TP53 gene mutations and protein accumulation in primary vaginal carcinomas.

Primary carcinomas from 46 patients were screened for TP53 alterations. Immunohistochemistry demonstrated nuclear TP53 protein accumulation in 22 (48%) cases using the polyclonal CM1 antiserum, whereas 15 (33%) cases showed positive nuclear staining with the mononuclear antibody PAb 1801. Constant denaturant gel electrophoresis (CDGE) was used to screen 27 of the vaginal carcinomas for mutations in the conserved regions of the TP53 gene (exons 5-8). Six of these tumours (22%) contained mutations: four were found in exon 5 and two in exon 8. A total of 50% of the primary vaginal carcinomas carried a TP53 alteration. These results indicate that TP53 abnormalities may be involved in the development of these tumours. However, there was no significant association between TP53 abnormalities and survival.

The TP53 tumour-suppressor gene, which encodes a 53 kDa cell cycle regulator nuclear phosphoprotein, is located on the short arm of chromosome 17. The product of this gene has been implicated in the control of the cell cycle, DNA repair and synthesis, cell differentiation and programmed cell death (Harris and Hollstein, 1993). Some mutant forms of the gene can act as dominant oncogenes, whereas wild-type TP53 has characteristics of a recessive tumour-suppressor gene (Lane and Benchimol, 1990). Although the precise mechanism by which the TP53 protein participates in these cellular functions is not fully understood, several biochemical features of TP53 have been elucidated. The TP53 protein is able to regulate transcription directly (Kern et al., 1991;El-Deiry et al., 1992) or by interacting with other transcriptional regulatory factors, such as the TATA-and CAAT-binding proteins (Seto et al., 1992;Agoff et al., 1993). The TP53 protein has also been shown to act as a specific transcription factor controlling the expression of growth arrest genes such as GADD45 (Kastan et al., 1992) and WAF-I (El-Deiry et al., 1993).
Mutations in the TP53 gene are the most frequent genetic alteration found in human tumours (Hollstein et al., 1994;Levine et al., 1994). The ability to transactivate gene expression from a specific promoter sequence is lost in most TP53 mutants associated with cell transformation and oncogenesis (Kern et al., 1991). The mutations are usually missense and are frequently accompanied by loss of the remaining normal allele. Furthermore, mutated proteins are able to bind and activate wild-type TP53 protein by forming oligomeric complexes (Nigro et al., 1989). Point mutations in the TP53 gene often result in increased stability of the mutant protein (Finlay et al., 1988), which can be detected by immunohistochemistry, whereas the wild-type TP53 protein is undetectable because of its short half-life (Gronstajaski et al., 1984). However, under certain circumstances accumulation of wildtype TP53 protein may occur, probably because of complex formation with other cellular proteins such as MDM2 , or virus proteins like the large T antigen of SV40 (Lane and Crawford, 1979).
There is increasing evidence that tumour-suppressor genes are involved in the development and/or progression of gynaecological malignancies. Mutations or loss of heterozygosity at the TP53 locus have been reported in ovarian (Bosari et al., 1993;Kohler et al., 1993a) (Kohler et al., 1993b;Yu et al., 1993) and cervical cancers (Helland et al., 1993;Holm et al., 1993). To our knowledge abnormalities of the TP53 tumour-suppressor gene have not been studied in vaginal carcinomas. Among gynaecological cancers, carcinoma of the vagina is relatively rare. It accounts for only 1-2% of all gynaecological cancers (Pride et al., 1979).
The aims of the present study were to determine the frequency of TP53 protein accumulation and TP53 mutations in a series of primary vaginal carcinomas. Furthermore, we wanted to correlate TP53 alterations with histopathological and clinical parameters and to evaluate whether these alterations provide prognostic information in vaginal carcinomas.

Twnour samples
Forty-six cases of vaginal carcinomas, diagnosed in the period 1973-94, were collected from the files of the Norwegian Radium Hospital. The mean age at diagnosis was 66 years (range 31-87 years). The median observation time of the patients still living was 43 months (range 1-214 months). Histopathological and clinical diagnoses are shown in Table I. Immediately after surgery the tissue was fixed in 10% formalin, embedded in paraffin and processed for light microscopy. Three of the tumour samples were also frozen and stored in liquid nitrogen for DNA analyses. Haematoxylin-eosin-stained sections were used to evaluate the approx-  Mies et al. (1991). Five to ten 10 jim tissue sections of the paraffin-embedded samples were collcted in a 2 ml Eppendorf tube, deparAffinised with Hioclear (Histolab, Sweden) and rinsed in 100% alcohol. The deparaffiised sections and crushed frozen tissue were did with proteinas K (Sigma, USA) at a final concentration of 0.5 mg ml-' in 0.05 M Tris-HCI buffer containing 0.15M sodium chbride, 5mM EDTA and 1% SDS (pH9.0). Digestion was performed at 55 C for 3-7 days. Protein was removed by phenol-chloroform extraction and DNA isolated by ethanol prcipitation. Samples were handled carefully to inimi mechanical stress, and wide-bore pipettes were used to transfer aqueous solutions containing high molecular weight DNA.
The four amplified products from each tumour were screened for TP53 mutations using constant denaturant gel eectrophoresis (CDGE) (Borresen et al., 1991;Hovig et al., 1991). Denaturing gels contained 12% acrylamide with varyig denaturant concentrations consisting of urea and formamide (fragment A, 45.5% and 55%; frgment B, 55%; fragment C, 49.5%; and fragment D, 49.5%; 100% denaturant corresponds to 7 M urea and 40% formamide). Gels were run submerged in TAE buffer (40mM Trisacetate, I mM EDTA, pH 8.0) at 56-C at 80 V for 3-4 h. After electrophoresis, gels were stained for a few minutes in ethidium bromide (2 mg 1-' TAE) and photogaphed ng a UV nsilluminator. Samps showing aberrant migration in CDGE were reamplified, and to confirm a true mutant denaturing gradiet edtrophoresis (DGGE) (Borfesen et al., 1991) was performed. The gadient gels were cast with a gravitational gradient mixer. The PCR product was loaded into a long well on top of the gel and run with the detrophoresis direction pependicular to the denaturant gradienLt The gels, which had the same chemicals and ekctrophoresis conditions as the constant denaturing gels, were run for 2 h with the gradient spanning from 10% to 70% denaturant. Gradient gels were stained and photographed usng both ethidium bromide, as described above, and SYBR green I (Molkuar Probes, Eugene, OR, USA) diluted 1:10000. Samples that carried a mutation were amplified with one biotinylated primer. The PCR products were sequenced with dideoxy sequencing reactions using Dynabeads M280-Strptavidin (DynaL Norway) as solid support (Hultman et at., 1989). Oligonucleotides flaning each of exons 5, 7 and 8 were used to prime the reactions, which were performed by first heating the primer-template mix (70 C). Then the samples were labeLlld for 10 min with [3SJdCTP, and the termination reactions were run with Sequenase 2.0 17 DNA polymerase (US Biochemicals) at 3TC for 10 min. The reaction products were electrophoresed on a 4.3% polyacrylamide gel, which was dried and autoradiographed with Kodak Hyperfilm-MP beta-max overnight.

Immunohistochemistry
Formalin-fixed paraffin-embedded tissue specimens from 46 cases were used for munhistochemical staining with the avidin-biotin-peroxidase complex (ABC) method (Hsu et al., 1981). Deparaffinised sections were treated with 0.3% hydrogen peroxide in methanol for 30 min to block endogenous peroxidase. The sections were incubated for 20min with normal serum from the species in which the secondary antibody was made. This was done to eliminate non-specific staining. Excess normal serum was blotted from the slides before incubation with a polyclonal TP53 antiserm (NCL-CM1, Novocastra Laboratory, UK) diluted 1:300 and a monoclonal TP53 antibody (PAb 1801, Oncogene Science, NY, USA) diluted 1:100 (1 Mg of IgG, per ml) for 18-22 h at 4C. Both antibodies detete mutant and wild-type TP53. The sections were then incubated with a 1:200 dilution of biotin-labeled secondary antibody for 30min and ABC (10 igml-' avidin and 2.4 jgml ' biotinlabelled peroxidase) (Vector, Burlingame, CA, USA) for 60min. The tissue was stained for 5min with 0.05% 3,3'diaminobenzidine tetrahydrochloride (DAB) freshly prepred in 0.05 M Tris buffer at pH 7.6 containing 0.01% hydrogen peroxide. Sections were then countetained with haematoxyfin, dehydrated and mounted in Diatex. All the dilutions of normal sera, antisera, biotin-labeiled secondary antibodies and ABC were done with phosphate-buffered saline (PBS), pH 7.4, containing 5% bovine serum a in All series incuded positive controls Negative controls included replcement of polyclonal prmary antisrum with rabbit serun diluted 1:300, whereas negative controls for the monoclonal antibody were performed usng mouse myeloma protein of the same subclass and concentration as the monoclonal antibody. All controls gave satisfactory results.

Statistical analysis
Differences in proportion were evaluated by the chi-square test Cancer-related survival was lalcted from start of treatment to death of diseas, or 31 May 1994, usng the method of Kaplan and Meier (1958). Diffaences in surival were evaluated usng the log-rani test (Tarone and Ware, 1977). A P-ievel less than 0.05 was conidere atistically signifiCant.
Rests TP53 mutation analysis Of the 46 samples that were subjected to immunotaining, sUffiCient material for CDGE and DGGE analyses were available for 27 samples. Six (22%) of these tumours contained mutations: four were identified in exon 5 and two in exon 8 (Table II, Figure la and   --ND -, No immunoreactive cells or mutation not detected; +, < 5% cells with immunoreactive nucleus; + +, 5 -50% cells with immunoreactive nucleus; + + +, > 50% cels with immunoreactive nucleus. ND, not done (suitable material not available). 'Mutation detected by abberant migrating bands in CDGE and DGGE. the polyclonal CM1 antiserum (Table II, Figure 2), whereas 15 of 46 (33%) cases showed positive staining with the monoclonal antibody PAb 1801 (Table II). The cases that were positive with the monoclonal antibody were all imunoreactive with the polyclonal antiserum. The TP53 protein-immunopositive cases exhibited granular or diffuise nuclear staining, and the proportion of immunoreactive cells varied between tumours (Table II). No positive staining was observed in normal tissues adjacent to tumours.
Correlation between mutation and immunohistochemical data The concordance between mutation and immunohistochemical data from samples that were subjected to both types of analyses was 70% when both positive and negative results were taken into consideration. In 14 tumours neither TP53 mutation nor TP53 protein accumulation was observed, whereas five cases exhibited both TP53 protein accumulation and mutations. One of the mutated tumours did not exhibit an elevated level of TP53 protein. In seven of the tumours that were TP53 protein positive a mutation was not identified. Of these seven cases, three accumulated TP53 protein in less than 5% of the tumour cells. In total, 50% of primary vaginal carcinomas carried TP53 alterations.

TP53 alterations and clinical parameters
The frequency of TP53 mutations and protein overexpression seemed to increase with increasing FIGO stage. However, the difference was not statistically significant (P = 0.41). There were no differences between the TP53 mutant/immunohistochemical positive and negative cases regarding histological type or grade of differentiation (Table I). The 5 year cancerrelated survival in the two groups was 47% and 42% respectively (P = 0.95).
Among gynaecological cancers, carcinoma of the vagina is relatively rare, accounting for 1-2% of all gynaecological studied in these malignncies. The present study demonstrated TP53 protein accmulation in 22 of 46 (48%) cases and mutations in six of 27 (22%) cases. A total of 50% of the primary vaginal caranomas showed TP53 gene mutation, and/or protein accumulation. The rates of TP53 alterations in these vaginal carcinomas are similr to what has been observed in other gynaecological cancers. TP53 alterations are found in 55% of ovarian cancers (Bosari et al., 1993;Marks et al., 1993), 59% of endometrial cacrs (Bur et al., 1992) and 62% of cervical cancers (Hohm et al., 1993). Our results indicate that TP53 abnormalities also may be involved in the development of vaginal carcinomas.
To our knowledge, no study has investigated the relationship between HPV DNA and TP53 alterations in vaginal carcinomas. Previously, an inverse relationship between the presence of HPV DNA and TP53 gene mutation in cell lines (Sheffner et al., 1991) and in primary cervical carcinomas has been demonstrated (Crook et al., 1992), whereas others have identified TP53 alteration and HPV DNA in the same cases of cervical carcinomas (Busby-Earle et al., 1992;Helland et al., 1993). In vaginal carcinomas we did not find an inverse relationship between HPV DNA and TP53 alterations. In 11 of 14 cases with TP53 alteration, HPV 16 was also detected (unpublished findings).
In the present study there was a 70% correlation between mutation and immunohistochemical data. This is in contrast to earlier studies in which a highly signnt association between the presence of TP53 mutations and TP53 protein accumulation was observed (Andersen et al., 1993;Marchetti et al., 1993). In previous studies, an increasing number of tumours with TP53 protein accumulation without altered DNA have been found (Helland et al., 1993;Lae, 1994). Seven of the tumours that were TP53 protein positive by immunohistochemistry were not found to be mutated. Of these seven cases, three showed a very small fraction of TP53 protein-positive cells. Therefore, the number of mutated cells may have been too low to be de by CDGE, although as few as 1-5% of mutated cells could be detected by this method (Andersen and Borresen, 1995). Furthermore, some of the tumours could have mutations outside the four screened regions of the gene, or have alterations in the TP53 regulator sequences. Another explanation is that the tumours may have accumulated wild-type TP53 protein. It has recently been shown that cell stress resulting from external DNA-damaging agents can lead to accumulation of wild-type TP53 protein in normal skcin (Hall et al., 1993;Fritsche et al., 1993). Nevertheess, this phenomenon is unlikely, as we never found TP53 protein in normal tissues surrounding the tumours. However, internal DNA damage could be limited to the tumour cells. Alternatively, wild-type TP53 protein may have formed complexes with other proteins such as MDM2, resulting in a higher level of inactive TP53 protein (Oliner et al., 1992). In one of the mutated samples we failed to find a positive immunoreaction. A mutation implying a shift in reading frame or a stop codon could lead to a change in a large proportion of the quatemary structure of the protein which would result in the absnce of TP53 immunostaining (Andersen and Borresen, 1995). Lack of immunostaining could also be explained by the presence of a sense mutation that does not stabilise the protein sufficiently.
CDGE and DGGE analysis identified TP53 gene mutations in six of 27 (22%) vaginal carcinomas. We were able to determine the exact nature of the mutation by direct sequencing only in the DNA extracted from the fresh-frozen tissue. This method required more template DNA and PCR product than CDGE and DGGE analysis. The sequencing prims were different from the CDGE and DGGE primers. It is therefore possible that formalin fixation had degraded or modified the DNA in a way that disturbed anling of the sequencng primers or, alternatively, the chain elongation. The amount of tissue from each sample was scarce, and thus sequencing would most likely be more successful when done on a largr amount of tissue. Previously, the CDGE/DGGE technique has proved to be highly rehable, with a detection rate of 100% of mutants in exon 5, 7 and 8 under optimal conditions (Condie, 1993). All samples with aberrant migrating bands in DGGE displayed heteroduplex formation. The heteroduplexes that are easily recognised in melting gels enable detection of mutations when present in 1-5% of the cell population (Andersen and Borresen, 1995).
Numerous antibodies detecting TP53 protein are available. We observed positive immunoreactivity more often with the polydonal antiserum CM1 than with the monoclonal antibody PAb 1801. This discrepancy may be due to aceumulation of TP53 protein with a configuration detted by the polyclonal but not the monoclonal antibody. In addition, the failure of PAb 1801 to detect TP53 protein in some specmens immunoreactive with CM1 could be because the epitope recognised by PAb 1801 antibody is not stable in formalin-fixed tissu (Purdie et al., 1991).
No correlation was ssen between TP53 alteration and survival of patients with vaginal carcinomas. This is in agreement with other studies that fail to find prognostic significae in cancers of the ovary (Marks et al., 1992;Kohler et al., 1993a) and cervix (Helland et al., 1993;Oka et a!., 1993). In contrast, other groups of investigators observed a relationship between behaviour and TP53 alteration in cancers of the ovary (Bosari et al., 1993) and endometrium (Bur et al., 1992). Our study inluded a limited number of cases, and therefore further studies of a largr amount of material are needed to better define the prognostic significae of TP53 alterations in patients with vaginal cancer.
In conclusion, TP53 alterations were dedtet in 50% of primary vaginal carcinomas by use of genetic and immunohistochemical techniques. These results indicate that TP53 abnormalities may be involved in the development of these tumours. However, there was no signnt correlation between TP53 alteration and survival.