiOAc solution. For each transformation, 50 μl of yeast suspension were mixed with 100 ng of pTSHP53, 100 ng of mutant p53-containing pSS16 (which had been sequence-verified), 50 μg of sonicated single-stranded salmon sperm DNA, and 300 μl of LiOAc containing 40% of polyethylene glycol 4000 (Kanto Chemical, Tokyo, Japan). The mixture was incubated at 30°C for 30 min and heat-shocked at 42°C for 15 min. Yeast were then plated on synthetic dropout (SD) medium minus leucine and tryptophan but containing a limited amount of adenine (5 μg/ml) and incubated for 48 h in a 30°C humidified atmosphere. Double-transformant clones (Leu+, Trp+) giving rise to white (Ade+) or pink/red (Ade−) colonies were interpreted as expressing recessive and dominant-negative mutations, respectively (Figs. 1 and 2).

At 48 h after incubation on the selection medium subsequent to the transformation with wild-type and mutant p53 expression vectors, yeast colonies were inoculated to YPD medium supplemented with a sufficient adenine (200 mg/liter) and cultured for 16–24 h. The yeast were pelleted by centrifugation and resuspended in yeast lysis buffer [50 mm Tris (pH 8.0), 0.1% Triton X-100, and 0.5% SDS] supplemented with protease inhibitors (10 mg/ml leupeptin, 1 mg/ml pepstatin A, and 174 mg/ml phenylmethylsulfonyl fluoride). The yeast protein was extracted by the glass beads method according to the standard protocol (27). Protein concentration was adjusted to 1 μg/μl. Yeast lysates were electrophoresed in 10% SDS-PAGE and transferred to a nylon membrane (Millipore). The membranes were first incubated with blocking buffer (TBST containing 3% BSA fraction V) overnight at 4°C, washed three times in 20 ml of TBST [20 mm Tris (pH 7.5), 137 mm NaCl, and 0.1% Tween 20], and then reacted with a p53-specific monoclonal antibody PAb1801 (Calbiochem) for 1 h at room temperature. After washing three times in 20 ml of TBST, the membranes were incubated with a sheep antimouse IgG antibody labeled with horseradish peroxidase for 1 h at room temperature. After washing three times in TBST, the membranes were developed using an ECL chemiluminescence detection kit (Amersham).

To ascertain whether both wild-type and mutant p53 are expressed equally in yeast, we cotransfected yeast with the wild-type p53 plasmid (TRP1 marker) and mutant p53 plasmid (LEU2 marker) with a nonsense mutation at the 3′ end. The two mutant plasmids used in this experiment, pLSΔ51 and pLSΔ57, have nonsense mutations at codons 342 and 336, resulting in truncated proteins lacking 51 and 57 COOH-terminal amino acids, respectively. Yeast cotransfected with pTSHP53 and either pLSΔ51 or pLSΔ57 formed white colonies on a minimal plate supplemented with 5 μg/ml adenine but lacking tryptophan and leucine. An anti-p53 antibody, PAb1801, was used to detect the expression of wild-type and mutant p53 protein on immunoblots of the yeast lysate. Both wild-type and truncated p53 were detected (Fig. 3).

A total of 76 p53 mutants identified in brain tumors, GBM-derived cell lines, breast cancers, and premalignant lesions and SCC of oral epithelium were tested by the transdominance assay. Of 30 mutations in brain tumors, 9 gave red or pink colonies (positive), and 21 gave white colonies (negative). Likewise, in breast cancers, 6 mutations were positive and 23 were negative, and in oral SCC and premalignant lesions, 6 were positive and 8 were negative. Five mutants identified in GBM cell lines all gave white colonies. In addition to these naturally occurring mutations, 28 mutations artificially introduced with Taq polymerase amplification were also analyzed. A summary of the data from transdominance test is shown in Tables 1 and 2.

Of the 42 patients with GBM (22), 1 patient was excluded because the tumor exhibited compound (recessive 158His and dominant-negative 248Gln) heterozygous mutants. The age at diagnosis was compared among those with dominant-negative p53 mutations (n = 7) and recessive mutations (n = 9) and those without mutations (n = 24). The average age at diagnosis was 30.4 ± 14.7 years in patients with dominant-negative mutations, which was significantly younger than that of the patients with recessive mutations (55.2 ± 18.6 years, P < 0.012) and those without mutations (54.7 ± 17.1 years, P < 0.003). Even if the excluded 27-year-old patient with both a recessive and transdominant mutations is included in either group, difference of the ages between patients with dominant-negative mutations versus patients with recessive mutations was statistically significant (P < 0.007 or P < 0.025).

In patients with anaplastic astrocytoma with p53 mutations, the age at diagnosis was younger compared with those without p53 mutation, regardless of its transdominance (33.2 ± 12.2 years), although significant difference was not observed due to the small numbers of patients. On the other hand, ages at diagnosis were not significantly different among breast cancer patients (49.5 ± 9.8 years, n = 6 versus 50.3 ± 10.3 years, n = 23) or oral cancer patients (64.7 ± 11.5 years, n = 6 versus 57.0 ± 13.9 years, n = 8) for patients with dominant-negative versus recessive p53 mutants, respectively.

In this study, we examined dominant-negative potential of various p53 mutants with regard to DNA binding and transcriptional activity, using a yeast-based assay. p53 mutations with amino acid substitution at the DNA binding domain showed dominant-negative potential over the coexpressed wild-type p53. Furthermore, we demonstrated that, in patients with GBM, the average age at diagnosis was significantly younger in those with dominant-negative p53 mutations than it was in those with recessive mutations or without p53 mutations.

p53 is active as a tetramer (28). Most of the p53 mutations in human cancers are found outside the oligomerization domain, hence sparing the function of tetramerization. Experimental evidence showed that wild-type p53 tetramerizes with mutant proteins to form a heterotetramer (18, 29). One of the main problems in the study on the biological interaction between mutant p53 and wild-type p53 has resided in the difficulty in expressing both RNAs at the same level. Most of the previous experiments were performed under conditions in which mutant p53 was overexpressed compared with the wild-type p53, which is totally different from the in vivo situation, in which both mutant and wild-type alleles are equally transcribed (11, 26). To reproduce this in vivo condition in yeast, we have used two vectors, one for expression of a wild-type p53 and the other for a mutant p53. Because the two vectors are identical except for the selection markers and have the same promoter and centromere/autonomous replication sequence (CEN/ARS) of the yeast chromosome, equal expression of mutant and wild-type p53 is warranted at least in mRNA level. Western blot analysis of the yeast lysates clearly showed that the yeast permitted efficient expression of both alleles.

Even in this situation, however, assessment of dominant-negative potential of the mutant p53 might depend on specificity in the binding to the p53-responsive element in the promoter and sensitivity of the ADE2 reporter gene assay. Thus, we have compared our results with the previous ones in which a yeast-based assay was used. Among 27 mutants for which dominant-negative potential had been evaluated by a similar assay (19), 9 mutants were further evaluated in this study, and the results were the same. Our results are also generally consistent with those obtained by a URA3-based yeast assay (18). In the URA3 selection method, the dominant-negative effect was classified into two categories: class 1, which is able to interfere with two copies of wild-type p53, and class 2, which can only override the activity of a single wild-type allele. The class 1 mutants were assessed as positive in our ADE2-based method, whereas class 2 mutants were scored negative in our assay. All of the transdominant mutations demonstrated by our assay are considered to affect the amino acids that are essential to the maintenance of DNA-binding surface, according to the crystallographic structure reported by Cho et al. (30). This is consistent with the hypothesis that the transdominant effect is due to alteration of the whole structure constituted by the DNA-binding surfaces of four p53 molecules (15).

Our results demonstrated that dominant-negative p53 mutations are associated with early onset of GBM. This is consistent with an observation in families carrying germ-line mutations of p53 gene, in which families with missense mutations in the core DNA-binding domain showed a cancer phenotype characterized by a higher incidence and earlier onset compared with those carrying protein-truncating or other inactivating mutations (31). Indeed, 9 of 12 mutations reported by them were shown as dominant-negative by this study. The most plausible explanation for the early onset of glioblastomas harboring dominant-negative mutations is that, in glial cells harboring heterozygous recessive p53 mutation, wild-type p53 is phenotypically dominant to mutated p53 and suppresses the neoplastic phenotype, but dominant-negative mutant abolishes the cooperativity of wild-type p53 with p53-inducible downstream effector genes, such as cell cycle regulatory and apoptotic genes. Because gliomas originate from glial cells that complete their division during early lifetime, most somatic mutations in glial cells are likely to be introduced in the limited time of life. Provided with an equal start time, cells with dominant-negative mutation have the advantage of cellular proliferation (32).

Unlike the situation in glioblastomas, dominant-negative p53 mutations did not give a difference in age at diagnosis in the patients with breast cancers or oral cancers. This may be because: (a) occurrence of somatic mutations in mammary gland cells and oral epithelial cells is not restricted to a specific period of life and, therefore, results in diverse ages of onset; and/or (b) p53 mutation is not a primary factor essential for initiation of tumor development but rather for tumor progression. Because germ-line p53 mutation predisposes breast cancers to develop at an early age (33), the first reason is most likely for breast cancers. In oral cancers, however, because topical exposure of carcinogens is not a time-restricted event, the second reason appears to be probable.

The late onset of the glioblastomas which do not harbor p53 mutation suggests that an alternative carcinogenetic pathway without p53 mutation requires accumulation of genetic events that takes a long period of time. Such genetic alterations include mutations in PTEN gene, loss of the p16 gene, and amplification of epidermal growth factor receptors (34). This pathway has been advocated to be characterized by de novo occurrence of glioblastomas without preceding more benign lesions as astrocytomas (35, 36).

Certain p53 mutants may act in a dominant-negative fashion not only to the wild-type p53 but also to the newly identified p53 homologue, p73 (37, 38). Restricted-tissue specific-expression pattern and monoallelic expression of p73 in cancers including neuroblastoma (38) suggest that the protein has a potential role as a tumor suppressor. One study indicates that expression of p73 gene is altered in glial origin tumors, i.e., low in medulloblastomas and glioblastomas but high in ependymomas (39). It has been shown that p53 mutants (R175H and R248W) reduced the transcriptional activity of p73 (40). Taking these findings into account, it is possible that some dominant-negative p53 mutants reduce gene-specific transactivating function of p73 or other yet unknown tumor suppressor proteins, thereby causing progression of tumors such as glioblastomas.

We thank M. Yanome for her help in preparing this manuscript.