SEARCH

SEARCH BY CITATION

Keywords:

  • breast cancer;
  • p53;
  • TP53;
  • mutation patterns

Abstract

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Tumor protein p53 (TP53) is mutated in approximately 30% of breast cancers, but this frequency fluctuates widely between subclasses. We investigated the p53 mutation status in 572 breast tumors, classified into luminal, basal and molecular apocrine subgroups. As expected, the lowest mutation frequency was observed in luminal (26%), and the highest in basal (88%) tumors. Luminal tumors showed significantly higher frequency of substitutions (82 vs. 65%), notably A/T to G/C transitions (31 vs. 15%), whereas molecular apocrine and basal tumors presented much higher frequencies of complex mutations (deletions/insertions) (36 and 33%, respectively, vs. 18%). Accordingly, missense mutations were significantly more frequent in luminal tumors (75 vs. 54%), whereas basal tumors displayed significantly increased rates of TP53 truncations (43 vs. 25%), resulting in loss of function and/or expression. Interestingly, as basal tumors, molecular apocrine tumors presented with a high rate of complex mutations, but paradoxically, these were not associated with increased frequency of p53 truncation. As in luminal tumors, this could reflect a selective pressure for p53 gain of function, possibly through P63/P73 inactivation. Collectively, these observations point not only to different mechanisms of TP53 alterations, but also to different functional consequences in the different breast cancer subtypes.

With more than one in ten women estimated to develop breast cancer in her life, breast cancer is the most frequent cancer of women. Breast cancer is a heterogeneous disease, usually classified into Luminal-A, Luminal-B, Basal-like, Normal-like and human epidermal growth factor receptor 2 (HER2)-enriched subclasses.1 However, at the moment, there is not a strict parallelism between subclasses and therapy, and in clinical practice, patient diagnosis is routinely based on estrogen receptor alpha (ERα) progesterone receptor and HER2 statuses. A simplified molecular classification based on ERα, androgen receptor (AR) and forkhead box protein A1 (FOXA1) expression has been recently proposed.2, 3 The first class, ERα+, corresponds to luminal tumors which present active estrogen signaling. The second class (ERα-, AR- and FOXA1-tumors), called basal, shows much higher proliferative and aggressive behaviors. Tumors of the third class (ERα-, AR+ and/or FOXA1+), called molecular apocrine, present androgen and estrogen signaling despite ERα- status, and frequently overexpress HER2.2

Tumor protein p53 (TP53) encodes a multifunction transcription factor whose loss promotes tumor formation.4 It is the most commonly mutated gene in human cancers. Approximately, 30% of breast cancers display TP53 mutation, but this frequency fluctuates from more than 80% in basal-like to less than 15% in luminal-A subgroups.5 Mutations result from misrepair of DNA damages. The main mutation types are substitutions and complex mutations. Substitutions correspond to the replacement of a nucleotide pair by a different pair. They result essentially from replication of mismatches generated by DNA polymerase, or from replication of altered bases such as deaminated cytosines, leading to C/G to T/A substitutions, or oxidized bases leading most frequently to C/G to T/A, C/G to G/C and G/C to T/A substitutions. Complex mutations are deletions or insertions of one or more nucleotides. They result from DNA polymerase slippage or from reparation of DNA breaks by mutagenic pathways such as single-strand annealing or nonhomologous end joining. Substitutions lead to silent, missense or nonsense mutations, whereas complex mutations may lead to frameshift mutations. Interestingly, unlike most other tumor suppressor genes, TP53 essentially displays missense substitutions, resulting often in a protein that is not only functionally altered, but also abnormally stable. Such mutants may be associated with a dominant-negative effect over wild-type p53 function or may exert gain of function via inactivation of other family members (p73 and p63).6 This dominant negative role has been well established by comparing tumorigenesis in p53−/− versus p53 mutant mice. Thus, TP53 mutations can provide specific features, influencing tumorigenesis as well as chemotherapy response.7

Here, we classified 572 breast tumors into three molecular subgroups and assessed whether any characteristics of TP53 mutations would be preferentially associated to a specific subtype of breast cancer. We found that not only frequency, but also type and effect of TP53 mutations discriminate between breast cancer subclasses, suggesting the existence of different mutational events and different selective pressure in each of them.

Material and Methods

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Breast tumors

A cohort of 572 breast tumors, enriched in T2T3 tumors, was collected: 117 tumors from patients treated in adjuvant setting in Saint Louis Hospital and 455 tumors from patients treated by neoadjuvant chemotherapy.8–10

Molecular classification

Tumors were analyzed by microarrays. RNA was amplified, labeled following the manufacturer's one-cycle target labeling protocol (http://www.affymetrix.com) and hybridized to HG-U133A or HG-U133 plus 2.0 Affymetrix GeneChip arrays (Affymetrix; GeneChip Fluidics Station 400). Chips were scanned with the Affymetrix GeneChip Scanner 3000, and images analyzed using GCOS 1.4. RNA expression normalization was performed by GC-RMA with R package (gcrma package, http://bioconductor.org/). Affymetrix expression arrays are publicly available (ETABM-43, E-MTAB-365 in array-express, GSE16716, GSE16446 and GSE26639 in GEO database3, 10–13). ERα, AR, FOXA1, HER2 statuses and TP53 transcriptional levels were established according to probesets ESR1_205225_at, FOXA1_204667_at, AR_211621_at, ERBB2_216836_s_at and p53_201746_at, identified with the classInt R package (R version 2.12.0) using the Fisher–Jenks algorithm.

P53 typing

The p53 status was assessed by functional analysis of separated alleles in yeast (FASAY)14 in tumoral samples used for diagnosis, thus before any therapy. TP53 cDNA was obtained from tumor RNA and expressed in a Saccharomyces cerevisiae strain carrying the ADE2 open reading frame under the control of a p53-responsive promoter. Consequently, the colonies turn red or white color according to p53 status. These analyses were confirmed by the split versions of the assay, which identifies defects in the 5′ or 3′ part of the gene. P53 was considered functional when <10% of yeast colonies were red, and dysfunctional (positive) when >15% were red. When 10–15% of yeast colonies were red, p53 has been considered dysfunctional if confirmed by sequencing. TP53 sequencing was subsequently performed on yeast-derived plasmids of positive cases (p53 dysfunction), using dideoxynucleotides (BigDye Terminator v.1.1, Applied Biosystems) and specific primers.

Statistical methods

Statistical analyses were performed using R 2.13.1 statistical software. Results were compared using the χ2 test. Bonferroni correction was applied in case of multiple comparisons.

Results

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

A cohort of 572 breast tumors was classified using probesets for ESR1, FOXA1 and AR in the available transcriptomes (Supporting Information Fig. 1). We found 53% (ERα+) luminal, 24% (ERα−, AR− and FOXA1−) basal and 23% (ERα−, AR+ and/or FOXA1+) molecular apocrine tumors (Table 1). As expected, molecular apocrine tumors frequently overexpress HER2 (75%, Table 1).2

Table 1. Tumor characteristics1
inline image

By FASAY, we detected p53 dysfunctions in 51% of cases, in line with the fact that this specific population is significantly enriched in large high-grade tumors treated by neoadjuvant regimen. Although FASAY is more effective than immunohistochemistry and more sensitive than direct sequencing,15 this method had also limits. False-negative cases can occur because of rare mutants outside the domain screened or alterations, leading to the absence of mRNA expression. However, this test is one of the most sensitive tests developed to date.

As expected, TP53 mutation frequency was lower in the luminal (26%) than in the basal (88%) and molecular apocrine (69%) subgroups (p < 0.0001, Table 1). We investigate the TP53 mutations within tumor subgroups by sequencing.

We examined the types of TP53 mutations, substitutions (point mutations) and complex mutations (deletions or insertions), according to tumor subgroups (Supporting Information Table 1 and Fig. 1). Luminal subgroup presented higher frequency of substitutions when compared to basal and molecular apocrine subgroups (82% of mutations for luminal vs. 65% for others; p = 0.02), with increased rate of A/T to G/C (31% of substitutions vs. 15% for other tumors; p = 0.02) and also C/G to A/T mutations (15% of substitutions vs. 5% for other tumors; p = 0.05). Using a set of published genes (PAM50),16 we subclassified ERα+ tumors in luminal-A or luminal-B subgroups. In line with the previous data,17 we found that luminal-B tumors were much more often TP53 mutated than luminal-A ones (41 vs. 17%; p < 0.0001, Supporting Information Fig. 2A). Notably, TP53 mutations pattern seemed also different in these two subgroups. Indeed, luminal-B tumors presented more complex mutations than luminal-A (26 vs. 6%; p = 0.03; Supporting Information Fig. 2B). Concerning basal and molecular apocrine subgroups, both of them exhibited increased frequencies of complex mutations (35% for molecular apocrine and basal tumors vs. 18% for luminal ones; p = 0.02). Furthermore, they present high rates of C/G to T/A mutations (60% of substitutions for molecular apocrine and basal vs. 37% for luminal tumors). Interestingly, in basal tumors these C/G to T/A mutations occurred preferentially at CpG sites (80% for basal vs. 47% for molecular apocrine, p = 0.002).1

thumbnail image

Figure 1. Assessment of TP53 mutations types in breast cancer tumors. TP53 mutations are represented according to types: substitutions (point mutations) (black columns); complex mutations (insertions or deletions) (white columns); C/G to T/A mutations at CpG sites (gray columns). Stars indicate significant difference between groups (p < 0.05).

Download figure to PowerPoint

We then examined the functional effects of TP53 mutations according to tumor subgroups (Supporting Information Table 1 and Fig. 2). We found significantly more missense mutations in luminal tumors (75% for luminal vs. 54% for others, p = 0.004), and more nonsense substitutions in basal tumors (15% for basal vs. 6% for others, p = 0.04). Critically, basal tumors displayed almost twofold more truncating mutations than luminal or molecular apocrine tumors (43% for basal vs. 25% for others, p = 0.006; Fig. 2a). Unexpectedly, we observed more in-frame complex mutations in molecular apocrine tumors (14% for molecular apocrine vs. 5% for all others, p = 0.02), suggestive for a selective pressure toward conserved p53 expression. Differentially expressed genes according to TP53 mutation types and effects were determined with a general linear model (lme4 R package). The most significant result was obtained with TP53 gene itself (p = 2 × 10−48) for which a low expression was associated with nonsense and frameshift mutations and a high expression with in-frame and missense mutations (Fig. 2b). This could be explained by nonsense-mediated mRNA decay, as both nonsense and frameshift mutations are associated to protein truncation or loss.18 Collectively, our data unravel the existence of a strong selective pressure for distinct effects of mutations in the different subgroups.

thumbnail image

Figure 2. Assessment of TP53 mutations effects in breast cancer tumors. (a) TP53 mutations are represented according to their effects: truncating mutations (nonsense substitutions and frameshift deletions or insertions) (black columns), and nontruncating mutations (missense substitutions and in-frame deletions or insertions) (white columns). Stars indicate significant difference between groups (p < 0.05). (b) TP53 transcriptional levels according to the types of mutations were evaluated using Probeset p53_201746_at and are represented according to TP53 effects.

Download figure to PowerPoint

Discussion

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Differences in the types of TP53 mutations reflect different mechanisms of DNA damaging. The high frequency of complex TP53 mutations observed in basal and molecular apocrine tumors likely reflects enhanced DNA breaks, as previously suggested in BRCA1-mutated tumors which present multiple genomic rearrangements in comparative genomic hybridization analysis.15, 19 The C/G to T/A transitions are very common in human cancers and result essentially from the spontaneous deamination of cytosines or 5-methylcytosines.20, 21 Indeed, 5-methyl cytosines are more susceptible to deamination than nonmethylated cytosines, making CpG sites hotspots for mutations. In basal tumors, most of these C/G to T/A mutations involved CpG dinucleotides, which may suggest differences in DNA methylation status and/or deaminated-5-methylcytosine repair pathway. The high prevalence of A/T to G/C and C/G to A/T mutations in luminal tumors could suggest involvement of exogenous carcinogens such as tobacco smoke.20, 22, 23 The higher frequency of TP53 mutations (particularly complex ones) in luminal-B than in luminal-A tumors suggests that TP53 may be an important feature in progression of luminal breast cancers and that also DNA breaks may be more common in luminal-B, possibly reflecting their higher proliferation rates.

Importantly, despite similarities in mutations patterns (complex alterations or insertion/deletion) in the basal and molecular apocrine subgroups, p53 truncation was significantly more frequent in basal tumors. This strongly suggests that, although similar mutational events occur in basal and molecular apocrine tumors, selective pressure favors nontruncating p53 mutations in molecular apocrine tumors and/or truncating mutations in basal tumors. Similarly, the high rate of missense mutations (vs. nonsense), observed in luminal tumors, could reflect a selective advantage for the presence of p53 mutant protein. Indeed, many TP53 mutants with missense mutations result in highly stabilized proteins that interfere with p73 and/or p63. Recent studies have demonstrated that p63 inactivation is associated with invasion and metastatic progression, by involvement of integrin and epidermal growth factor receptor signaling, or by transforming growth factor-β cooperation.24 Our findings, therefore, suggest that inactivation of p63 pathway may be advantageous for luminal or molecular apocrine tumorigenesis. On the contrary, basal tumors, possibly owing to their intrinsic very high proliferation rate, metastatic potential and stem cell-like gene expression pattern, may not be sensitive to p63 inactivation.25

In conclusion, although the different prevalences of TP53 mutations among the different breast cancer subgroups were well-known, our studies demonstrate that difference in mutational events and/or selective pressures also exists. The molecular mechanism underlying the different types of mutations in these three molecular subtypes of breast cancer could suggest that they may occur at different phases of tumor development.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

The authors thank all the technicians of the Biochemistry Department, Saint Louis Hospital, for the determination of the p53 status (Claire Bocquet, Catherine Brunin, Dominique Chapelin, Audile Flinois, Brice Geslot and Martine Legrand). The authors thank Jean-Claude Gluckman and Patricia de Cremoux for critical reading and constructive comments on the manuscript.

References

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Supporting Information

  1. Top of page
  2. Abstract
  3. Material and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  8. Supporting Information

Additional Supporting Information may be found in the online version of this article.

FilenameFormatSizeDescription
IJC_27767_sm_SuppFig1.tif92311KSupporting Information Figure 1
IJC_27767_sm_SuppFig2.tif92312KSupporting Information Figure 2

Please note: Wiley Blackwell is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.