Loss of heterozygosity (LOH) atp53 is correlated with LOH atBRCA1 andBRCA2 in various human malignant tumors

Mutations in BRCA1 and BRCA2 predispose to breast and ovarian cancers. Several studies have demonstrated that BRCA1 and BRCA2 share similarities not only in their biochemical properties but also in the biochemical pathways by which they are regulated. The protein products of both BRCA1 and BRCA2 genes have transcription-activation function (Chapman and Verma, 1996; Milner et al., 1997). Both bind to Rad51, a protein involved in maintaining the integrity of the genome (Scully et al., 1997; Sharan et al., 1997). Frequent LOH of both BRCA1 and BRCA2 genes was observed in breast and ovarian cancers. Furthermore, concordant loss of both genes was also reported (Kelsell et al., 1996; Silva et al., 1999). The combined loss of BRCA1 and BRCA2 leads to the hypothesis that both genes may be induced by or function in overlapping regulatory pathways involved in tumorigenesis. Rajan et al. (1996) demonstrated that BRCA2 mRNA expression was tightly regulated during mammary epithelial proliferation and differentiation and that this regulation occurred coordinately with BRCA1. It has also been demonstrated that the mRNA levels of BRCA1 and BRCA2 were coordinately elevated by the steroid hormone estradiol in breast cancer cell lines (Spillman and Bowcock, 1996). In the ovary, BRCA1 and BRCA2 exhibited a comparable hormone-independent pattern of expression in oocytes, granulosa cells and thecal cells of developing follicles (Blackshear et al., 1998). In addition, the products of both genes form a stable interaction in both mitotic and meiotic cells (Chen et al., 1998). BRCA genes are referred to as caretakers rather than gatekeepers. Inactivation of a caretaker gene does not promote tumor initiation directly; it rather leads to genetic instabilities that result in an increased mutation rate affecting all genes including gatekeepers, which directly regulate the growth of tumors by inhibiting growth or promoting death (Kinzler and Vogelstein, 1997). High frequencies of p53 mutations were found in both breast and ovarian cancers (Hollstein et al., 1996; Kupryjanczyk et al., 1993). Somatic mutations of p53 are common in association with germline mutations of BRCA1 and BRCA2 (Rhei et al., 1998; Crook et al., 1997; Gretarsdottir et al., 1998). In breast cancer, a higher frequency of LOH at 17q and 13q was found in association with p53 gene alterations. The increased LOH rate in relation to p53 abnormalities was statistically significant for the BRCA1, but not for the BRCA2, locus (Tseng et al., 1997). According to the tumor-suppressor hypothesis, LOH of 1 allele is frequently accompanied by mutational or regulatory inactivation of the remaining allele and, thus, leads to complete inactivation of a gene (Knudson, 1971). To find out whether p53 inactivation is specifically correlated with that of BRCA1 and BRCA2 in various tumor types of epithelial origin, we examined LOH of these 3 genes in 193 breast carcinomas, 66 ovarian carcinomas, 30 endometrial carcinomas and 39 head-and-neck carcinomas. Fresh biopsies from 193 sporadic breast carcinomas, 66 sporadic ovarian carcinomas and 30 sporadic endometrial carcinomas together with the corresponding EDTA blood samples were collected in the Department of Obstetrics and Gynecology, University of Vienna, from 1994 to 1998. Thirty-nine tumor biopsies from head-and-neck squamous-cell carcinomas and the corresponding EDTA blood samples were collected in the department of Otolaryngology, University of Vienna, at the same time. The histopathologic characteristics of the tumors are shown in Table I. Tumor samples were examined for contamination of non-tumorous portions. Only samples consisting of >90% tumor cells were analyzed. DNA was isolated from blood or tumor tissue using commercially available kits (DNA Extraction System I and II; ViennaLab, Vienna, Austria). Three microsatellite markers, D17S1322, D17S1323 and D17S855, which are located in introns 12, 19 and 20 of BRCA1, respectively, together with the marker D17S1327, which lies adjacent to BRCA1, were used for the examination of LOH at the BRCA1 locus. Two markers, D13S260 and D13S267, which flank the BRCA2 gene, were used to detect LOH at the BRCA2 locus. Marker TP53 was used to detect LOH at p53, and D8S535 was used to detect LOH at chromosome 8p11-p21. DNA (50 ng) was used as template for the amplification. The reaction mixture included 25 pmol of both sense and anti-sense primers, 250 μM dNTPs (ViennaLab), 5 μl of 10× amplification buffer [10 mM Tris-HCl (pH 9.0), 50 mM KCl, 0.01% (w/v) gelatin, 1.5 mM MgCl2, 0.1% Triton X-100; ViennaLab) and 1.6 units of Super Taq Polymerase (HT Biotechnology, Cambridge, UK). PCR was performed on a Perkin-Elmer (Foster City, CA) GeneAmp PCR system 9600 with 35 cycles at 94°C for 30 sec, the corresponding annealing temperature for 30 sec and 72°C for 30 sec. Annealing temperatures were 65°C, 55°C and 50°C for amplification of TP53, D8S535 and all BRCA1 and BRCA2 markers, respectively. The reaction was preceded by a primary denaturation step at 94°C for 1 min. Sense primers were labeled with Cy5. PCR products were resolved on a 6% polyacrylamide gel using an automated laser fluorescent sequencer (ALF Express DNA Sequencer; Pharmacia, Uppsala, Sweden). Sizes of the fragments were calculated using an external standard and the Fragment Manager software (Pharmacia). The ratio of relative allelic peak intensities in the tumor DNA to that in the corresponding normal DNA was calculated. An allelic loss was considered significant when this ratio was <0.67 or >1.5. LOH affecting BRCA1 was defined when at least 1 of the markers used for each locus showed allelic loss. LOH affecting BRCA2 was defined when both markers showed allelic loss. The χ2 test was used to examine the correlation of LOH at p53, BRCA1 and BRCA2 loci. p < 0.05 was considered statistically significant. LOH at p53 was found in 23.5% (n = 132 informative cases), 63.5% (n = 52), 26.9% (n = 26) and 64.9% (n = 37) of breast, ovarian, endometrial and head-and-neck carcinomas, respectively. LOH at BRCA1 was found in 21.7% (n = 157), 56.3% (n = 64), 24.1% (n = 29) and 22.2% (n = 36) of breast, ovarian, endometrial and head-and-neck carcinomas, respectively. LOH at BRCA2 was found in 26.2% (n = 183), 57.1% (n = 63), 28.6% (n = 28) and 57.9% (n = 38) of breast, ovarian, endometrial and head-and-neck carcinomas, respectively. Statistical analysis revealed that LOH at p53 was significantly correlated with LOH at BRCA1 and BRCA2 in all types of carcinoma examined, except for p53 and BRCA1 in endometrial carcinomas (Table II). The χ2 analysis revealed that LOH at BRCA1 and BRCA2 was significantly correlated only in ovarian carcinomas (n = 61, χ2 = 9.86, p = 0.0017). No such correlation was observed in breast carcinomas, squamous-cell carcinomas of the head and neck and endometrial carcinomas (n = 150, n = 35, n = 27, respectively). Since there was a report of combined loss of BRCA1 and BRCA2 in grade 3 breast carcinomas (Kelsell et al., 1996), we also examined the correlation in this special group. However, we did not find a significant correlation of LOH at BRCA1 and BRCA2 (n = 42). To investigate whether this correlation was specific for the analyzed genes, marker D8S535 was also examined for LOH in ovarian carcinomas. D8S535 was mapped to chromosome 8p11-p21, a region where the existence of three tumor-suppressor genes has been suggested (Macoska et al., 1995). No correlation between LOH of D8S535 and LOH at either BRCA1 (n = 44) or BRCA2 (n = 42) loci was found. Combined LOH affecting all 3 genes was found in 5% (5/103) of breast carcinomas that were informative for all 3 loci, in 33% (18/55) of ovarian carcinomas, in 9% (2/23) of endometrial carcinomas and in 15% (5/33) of head-and-neck carcinomas. Frequencies of LOH affecting p53, BRCA1 and BRCA2 in sporadic breast and ovarian carcinomas have been widely studied. However, the results are not consistent, probably due to different techniques used. In contrast, such studies in endometrial and head-and-neck squamous-cell carcinomas are quite rare. Our results are in accordance with most reports regarding the frequency of LOH at these 3 gene loci detected in ovarian carcinomas (Saretzki et al., 1997; Lancaster et al., 1996) and endometrial carcinomas (Kihana et al., 1995), at the BRCA gene loci in breast carcinomas (Rio et al., 1998; Cleton-Jansen et al., 1995; Niederacher et al., 1997) and at p53 and BRCA1 in squamous-cell carcinomas of the head and neck (Gonzalez et al., 1995). However, the frequency of LOH at p53 in breast cancer that we found was lower and the frequency of LOH at BRCA2 in head-and-neck carcinomas was higher than those reported (Niederacher et al., 1997; Chen et al., 1995; Kirkpatrick et al., 1997). A correlation of LOH at p53 with LOH at BRCA1 and BRCA2 was observed in all malignancies examined. These correlations were statistically significant, with the exception of the correlation between LOH at p53 and LOH at BRCA1 in endometrial carcinomas. The correlation between LOH at markers in BRCA1 and p53 may be partly due to the fact that these genes are located on the same chromosome. However, p53 and BRCA2 are located on different chromosomes. At the initiation of this study, we expected to find a correlation between LOH of p53 and that of BRCA1 and BRCA2 in breast and ovarian carcinomas only. Interestingly, we found such correlations also in squamous-cell carcinomas of head-and-neck and endometrial carcinomas. This indicates that the correlation is not specific for certain tumor types; it rather represents a general phenomenon in the initiation and/or progression of tumors derived from epithelial cells. Our finding supports the gatekeeper–caretaker hypothesis. When a caretaker, such as the BRCA genes, is inactivated, tumor initiation will not be promoted directly; but the genetic instability results in increased inactivation of all genes, including gatekeepers such as p53, which directly regulate the growth of tumors by inhibiting growth or promoting death (Kinzler and Vogelstein, 1997). Thus, a correlation of LOH at BRCA genes with LOH at p53 was found in various human malignancies. Another interesting result is the concordant LOH at both BRCA1 and BRCA2 in ovarian carcinomas, suggesting that these 2 genes might play a role co-ordinately. Combined loss of the function of both genes could be the result of an early genetic event in the onset and/or progression of ovarian carcinoma. There were no correlations of LOH of the 2 BRCA genes with LOH at chromosome 8p. Thus, it is very likely that the BRCA genes are specific targets for inactivation rather than that the results are a non-specific consequence of tumor progression. Further investigations are needed to reveal the mechanism of this combined LOH at BRCA1 and BRCA2. Taken together, our findings indicate that the combined allelic loss at BRCA1 and BRCA2 is specific for ovarian carcinoma. However, the correlation of LOH at p53 with LOH at either BRCA1 or BRCA2 is common in various epithelial carcinomas. Dan Tong*, Elisabeth Kucera*, Eva Schuster*, Rita K.