Background

Li-Fraumeni syndrome (LFS; OMIM #151623) is an autosomal dominant highly penetrant cancer predisposition syndrome characterized by a variety of early onset tumors [1, 2]. LFS arises from germline mutations in the TP53 gene, which codes for a transcription factor implicated in cell proliferation, apoptosis, and genomic stability [3, 4]. Although decreases in the age of cancer onset and increases in cancer incidence over successive generations have been documented in LFS [5, 6], the molecular basis of this anticipation is still a matter of debate.

Most mutations in the TP53 gene affect the DNA binding domain (DBD) and result in missense substitutions giving rise to altered proteins [7] that have a considerable longer half-life than the wild-type protein, resulting in accumulation of the mutant protein, at least in transfected or neoplastic cells [8].

It is known that different mutant p53 proteins may have diverse functional and biological effects, which could partially explain the heterogeneity reported between Li-Fraumeni families [9]. Not only the type of TP53 mutation determines the tumor spectrum and severity of phenotype, but other germline genetic factors such as polymorphisms for TP53 and MDM2[10] and telomere length [11, 12] are assumed to modulate the cancer phenotype. It has been more recently observed that individuals carrying TP53 mutations show a ~3-fold increase in DNA copy number variation (CNV) compared to controls, suggesting that CNVs make an additive contribution to cancer risk [13].

In Brazil, the Li-Fraumeni syndrome also occurs as a frequent cancer predisposition alteration due to a germline mutation at codon 337 (c.1010G4A, p.R337H) that has been identified in families matching LFS definitions [14, 15]. The frequency of p.R337H in the population of Southern Brazil is about 1:280 individuals, about 300 times higher than any other single germline TP53 mutation [16]. Analysis of 28 polymorphic markers showed p.R337H to have arisen from a founder mutation that has been spreading in the Brazilian population since the XVIII century [17, 18]. This mildly pathogenic mutation shows a penetrance for cancer of less than 20% by age 30, compared to about 50% in classic LFS [18]: this provides more opportunity for transmitting the mutation to the next generations, which explains, at least in part, the maintenance in the Brazilian population.

Structural and functional studies have demonstrated that the p.R337H protein displays a pH-dependence, rendering the protein inactive only under conditions of increased intracellular pH, and the majority of the time behaves normally, which could explain its lower overall malignant potential [19].

It is accepted that differences in severity and age of onset of cancer manifestation linked to various germline TP53 mutations not only result from the TP53 mutations themselves, but also from their interaction with other genetic variants, of which CNVs are now becoming recognized as a significant contributant (reviewed in [20]). Based on the foregoing considerations we hypothesize that different types of TP53 mutations exhibit different CNV profiles reflecting the genotype- phenotype correlation and also possibly playing a role in the anticipation observed in LFS families.

Methods

Patients

The patients were seen in the Department of Oncogenetics in the A. C. Camargo Hospital, São Paulo, Brazil. Research protocol approval was provided by the ethics committee of the institution, and informed consent obtained from all subjects and their families. The cohort comprised 21 probands with TP53 germline mutations [14]. The characteristics of the mutations, cancer type, age at onset and clinical classification of the patients are described in Table 1.

Table 1 Clinical data of the patients and characteristics of the detected mutations
Full size table

Controls

The CNV control data used in this study was obtained from a previously reported group of 100 random individuals from the urban area of São Paulo, Brazil [21].

Array-CGH

We performed oligonucleotide microarray-based CGH (array-CGH) using a 4 × 180 K whole-genome platform (design 22060, Agilent Technologies, Santa Clara, USA), which has an average spacing of 18 Kb between probes. Briefly, samples were labeled with Cy3- and Cy5-dCTPs by random priming; purification, hybridization, and washing were carried out as recommended by the manufacturer. Scanned images of the arrays were processed and analyzed using Feature Extraction software and Genomic Workbench software (both from Agilent Technologies), with the statistical algorithm ADM-2, and a sensitivity threshold of 6.7. We applied a′loop design′ in our hybridizations as previously described [22], resulting in two reverse labeling hybridizations per sample. We considered a gain or loss in copy number when the log2 ratio of the Cy3/Cy5 intensities of a given genomic segment was >0.5 or <−0.8, respectively. Alterations had to encompass at least three consecutive probes to be called by the software, and those not detected in both dye-swap experiments of the same sample were excluded from the analysis.

Analysis

The detected copy number variations were compared to CNVs reported in the Database of Genomic Variants (DGV; http://projects.tcag.ca/variation/; freeze December, 2011). We classified the CNVs into ″rare″ and ″common″, considering as ″rare″ those CNVs where the imbalances encompassed coding sequences which were never or only once reported in DGV. Mann–Whitney and Fisher-exact tests were used to evaluate CNVs regarding proportions of total and rare CNVs, frequency of deletions and duplications and size of the variation.

The frequencies were then calculated for each rare variant in the total population described in the DGV studies. For estimating the total population in DGV, BAC array publications have been excluded because they are known to overestimate the CNV boundaries; the remaining 22 studies comprised a total of 8058 individuals investigated. The frequencies of these rare variants were than cross-checked with a study comprising 8148 controls [23], which is not documented in DGV but is registered in dbVar (http://www.ncbi.nlm.nih.gov/dbvar).

Results

The CNV data of this study are summarized in Table 2; a full description of the CNVs identified in the patients can be found in Additional file 1. The overall frequency of CNVs per genome did not differ either between patients and controls or between p.R33H and DBD mutants (Mann–Whitney test). No significant differences were detected between the groups regarding CNV size or frequency of deletions and duplications.

Table 2 Copy number variation data in controls and LFS/LFL patients classified according to TP53 mutation type
Full size table

However, with regards to rarity, we found 13 rare CNVs among all LFS patients, including 9 deletions and 4 duplications. Additional file 2 describes the rare CNVs, the studies in which they have been observed and their corresponding frequencies in the combined individuals of DGV and dbVar. The calculated frequencies of these rare variants in normal populations as estimated based both on DGV and dbVar were < 0.1%, a value much inferior to the 1% threshold often associated with polimorphisms.

In comparison to our control group, TP53 mutated individuals (p.R337H and DBD) did not exhibit differences in the total number of CNVs/genome (Figure 1A). However, we did find a highly significant increase in the rare/common CNV ratio in the LFS patients carrying mutations in the TP53 DBD when compared to both controls (p=0.0002) and to patients carrying the p.R337H mutation (p=0.0156; Fisher-exact test) (Figure 1B). Interestingly, no significant difference was found between controls and p.R337H patients.

Figure 1
figure 1

Distribution of CNVs according to TP53 mutation status. The graphs show no differences in the frequency of total CNVs but an increased frequency of rare CNVs in DBD mutation carriers. (A) Frequency of total CNVs in LFS patients and controls; (B) Rare/common CNV ratio in LFS patients and controls. White bar represents the control individuals, grey bar represents p.R337H mutated carriers and dark black bar the TP53 DBD mutated carriers. Fisher-exact test; *p=0.0156 and ***p=0.0002.

Full size image

Discussion

The vast majority of TP53 mutations lies within the DNA binding domain and target critical sites that either make contact with DNA or are required to maintain appropriate p53 protein conformation. Such alterations lead to a significant loss of TP53 transcription and tumor suppressive activity, with great impact on genomic stability and apoptosis [24–26]. In contrast, the p.R337H mutation, which affects a residue of the oligomerization domain of the p53 protein, has a markedly less severe impact on tumor predisposition, which correlates with its remaining p53 activity [19, 27].

Shlien et al. [13] reported a ~3-fold increase in the frequency of CNVs in TP53 germline mutation carriers compared to normal individuals, and proposed that this increase might result from enhanced genome instability in the presence of TP53 mutations. In our study, we found no difference between the total number of germline CNVs present in LFS patients and controls; however, we noted a highly significant increase (>5 fold) in the rare CNVs in TP53 DBD mutation carriers as compared both to controls and to p.R337H carriers. A difference between Shlien et al. [13] data and ours refers to the controls samples: in their work data previously assembled in other studies were used to establish a baseline CNV control frequency, whereas our control and patient samples have been investigated in parallel in the same laboratory. We should also point out that the studies were carried out using different microarray technologies, namely array-CGH and SNP platforms. In fact, a large study comparing different platforms found that different analytic tools typically yield CNV calls with <50% concordance, overall regarding smaller CNVs [28].

An increase of only rare germline CNVs in the DBD TP53 carriers, rather than an overall increase in CNVs, suggests that a selective mechanism would be involved in this event. Under the assumption that pathogenic variants would only remain in the population at low frequency, the selection against new pathogenic CNVs (in the context of studies on schizophrenia and autism, respectively) is extremely high [29, 30]. The finding of an increased proportion of rare CNVs, potentially pathogenic, among TP53 DBD mutants may be the result of an inefficient selection against new CNVs (deficient apoptosis). This hypothesis could be tested by determining the pattern of inheritance of the CNVs [13], but unfortunately parents were not available for investigation.

Anticipation, characterized by an increase of penetrance and earlier age of onset over subsequent generations, has been well documented for LFS [5, 6]. An accumulation of germline CNVs impacting cancer predisposition in LFS provides a logical explanation for anticipation. How CNVs and other genetic modifiers interact and modulate TP53 tumor suppressor activities remains to be determined. Elucidating these mechanisms may hold the key to defining evidence-based strategies for diagnosis, counseling, follow-up and therapeutic care based on a detailed understanding of the nature of combined risk between TP53 and other variants, including CNVs.

Conclusions

In this work we show a highly significant increase of rare CNVs in patients with mutations in the TP53 DNA binding domain site (DBD) relative both to controls and to carriers of the mild TP53 Brazilian variant. An accumulation of rare CNVs over generations in the presence of penetrant TP53 mutation provides a plausible explanation for anticipation in LFS: in contrast, lack of this increase in rare CNVs combined with a longer lifespan and likelihood of reproduction probably contributes to the frequency of the p.R337H mutation being ~300 xs larger than reported for any other TP53 mutation.

Availability of supporting data

The data set supporting the results of this article is included within the article and its additional files.