Increased chromosomal radiosensitivity in asymptomatic carriers of a heterozygous BRCA1 mutation

Blood samples were collected from individuals consulting the clinic of the Centre for Medical Genetics, Ghent University Hospital, Belgium (CMGG). Both ethylenediaminetetraacetic acid (EDTA) and heparin blood samples were collected. EDTA samples were used for mutation analysis at CMGG, whereas the G₂ MN assay was performed on heparinized blood samples. In addition, we collected heparinized blood samples from healthy volunteers (n = 20) without a personal or family history of BC to determine the normal distribution of micronucleus yields in controls. Lymphocytes were isolated from the blood samples using Lymphoprep™ (STEMCELL Technologies, Vancouver, BC, Canada) and were preserved in liquid nitrogen for analysis of NMD of the mutant allele. For a number of mutation carriers (n = 4) and healthy volunteers (n = 7), a second blood sample was taken to determine the reproducibility of the results obtained with the G₂ MN assay at different time points.

This study was approved by the ethics committee of the Ghent University Hospital (B670201111641 d.d. 20/09/2011). All study participants (n = 18) were counselled by clinical geneticists in the context of a predictive (n = 16) or diagnostic (n = 2) test for hereditary BC and signed an informed consent form. Two of the eighteen BRCA1 mutation carriers had developed BC, but their cancer treatment had finished more than 2 years ago; these women are both carriers of a substitution affecting the start codon (M01 and M02). The mean ages of the mutation carriers and healthy volunteers were 40.9 and 35.4 years, respectively (p = 0.26, t test).

For the G₂ MN assay, a large blood culture was set up in the presence of the mitogen PHA. PHA stimulates T-lymphocyte division, resulting in a population of cycling lymphocytes (G₁, S₁, G₂ and M phase) after 3 days of incubation, when the blood culture is irradiated. On the contrary, in the G₀ MN assay, blood is first irradiated and then cultured in the presence of PHA, resulting in the irradiation of T lymphocytes in G₀ phase. The addition of cytochalasin B (cyto B), an agent that blocks cytokinesis, to the blood cultures allows the identification of first-division cells as binucleated (BN) cells. A non- or misrepaired DSB can result in an acentric chromosomal fragment, which is detected as a micronucleus in the cytoplasm of the BN cell [17, 28].

More precisely, for the G₂ MN assay, a 50-ml blood culture was set up using 5 ml of heparinized blood, 45 ml of complete RPMI (RPMI with 1 % l-glutamine and 0.5 % penicillin/streptomycin) containing 10 % foetal calf serum (FCS) and 1 ml of PHA (all Gibco®; Thermo Fisher, Rockford, IL) in a T75 culture flask [27]. This large culture was set up to avoid interculture variation during the first days of incubation. After 3 days, the culture was split, and sham irradiations as well as irradiations with 2 and 4 Gy ⁶⁰Co gamma rays were performed. An overview of the experiments is shown in Fig. 1a. Immediately after irradiation, cyto B (6 μg/ml; Sigma-Aldrich, St. Louis, MO, USA) was added. To half of the irradiated cultures, caffeine (CAF, 4 mM; Sigma-Aldrich), an agent that abrogates the G₂/M checkpoint, was added to determine the G₂/M checkpoint efficiency ratio [29, 30]. On the basis of 5-bromo-2′-deoxyuridine (BrdU) results obtained in our previous study on ATM [27] and new experiments performed for this study, a post-irradiation incubation period of 8 h was selected for detecting DNA damage induced in cells in G₂ phase (Fig. 1b). The cultures were then treated with 75 mM KCl and fixed once using a combination of methanol, acetic acid and Ringer’s solution (9 g NaCl + 0.42 g KCl + 0.24 g CaCl₂ for 1 L of dH₂O) (ratio 4/1/5, 4 °C) and thrice with methanol and acetic acid (ratio 4/1, 4 °C). Finally, the cell suspension was concentrated and spread on slides.

After 4′,6-diamidino-2-phenylindole staining, the slides were scanned with a Metafer 4 (MN score software module; MetaSystems; Altlussheim, Germany). This automated image analysis system selects BN cells and determines the number of MN per BN cell (see Fig. 1a). For each condition, two cultures were set up and two slides per culture were analysed. A minimum of 600 BN cells were scored on each coded slide for the presence of MN. BN and MN selected in the automated setting were manually checked for false-positives and false-negatives. To determine the G₂/M checkpoint efficiency ratio, the quotient was determined for the MN yield obtained in the presence versus the absence of caffeine (MNCaf⁺/MNCaf⁻). The lower this ratio was, the more radiosensitive an individual was.

To assess the overall radiosensitive phenotype of each patient, a RIND score was calculated. The mean and standard deviation (SD) of the micronucleus yield and the G₂/M checkpoint efficiency ratio assessed in the cohort of healthy volunteers served to determine cut-offs to define the radiosensitivity of individual BRCA1 mutation carriers. An individual value equal to or greater than the mean MN yields in controls + 1 SD scored 1 point (colour-coded orange). A value equal to or greater than the mean MN yields of controls + 2 SD scored 2 points (colour-coded red). An individual value less than the mean of controls + 1 SD was scored as naught (colour-coded white). For the G₂/M checkpoint, a similar conversion of the data was performed, with a value less than the mean checkpoint ratio assessed in controls − 1 SD scored 1 point (colour-coded orange) and the mean − 2 SD scored 2 points (color-coded red). An individual value greater than the mean of controls − 1 SD was scored as naught (colour-coded white). The combined scores from MN yields (0, 1 or 2 points) and checkpoint ratios (0, 1 or 2 points) were then added to form a single RIND score (range 0–4).

BrdU was added to blood cultures of three individuals to analyse the phase of the cell cycle at the time of irradiation. BrdU (0.01 mM; Sigma-Aldrich) was added to the cultures immediately after irradiation. Binucleated cells that incorporated BrdU during synthesis were in S or G₁ phase of the cell cycle at the moment of irradiation, whereas BrdU-negative binucleated cells were irradiated in G₂ phase. We corrected for daughter nuclei incorporating BrdU after binucleation by adding BrdU 2 h before fixation to another subset of cultures. BrdU was visualized by fluorescence immunostaining with a monoclonal BrdU-specific antibody (M0744; Dako, Carpinteria, CA, USA).

For this experiment, 2 × 10⁶ frozen lymphocytes were thawed and cultured in a mix of 1 ml of cRPMI, FCS (10 %), 2-mercaptoethanol (0.1 %, Gibco®) and sodium pyruvate (1 %; Life Technologies; Carlsbad, CA, USA). PHA (10 μl/ml) was added to stimulate cell division. At day 7, whole RNA was extracted using the QIAamp® RNeasy Mini Kit (QIAGEN, Valencia, CA, USA) according to the manufacturer’s instructions. Four hours before extraction, all cultures were split in two, and puromycin (200 μg/ml; Sigma-Aldrich) was added to one part. Puromycin was added to avoid NMD, a pathway responsible for the degradation of aberrant mRNA [31].

To determine the ratio of the mutant BRCA1 allele versus the wild-type (WT) allele at the mRNA level, we performed polymerase chain reaction (PCR) amplification and Sanger sequencing of the amplicon harbouring the germline mutation and a heterozygous single-nucleotide polymorphism (SNP) (c.2311 T > C), if present. An overview of the primers used can be found in Additional file 1. Sanger sequencing was performed using the BigDye® Terminator Cycle Sequencing Kit (Life Technologies). PCR was followed by ethanol precipitation, and fragments were dissolved in a mix of Hi-Di™ formamide (10 μl; Life Technologies) and GeneScan™ 500 LIZ™ Size Standard (0.5 μl; Applied Biosystems, Foster City, CA, USA). The fragments were analysed on the ABI PRISM® 3730KL Genetic Analyzer (Life Technologies). Results were evaluated using GeneMapper® software (Applied Biosystems) for visual inspection and determination of the ratio of the peak heights representing the WT and mutant alleles, respectively (Additional files 2 and 3). Data were normalized with a control peak in the near vicinity. To confirm the results, the PCR amplicons were also sequenced with the sequencing by synthesis technology on a MiSeq instrument (Illumina, San Diego, CA, USA). The library preparation was performed using an adapted Nextera XT protocol (Illumina) as described by De Leeneer et al. [32]. Reads were mapped with CLC Genomics Workbench (CLC bio/QIAGEN, Aarhus, Denmark). The variant allele frequency (VAF) (which reflects the ratio of WT versus mutant allele) of both the mutation and a SNP (if available) was determined in the mutation carriers and in controls without a germline BRCA1 mutation.

On the basis of data reported in the literature, the cell cycle of proliferating lymphocytes takes between 18 and 22 h. The cell cycle is presented in Fig. 1b, and the mean length for each phase is indicated [33, 34]. With the BrdU immunostaining experiments, we determined that a post-irradiation incubation time of 8 h after 2- or 4-Gy irradiation resulted in blood cultures in which 80 % (±13 %) and 87 % (±3 %) binucleated cells, respectively, were in G₂ phase (negative at BrdU staining) at the time of irradiation. In sham-irradiated samples, the percentage of BrdU-negative binucleated cells was much lower (56 ± 8 %). These observations suggest that synchronization of the cells in G₂ phase took place because of radiation-induced G₂ arrest. When caffeine, an agent that abrogates the G₂/M checkpoint [29], was added to the cultures, the percentage of cells in G₂ phase decreased (2 Gy 73 % ± 6 %; 4 Gy 79 % ± 2 %). The high percentage of cells irradiated in G₂ phase indicates that the harvest of BN cells 8 h post-irradiation is optimal.

The MN yield obtained for the sham-irradiated samples did not show a significantly different mean result between mutation carriers and healthy volunteers (Table 1 and Fig. 2). The mean values obtained for the radiosensitivity analysis are shown in Table 1 and Fig. 2. In BRCA1 mutation carriers compared with healthy volunteers, a significantly increased radiosensitivity was observed for both endpoints at both radiation doses of 2 Gy and 4 Gy. Our results also show that the 4 Gy dose point was the most discriminative (highest significance based on p value) for both endpoints. We thus selected this dose to determine the individual overall radiosensitive phenotype. Results of the mutation analysis and individual results obtained for the two radiosensitivity endpoints upon 4 Gy irradiation are shown in Table 2. Each value received a colour code corresponding to a radiosensitivity score, from which a RIND score was calculated.

Significantly different RIND values were found for BRCA1 mutation carriers (median 2) and healthy volunteers (median 0) (p = 0.0076, Mann-Whitney U test). Figure 3 shows the distribution of the BRCA1 mutation carriers and healthy control subjects for the five different RIND scores. The distribution amongst the RIND scores between the two groups is significantly different (p = 0.034, Fisher’s exact test). The significantly different median and the significantly different distribution over the RIND scores obtained for both groups point towards a difference in response to radiation.

Repeated assessments were performed on blood specimens taken on two different occasions from a random sample of seven healthy volunteers and four mutation carriers to exclude intraindividual variations. Although minor variations were observed, no significant differences were observed between repeated measurements (p > 0.05 for both MN yield and G₂/M checkpoint efficiency; repeated-measures analysis of variance).

We also investigated whether relatives of the subjects enrolled in this study, carrying the same mutation, had similar RIND scores. Individuals with the same family ID are known to be related (Table 2). For families BR-32-0196, BR-32-1028 and BR-32-2256, we had access to data and samples of several relatives. Both father and daughter of family BR-32-1028 showed a RIND score of 3. The siblings from family BR-32-2256 show RIND scores of 1 and 2, and we obtained scores of 0 and 1 for the third-degree relatives in family BR-32-0196. We thus conclude that there were no major differences in RIND scores within families. However, different RIND scores (ranging from 0–4) were observed between four individuals carrying the same mutation (c.2359dup), but we were unaware of a close relationship.

The results of fragment analysis and massive parallel sequencing were comparable (Table 3). All results for cDNA, but not for gDNA, analyses are obtained by in duplicate experiments. The results of the MiSeq analysis are expressed as the VAF in percent; 50 % means equal expression of both alleles. The results of the fragment analysis are shown as a ratio of peak heights of mutant to WT allele. A value of 1 equals no loss of expression of the mutant allele. cDNA samples not treated with puromycin were scored as “evidence of NMD” (Table 3) when an average VAF ≤31 % or a ratio of peak heights <0.7 was obtained. We observed more variation for deletions/insertions than for substitutions. This can be explained by the more complex mapping of reads containing deletions/duplications. A large deletion and software struggling to map correctly to the reference sequence explains a VAF <50 % for gDNA in some patients (Table 3). Nevertheless, a drop in VAF for cDNA and not for cDNA with puromycin can still be distinguished.

For M02, M08 and M15, no lymphocytes were available to perform this assay. However, as we had access to cDNA from other individuals with the same mutation, we were able to gain insight into the stability of the mutant mRNA for these three mutation carriers. For M16, no gDNA data could be obtained; information with regard to mutant mRNA stability for this mutation carrier was obtained via cDNAp and M17.

For all truncating mutations studied in the central part of the gene (7 unique mutations in 15 individuals), we found evidence for NMD. In general, 25–30 % residual truncated mRNA was detected in carriers of a premature termination codon (PTC) mutation. Ten of fifteen patients with a truncating allele showed a radiosensitive phenotype (RIND score 1–4, median 1).

For mutations in the 5′ part of the gene (n = 3), we have no evidence for NMD. The effect at the mRNA level for the c.212 + 3A > G splice-site mutation has previously been studied with quantitative PCR by our group. We showed that no full-length transcript is formed, but that it leads to a significantly increased expression of an alternative transcript (out of frame skip of 22 last nucleotides of exon 5; r.190_212del), which is not subjected to NMD [35, 36]. As M03 is not heterozygous for any of the tested common SNPs, results could not be re-analysed with the approach described in this paper. Our results for M03 showed a RIND score of 2, which can be attributed to a decreased G₂/M checkpoint efficiency. For M01, carrier of the c.1A > G start codon mutation, no decline of the mutant allele could be detected when analysing the SNP data (Table 3). This suggests the conservation of the start codon or an alternative one. The mutation itself could not be quantified, owing to the presence of GC-rich areas near the start codon. The introduction of an alternative start codon could give rise to an aberrant protein with a dominant-negative effect [37]. Interestingly, both individuals carrying a mutation affecting the start codon had, respectively, RIND scores of 3 and 2. A median RIND score of 2 was observed for carriers of a mutation in the 5′ part of the gene.