High Frequency of TP53 Mutation in BRCA1 and Sporadic Basal-like Carcinomas but not in BRCA1 Luminal Breast Tumors

Breast tumors with a germ-line mutation of BRCA1 (BRCA1 tumors) and basal-like carcinoma (BLC) are associated with a high rate of TP53 mutation. Because BRCA1 tumors frequently display a basal-like phenotype, this study was designed to determine whether TP53 mutations are correlated with the hereditary BRCA1 mutated status or the particular phenotype of these tumors. The TP53 gene status was first investigated in a series of 35 BRCA1 BLCs using immunohistochemistry, direct sequencing of the coding sequence, and functional analysis of separated alleles in yeast, and compared with the TP53 status in a series of 38 sporadic (nonhereditary) BLCs. Using this sensitive approach, TP53 was found to be frequently mutated in both BRCA1 (34 of 35, 97%) and sporadic (35 of 38, 92%) BLCs. However, the spectrum of mutation was different, particularly with a higher rate of complex mutations, such as insertion/deletion, in BRCA1 BLCs than in the sporadic group [14 of 33 (42%) and 13 of 34 (9%), respectively; P = 0.002]. Secondly, the incidence of TP53 mutations was analyzed in 19 BRCA1 luminal tumors using the same strategy. Interestingly, only 10 of these 19 tumors were mutated (53%), a frequency similar to that found in grade-matched sporadic luminal tumors. In conclusion, TP53 mutation is highly recurrent in BLCs independently of BRCA1 status, but not a common feature of BRCA1 luminal tumors. [Cancer Res 2009;69(2):663–71]

Breast cancer is the most frequent tumor in women from developed countries. Numerous studies support the major role of genetic factors in breast cancer susceptibility (see ref. 1 for a review). Indeed, two major susceptibility genes, BRCA1 and BRCA2 (2, 3), were identified in the mid-1990s. Monoallelic deleterious germ-line mutations of these genes confer a major risk of developing breast and ovarian carcinoma (4). BRCA1 and BRCA2 genes behave as classic tumor suppressor genes in that, in heterozygous mutation carriers, the wild-type allele is deleted in the tumor (5). Breast tumors arising in BRCA1 mutation carriers (herein called “BRCA1 tumors”) present particular clinical and histologic features (6): they arise earlier than in the general population (4), they are frequently multifocal diseases (7), and they are associated with poor outcome (8). Histologically, BRCA1 breast tumors are mainly classified as high-grade invasive ductal carcinomas. A noteworthy characteristic of BRCA1 tumors is the frequent alteration of the TP53 gene. TP53 is a major tumor suppressor gene frequently mutated in human cancers, including breast carcinomas (9, 10). This gene encodes a transcription factor playing a central role in DNA damage response, cell cycle control, and apoptosis (11). Its particular involvement in familial forms of breast cancers was first recognized by immunohistochemistry (IHC; refs. 12, 13) and was then rapidly completed by direct sequencing of the TP53 gene. In 1997, Crook and colleagues (14) showed TP53 mutations in seven of seven BRCA1 breast tumors analyzed. A meta-analysis of 12 studies confirmed the higher frequency of TP53 mutations in BRCA1/BRCA2 compared with sporadic breast tumors (15). The frequent TP53 alteration in the BRCA1 tumors is consistent with the prominent role of BRCA1 in the control of the genomic stability. Indeed, in murine cellular models, Brca1 loss of function leads to hypersensitivity to radiation, high genomic instability, and homologous recombination defects (see ref. 16 for review). In mice, biallelic inactivation of Brca1 leads to massive cell cycle arrest with early embryonic lethality that is partially rescued by inactivation of p53 or p21 [coded in mice by transformation-related protein 53 (Trp53) and Cdkn1a, respectively; ref. 17]. Trp53 (the murine ortholog of TP53) expression has also been found to be decreased in mammary tumors arising in a Brca1 conditional inactivation mouse model, and tumorigenesis was accelerated in this model when a Trp53 mutant background was used (18). These results suggest that inactivation of TP53 is necessary for survival of BRCA1^−/− cells and that this event may be a mandatory step in BRCA1 tumorigenesis (19).

Interestingly, BRCA1 tumors frequently display histologic features of basal-like carcinoma (BLC; refs. 20, 21). This uncommon subtype of breast carcinoma was first described by the Stanford group (21–23) and characterized by markers normally expressed by the normal basal/myoepithelial cells of the mammary gland. BLCs are generally described as high-grade carcinomas (including high proliferation, lack of differentiation, and high nuclear pleiomorphism), with peripheral lymphocytic infiltrate and pushing borders. Although no immunophenotypic consensus definition of BLC has yet been established and accepted by the scientific community, the so-called triple-negative status [lack of expression of estrogen receptor (ER), progesterone receptor (PR), and ERBB2] combined with expression of cytokeratins (KRT) 5/6 and/or epidermal growth factor receptor (EGFR) and/or KRT14 can be used for the diagnosis of BLC (24, 25). The TP53 gene is also frequently mutated in this group of tumors (23, 26–29). Interestingly, medullary features, including syncytial growth, well-defined border, and diffuse lymphoplasmocytic infiltrate, are found in a fraction of basal-like tumors. These carcinomas can be individualized as medullary carcinoma (30), in which we have previously shown that TP53 is consistently mutated (31).

This study, based on a thorough analysis of TP53 mutations in a large series of BRCA1 carcinomas and sporadic controls, was designed to determine whether TP53 mutations are associated with BRCA1-null status or with the preferential basal-like phenotype of BRCA1 tumors.

Patients and DNA. A series of 58 independent BRCA1-associated breast tumors was assembled from 54 deleterious germ-line BRCA1 mutation carriers (Supplementary Table S1) who had surgery for invasive breast carcinoma at Institut Curie, Centre René Huguenin or Institut Gustave Roussy and for which frozen samples or nucleic acids were available at the institutional tumor banks. A first control series of 38 sporadic breast carcinomas with a basal phenotype, composed of 24 invasive ductal carcinomas and 14 medullary tumors, has been described in part in a previous report (29). These tumors occurred in patients with no family history of breast or ovarian cancer and/or no detected BRCA1/2 mutations. A second control series was composed of 157 luminal tumors without a suggestive familial history.

The proportion of tumor cells was evaluated on 3-μm sections of frozen samples. DNA was extracted according to standard protocols. In five cases with high stromal content, the frozen blocks were macrodissected before DNA extraction.

IHC analysis of the tumors. In addition to the histologic morphology, Elston-Ellis grading, and mitotic index, immunostaining was performed as previously described (29). Expression of TP53 (Dako A/S), ER and PR (Novocastra), ERBB2/HER2 (Novocastra), EGFR (Zymed, Invitrogen), KRT5/6 (Dako), KRT14 (Biogenex), and γH2AX (clone JBW301, Upstate) was assessed. Internal and external controls for each antibody were included in the experiments.

Mutation screening of the TP53 gene. The TP53 gene sequence was screened from exon 2 to 11 in each tumor. Each PCR was performed on 50 ng of tumor DNA (or cDNA in nine cases). Primer sequences and protocols are available on request. Purified PCR products were sequenced with dideoxynucleotides (BigDye Terminator v1.1, Applied Biosystems), and 3.2 pmol of specific primer in 20 μL, purified on a Sephadex G50 column, and analyzed with a capillary sequencing machine (3130xl Genetic Analyzer, Applied Biosystems). Sequences were then examined with SeqScape v2.5 software (Applied Biosystems). Mutations were detected if the height of the mutated peak reached 20% of the height of the normal peak.

Functional analysis of separated alleles in yeast. TP53 status was determined by functional analysis of separated alleles in yeast (FASAY; ref. 32), which evaluates the transactivation activity of TP53 on a TP53-responsive promoter stably integrated in the yeast genome. Yeast colonies transformed with wild-type or mutated TP53 sequences appear white and large, or red and small, respectively. TP53 status was considered mutated when (a) >15% of the yeast colonies were red, (b) FASAY analysis confirmed the defect in the 5′ or 3′ part of the gene (split version of the test), and (c) sequence analysis from mutant yeast colonies identified an unambiguous genetic defect.

TP53 mutation analysis. Mutations identified by genomic sequencing or by FASAY were analyzed with MUT-TP53 software¹¹

(33). The deleterious effect of missense mutations was deduced from annotations available in the Universal Mutation Database (UMD)¹² database and the IARC¹³ database. The functional consequences of the mutations analyzed were predicted according to Kato and colleagues (34) and Mathe and colleagues (35) available on the IARC Web site. Statistical comparisons between BRCA1 and sporadic BLCs used a χ² test.

Evaluation of the BRCA1 wild-type allele in tumors. Direct sequencing of the known mutation was performed in patients with BRCA1 point mutations (primer sequences and protocols are available on request). The respective heights of the wild-type and mutated allelic peaks in the electrophoregram were compared with those observed using paired germ-line DNA extracted from peripheral blood mononuclear cells. For patients with large deletions of the BRCA1 gene, a multiplex PCR was performed with fluorescent primers located on the deleted and wild-type regions of BRCA1 and on control loci. The analysis included tumor DNA, paired germ-line DNA, and normal and deleted controls [quantitative multiplex PCR of short fluorescent fragments (QMPSF); ref. 36]. Sequences and protocols are available on request. The fluorescent fragments were then separated on a capillary sequencer (3130xl Genetic Analyzer) according to their size. For each sample, the electrophoregrams obtained, in which the height of the peak reflected the copy number of the amplified loci, were normalized with the normal control. The number of copies of BRCA1 in the tumor was compared with that of normal and deleted controls.

Evaluation of gene expression. Expression of ID4 and BRCA1 was evaluated by real-time quantitative PCR using the Taqman system. Briefly, amplification of 50 ng of DNase-digested retrotranscripted RNA was achieved in 20 μL using Taqman primers and probes (ID4: Hs01055311_g1; BRCA1: Hs01556194_m1; Applied Biosystems), and fluorescence intensity was measured by a 7500 Real Time PCR system (Applied Biosystems) and normalized by the 2^−ΔΔCT method. Gene expression in the BT474 cell line was used for baseline expression.

Evaluation of the methylation of the BRCA1 promoter. The methylated status of the BRCA1 promoter was evaluated by sodium bisulfite modification of 500 ng of genomic DNA before methylated-specific and unmethylated-specific PCR. The methylated and unmethylated reactions were performed using the previously published primers (37), leading to 86 and 75 bp products, respectively.

TP53 is consistently mutated in BLC independently of BRCA1 status. A series of 35 BRCA1-associated BLCs was assembled from the Institut Curie tumor collection. These tumors had occurred in 33 patients with well-characterized deleterious BRCA1 mutations (Supplementary Table S1). As determined by IHC characterization, these tumors were classified as “BLC” defined as ER/PR/ERBB2-negative and either KRT5/6- or KRT14- or EGFR-positive IHC, according to the criteria of Nielsen and colleagues (24) and Da Silva and colleagues (Supplementary Table S2; ref. 25). The subset of BRCA1 BLCs was compared with a series of 38 BLCs, which had occurred in patients without a family history of breast or ovarian cancer and/or without any detected BRCA1/2 mutations (subsequently named sporadic BLCs). Medullary features were observed in 5 BRCA1 and 14 sporadic BLCs.

The TP53 status of these BLCs was evaluated by a combination of IHC and genomic approaches. Firstly, positive immunostaining was found in 50% (17 of 34) and 81% (17 of 21) of tested BRCA1 and sporadic BLCs, respectively (Table 1). The coding exons of the TP53 gene were then screened from genomic DNA or cDNA. Mutations were found in 83% (29 of 35) of BRCA1 BLCs and in 89% (34 of 38) of sporadic BLCs (Table 1).

No mutation was detected in 10 tumors (6 BRCA1 and 4 sporadic BLCs), including 4 IHC-positive tumors. A DNA damage response-related TP53-positive staining (38) was excluded by negative γH2AX staining in tested cases. FASAY has been reported to be a more sensitive assay of TP53 mutations, especially in tumor samples with a high content of stromal cells (32). This assay was performed and found positive in 5 of these 10 tumors. TP53 mutations were obtained from the positive yeast colonies in four of these five cases. A posteriori analysis of the corresponding genomic sequences revealed abnormal peaks below the commonly used detection cutoff of 20% (Supplementary Fig. S1). In one case (B1-T11), macrodissection of the sample was necessary to reveal the frameshift mutation identified in FASAY.

Overall, indirect (positive IHC staining) and/or direct (direct sequencing or FASAY) evidence for TP53 mutations was found in nearly all BRCA1 and sporadic BLCs [34 of 35 (97%) and 35 of 38 (92%), respectively; Table 1]. These results confirm and extend the previously observed high rate of TP53 mutations in BLCs by showing an almost constant mutation of this gene in these tumors, whether or not it was associated with BRCA1 germ-line mutations.

Different spectrum of TP53 mutations in BRCA1-associated basal-like tumors. Hereditary breast cancers have been previously reported to be associated with a particular spectrum of TP53 mutations (15). The TP53 mutations identified in BRCA1 and sporadic BLCs were therefore compared in terms of location, type, and frequency in the UMD and IARC databases (Table 2). Firstly, all TP53 mutations occurred between exons 4 and 10 (Fig. 1A). Two mutations implicating splicing sites were found in the sporadic group. Significantly more mutations in exon 5 (13 of 33), and conversely less in exon 6 (2 of 33), were found in the BRCA1 group compared with the sporadic group (6 of 34 and 13 of 34, respectively; P_ex5 = 0.048 and P_ex6 = 0.002). Secondly, mutations were classified as transitions, transversions, and insertions and/or deletions (ins/del) and were compared between the two groups (Fig. 1B). The frequency of transversions was similar in the BRCA1 and sporadic BLC groups [9 of 33 (27%) and 8 of 34 (24%), respectively]. Transitions occurred in 10 of 33 (30%) BRCA1 tumors versus 23 of 34 (68%) sporadic cases (P = 0.002). Conversely, complex mutations (ins/del) accounted for more than one third (14 of 33, 42%) of BRCA1 cases but were rare in the sporadic group (13 of 34, 9%; P = 0.002). Finally, the identified mutations were matched to the UMD and IARC databases. Most of the mutations identified in the sporadic group were hotspot mutations (23 of 34, 68%) compared with only 24% (8 of 33) in the BRCA1 group (P = 0.0004). These results confirm that BRCA1 breast tumors display a particular TP53 mutation spectrum (Table 2; Fig. 1).

TP53 is not consistently mutated in luminal BRCA1 tumors. To determine whether the high frequency of TP53 mutations observed in the BRCA1 group was linked to the BRCA1-null status of these tumors or to their BLC phenotype, a large series of 23 luminal BRCA1 tumors from the tumor collections of three Cancer Centers of the Paris area were then assembled. These tumors had occurred in 21 patients with well-characterized deleterious BRCA1 mutations (Supplementary Table S1). These tumors were defined by IHC positivity for ER and/or PR and negativity for ERBB2. TP53 gene status was evaluated as for BLCs.

TP53 IHC staining was negative for 12 tumors, strongly positive for 4 tumors, and positive for a minority of tumor cells (∼20%) in 3 other cases. Direct sequencing from genomic DNA revealed 10 TP53 mutations (3 transversions, 4 transitions, and 3 ins/del; Table 2). No additional mutation was detected by FASAY, particularly in the three tumors with IHC-positive cells.

Because luminal tumors are rare in BRCA1 mutation carriers, possible coincidental sporadic breast carcinomas had to be distinguished from genuine BRCA1 carcinomas in these patients. BRCA1 tumors can be defined by null BRCA1 expression after loss of the wild-type allele in BRCA1 mutation carriers. The presence of the wild-type BRCA1 allele in tumor DNA was therefore determined by resequencing known BRCA1 mutations or by QMPSF in patients with large-scale BRCA1 rearrangements. All but four of the tumors showed evidence of loss of the wild-type BRCA1 allele (Supplementary Fig. S2). The presence of the wild-type BRCA1 allele in four luminal tumors could be due to low tumor content or tumorigenesis unrelated to the BRCA1 germ-line mutations. These four tumors were therefore excluded from further analyses. None of them displayed evidence of TP53 mutation.

In the remaining 19 BRCA1 luminal tumors (Table 1), 10 presented TP53 mutations. For comparison, TP53 mutations were evident in 4 of 75 (5%), 17 of 62 (27%), and 14 of 19 (74%) histologic grade 1, 2, and 3 sporadic luminal tumors, respectively.¹⁴

Although the size of this series was too small to allow statistical tests, the frequency of TP53 mutations found in BRCA1 luminal tumors (10 of 19, 53%) was similar than that observed in sporadic tumors (grade-matched frequency estimated at 58%).

Finally, it has recently been reported that breast tumors classified as luminal B subtype more frequently presented TP53 mutations than luminal A breast tumors (39). Luminal A and B subtypes were primarily distinguished by their proliferation expression cluster status (40), which is well correlated with the mitotic index (39). BRCA1 luminal tumors have therefore been further subclassified according to their low or high proliferation (mitotic index <10 or ≥10 mitotic figures per 10 high-power fields; 2 and 17 tumors, respectively). A high frequency of TP53 mutations was found in proliferative BRCA1 luminal tumors (presumably B subtype; 10 of 17, 59%), whereas neither of the two tumors with low mitotic index (presumably A subtype) was mutated.

TP53 status was analyzed in this series of BRCA1 tumors by using a robust and sensitive approach combining three different methods: TP53 staining by IHC, direct sequencing of the coding sequence, and FASAY. FASAY provided a major contribution to this analysis by revealing several TP53 mutations not detected by direct sequencing, principally in samples highly contaminated with stromal cells. This combined approach identified TP53 mutations in all but one of the BRCA1 BLCs. Whether the unique BRCA1 BLC without evidence of TP53 mutation is truly associated with a wild-type TP53 status or with an as yet unknown TP53 mutation remains to be determined.

The consistent mutation of TP53 in BRCA1 BLCs could fit with the current model according to which inactivation of the Trp53 pathway is mandatory for survival of Brca1^−/− embryos (17). Elegant demonstrations of this model were recently provided by the in vivo conditional deletions of both Brca1 and Trp53 in murine mammary gland, which led to carcinomas closely resembling human BRCA1 BLCs (41, 42).

The TP53 alteration has also been described as a common feature of sporadic BLCs (23, 26–29) and we further extend this correlation by identifying a TP53 mutation in the majority (92%) of sporadic BLCs of this series. This result is reminiscent of a previous study using FASAY screening on a particular type of BLC, the typical medullary breast carcinoma, and in which a constant TP53 mutation was identified (31). Although BRCA1 somatic mutations in breast carcinomas are extremely rare, it has been proposed that functional inactivation of this gene may be a key event in the malignant transformation of sporadic BLCs. Several mechanisms have been proposed, including epigenetic silencing by BRCA1 promoter hypermethylation (37, 43, 44) and overexpression of the BRCA1 transcriptional repressor ID4 (45), and both mechanisms were indeed observed in our series (Supplementary Table S3; Supplementary Fig. S3). The particular cross-relationship between TP53 and BRCA1 may therefore account for the survival and development of both BRCA1 and sporadic BLCs.

Nevertheless, analysis of the TP53 mutation spectrum in the present series highlighted several major differences between BRCA1 and sporadic BLCs. Greenblatt and colleagues (15) in a large meta-analysis of 12 studies were the first to show that the TP53 mutation spectrum differed between hereditary (BRCA1/BRCA2) and sporadic breast carcinomas in terms of frequency, location on TP53, and strand bias. Our study, which is focused on BRCA1-related hereditary tumors compared with sporadic tumors of the same tumor subtype, confirmed the peculiar nature of the spectrum of TP53 mutations in these tumors and showed the previously unreported overrepresentation of complex (insertion/deletion) mutations in BRCA1 tumors. It is interesting to note that this bias for complex TP53 mutations is also found in our series of BRCA1 luminal tumors. This high frequency of insertions/deletions in TP53, also found in radiation-induced sarcomas (46) and in breast tumors responsive to high-dose chemotherapy (47), could easily be linked to the homologous recombination deficiency in BRCA1 tumors. Similarly, specific gross rearrangements of the PTEN gene have been recently reported in BRCA1 tumors, in contrast with the frequent point mutations of this gene found in sporadic tumors (48). To reconcile the low frequency of TP53 complex mutations in sporadic BLCs with their high frequency of BRCA1 dysfunction (45), residual BRCA1 activity in sporadic BLCs or the occurrence of TP53 mutations before BRCA1 inactivation can be hypothesized in these tumors.

A surprising and new observation is that the frequency of TP53 mutation in BRCA1 luminal tumors was similar to that found in sporadic grade-matched luminal tumors (39).¹⁴ It is unlikely that this low frequency was due to a lack of sensitivity, as the same approach as that used for BLCs was applied. Luminal tumors are rare among BRCA1 tumors, and the risk for BRCA1 carriers of developing this subtype of breast carcinoma is close to that observed in the general population. These tumors may therefore have occurred independently of the BRCA1 germ-line status. However, analysis of the wild-type allele of BRCA1 showed that it was deleted or reduced in most BRCA1 luminal tumors. This BRCA1-null status is associated with a marked overrepresentation of high-grade, highly proliferative luminal tumors in the BRCA1 group compared with the sporadic group as well as large published series of patients (49). Thus, the specific features of those BRCA1 luminal tumors need further investigation. Although we cannot exclude other mechanisms of TP53 pathway inactivation in these BRCA1-null luminal tumors, such as alterations of ARF or ATM, or amplification of MDM2, no such evidence was found in high-resolution genomic profiling of these tumors.¹⁵

It is therefore likely that BRCA1 inactivation has different cellular effects depending on the cell lineage in which it occurs. This could account for the results reported here as well as for the extreme tissue specificity of the cancer predisposition in BRCA1 mutation carriers contrasting with the central and ubiquitous role of BRCA1 in DNA repair.

In conclusion, mutation of TP53 is strongly associated with a basal-like tumor phenotype and is not directly indicative of BRCA1 hereditary mutation. However, insertions, deletions, or complex rearrangements of TP53, if confirmed on larger series, might be a good surrogate clinical marker of BRCA1 mutation.

No potential conflicts of interest were disclosed.

Grant support: Centre National de la Recherche Scientifique, Institut National de la Sante et de la Recherche Medicale, Cancéropôle Ile-de-France, and Institut Curie and its Translational Research Department. M. Warcoin is a recipient of a Fellowship from the Cancéropôle Ile-de-France.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank A. Laugé, V. Moncoutier, D. Michaux, and C. Houdayer for BRCA1 characterization; M. Belotti for medical record analysis; P. Legoix-Né for sequencing analysis; L.F. Plassa, E. Wittmer, M. Legrand, and C. Brunin for their excellent technical assistance in the FASAY analysis; D. Williamson for critical reading; A. Chompret, S. Delaloge, and B. Bressac-de Paillerets for oncogenetic data; and J-M. Guinebretière for tumor phenotyping.

This work is part of the “Cancéropôle Ile-de-France Hereditary Breast Cancer” program coordinated by D. Stoppa-Lyonnet and M-H. Stern.