CHEK2 contribution to hereditary breast cancer in non-BRCAfamilies

We recruited 507 cases with HBC risk through the oncogenetic consultation department at the Centre Jean Perrin (Clermont-Ferrand, France). This group consisted of 258 families with 3 breast cancers in the same familial branch with at least 2 cases related in the first degree, 237 families with 2 cases of breast cancer in the same branch with at least 1 breast cancer diagnosed before age 40 years or with bilateral breast cancer, and 12 families with 2 cases of breast cancer and at least 1 male breast cancer. One affected patient per HBC family was screened for variants in CHEK2. Cases with HBC linked to BRCA1 or BRCA2 mutations were excluded by direct sequencing of both genes and by multiplex ligation-dependent probe amplification of BRCA1. A control group recruited from the same region of France consisted of 513 female volunteers in good health and without any personal or family history of breast or gynecologic cancers at the time of the recruitment. All subjects signed informed consent agreements that were approved by the CCPPRB Regional Ethics Committee (Auvergne, France). To assess the relationship between CHEK2 variants and breast cancer risk, logistic regression was used to obtain odds ratios (as estimates of relative risk) and 95% confidence intervals [28].

To identify variants in the CHEK2 gene, exons 2 to 14 were analyzed (exon 1 is noncoding, and exon 15, representing 89 bp of coding sequence, could not be analyzed for all the patients, owing to the presence of repeated sequences) in both patients and controls for the genomic sequence [GenBank:NG_008150.1] and for the cDNA sequence [GenBank:NM_007194.3] [29] DNA was extracted from 10 ml of peripheral blood collected on heparin/lithium using a Genomix blood DNA extraction kit according to the manufacturer's instructions (Talent srl, Trieste, Italy). Samples were resuspended with Tris-ethylenediaminetetraacetic acid (EDTA) (TE) (10 mM Tris, 1 mM EDTA, pH 8.0). Exons 2 to 10, including intron-exon boundaries, were amplified by using standard PCR techniques (conditions and primers available on request). Because of the multiple copies of CHEK2 pseudogenes, we used a nested PCR strategy, described previously by Sodha et al. [10], to specifically amplify exons 10 to 14 [30]. Sequence reactions were performed on PCR products purified by ExoSAP-IT (Affymetrix, Inc, Santa Clara, CA, USA) using BigDye v3 reagents (Applied Biosystems/Life Technologies, Foster City, CA, USA) (primers available on request), purified in Sephadex G-50 fine (G5080; Sigma-Aldrich, St Louis, MO, USA) and analyzed using a 3130xl capillary electrophoresis system (Applied Biosystems/Life Technologies). Alignment to the reference sequences was performed using SeqMan NGen software (DNASTAR, Inc, Madison, WI, USA).

For each missense variant, prediction of the impact of the mutation on the protein was assessed by calculating the SIFT (Sorting Intolerant From Tolerant), Align-GVGD and PolyPhen-2 (Polymorphism Phenotyping v2) software tool scores [31‐34]. Align-GVGD predictions and SIFT score were computed using the ortholog alignment of exons 2 to 14 of CHEK2 derived by using Alamut software (Interactive Biosoftware, Rouen, France) [32]. Included were human (Homo sapiens) [GenBank:NP_009125.1], chimpanzee (Pan troglodytes) [GenBank:XP_001172759.1], macaque (Macaca) [GenBank:XP_001101658.1], rat (Rattus norvegicus) [GenBank:NP_446129.1], mouse (Mus musculus) [GenBank:NP_057890.1], dog (Canis lupus familiaris) [GenBank:XP_543464.2], cow (Bos taurus) [GenBank:NP_001029703.1], chicken (Gallus gallus) [GenBank:XP_001232074.1], frog (Xenopus tropicalis) [GenBank:NP_001119996.1] and pufferfish (Tetraodon nigroviridis) [UniProtKB/TrEMBL:Q4TI84], all extracted from the Ensembl Compara database [35]. PolyPhen-2 score was calculated online using default settings and accession numbers [UniProtKB/Swiss-Prot:O96017] [36, 37]. The potential impact on splicing was studied using SpliceSiteFinder, MaxEntScan and GeneSplicer prediction software [38‐40].

The pDream2.1 cloning vector (GenScript USA Inc, Piscataway, NJ, USA) carrying the full-length human CHEK2 coding sequence tagged with an N-terminal FLAG extension under the control of the LacZ promoter for expression in prokaryotes was verified to contain the wild-type (WT) sequence. The Stratagene QuickChange II Site-Directed Mutagenesis Kit (Agilent Technologies, Inc, Santa Clara, CA, USA) was used to generate mutant constructs Ser39Phe, Pro85Arg, Arg117Gly, Arg145Trp, Glu161Del, Arg180His, Lys224Glu, Lys244Arg, Met367fsX15, Tyr390Ser and Thr476Met, with the corresponding primers (available on request) in accordance with the manufacturer's recommendations. All constructs were confirmed by sequencing of the entire coding region of the gene (primers available on request).

Escherichia coli strain BL21 was transformed with pDream plasmids(GenScript USA Inc) encoding WT or mutated Flag-CHEK2. Cultures were grown at 37°C in Luria Broth media containing 100 μg/ml ampicillin until absorbance at 600 nm reached 0.6 before isopropyl β-D-1-thiogalactopyranoside was added to a final concentration of 0.5 mM and incubated for 3 hours. Extraction of total bacterial proteins was performed as described previously [26].

Omnia kinase assay buffer (18 μl; Invitrogen/Life Technologies, Carlsbad, CA, USA) containing 10 μM Sox substrate peptide, 1 mM ATP, 0.2 mM dithiothreitol and 2 μl of 10 × Omnia buffer was incubated for 5 minutes at room temperature and aliquoted to a 96-well plate to ensure equal amounts of the chemosensor. For each assay reaction, 1.5 μg of total bacterial protein from induced cultures containing WT or mutated Flag-CHEK2 or from untransformed E. coli, were then added and mixed gently. CHEK2 protein was added at the moment of the fluorescence acquisition, allowing us to follow the kinetics of substrate phosphorylation. CHEK2 kinase activity was monitored with excitation at 360 nm and emission at 485 nm. Fluorescence was detected using an Infinite 200 PRO plate reader (Tecan Group Ltd, Männedorf, Switzerland) for 60 minutes at room temperature. For each mutation, an average of six wells and three independent experiments were conducted. Each curve was normalized by linear regression using the slope of the corresponding nontransformed bacterial protein extract curve. Thus the slope of the resulting curves represents the ability of CHEK2 recombinant protein to phosphorylate the substrate (see Additional file 1).

To evaluate the contribution of CHEK2 mutations to HBC, we sequenced the coding sequence of the gene, including intron-exon boundaries. We observed 13 different variants in 16 of 507 cases and 4 different variants in 4 of 513 controls (Table 1). In the case population, there were eight different novel missense mutations and one previously described in osteosarcomas [41], as well as one nonsense mutation, one novel frame shift mutation, one splice donor mutation and three patients (0.59%) with the c.1100delC (Met367fsX13) mutation (Figure 1). No mutation hotspots were observed (Figure 1). Mutations among controls included three missense mutations and one affecting a splice donor site. To the best of our knowledge, we are the first to report all mutations found in the control population. The missense mutation Lys244Arg was found in both cases and controls. The mutation frequency was higher for the cases (16 of 1,014 vs 4 of 1,026; P = 0.0065) (Table 2). The OR of CHEK2 mutation carriers was 4.15 (95% CI = 1.38 to 12.50), suggesting that CHEK2 contributes to hereditary risk of breast cancer.

Canonical splice donor and acceptor sites were evaluated using SpliceSiteFinder, MaxEntScan and GeneSplicer. All three programs provided consistent information that the two mutations affecting splice donor sites abrogate splicing of the exons concerned (Table 1). We thus considered these mutations to be deleterious. Because the effect of an amino substitution can be difficult to assess, a combination of three different in silico analyses (Align-GVGD class, SIFT prediction and PolyPhen-2 prediction) was used. For each missense variation, we compiled these three scores to propose a diagnosis. Missense variants were considered probably deleterious if at least one deleterious score was obtained and probably benign if three benign scores were obtained. Class above C35 was considered the threshold for deleterious variants in Align-GVGD.

Substitutions with a SIFT score less than 0.05 are predicted to be deleterious. A SIFT median sequence conservation score cutoff of 3.25 was used to measure the diversity of the sequences used for prediction, and a score greater than 3.25 could indicate that the prediction was based on closely related sequences. This would result in a low confidence score if the variant were considered deleterious. No SIFT median sequence conservation score reached this cutoff, indicating that the aligned sequences were diverse enough for confident prediction of substitutions that should affect protein function. One mutation, Lys244Arg, present in both cases and controls, was not considered to be potentially deleterious on the basis of the results of any of the algorithms used, suggesting it is a rare but benign variant. All other missense variants were considered potentially damaging on the basis of at least one measure.

To evaluate whether missense variants inhibit the function of the CHEK2 protein, an in vitro kinase activity test based on a CHEK2-specific substrate peptide carrying a C-terminal SOX was developed [42]. Overexpression of recombinant CHEK2 at high levels in bacteria is associated with CHEK2 autophosphorylation and activation in the absence of DNA damage [13]. This property was used to obtain recombinant activated CHEK2. Upon the phosphorylation of the SOX-specific substrate by CHEK2, the presence of the chemosensor SOX results in an increase in fluorescence at 485 nm. Activity was detected for the WT protein but not for proteins extracted from nontransformed bacteria or recombinant CHEK2 protein carrying c.1100delC (Figure 2). Only missense variants were tested for kinase activity. c.190G > T (Glu64X), c.825_826del, c.846+4_+7del and c.792+1dup were considered deleterious without further analysis. Four mutations tested for kinase activity in vitro by Sodha et al. [27] were included to validate the assay.

Three different classes of kinase activity were observed: WT-like, intermediate and null (Figure 2). Mutations Ser39Phe, Arg145Trp and Arg137Gln exhibited WT-like kinase activity, suggesting that these mutations do not affect the ability of recombinant CHEK2 to recognize, bind and phosphorylate its substrate (Figure 2). The mutants Pro85Arg, Arg180His and Lys244Arg had significantly lower, but not null, kinase activity (Figure 2), which placed them in the intermediate class. The mutations Glu161Del, Lys224Glu, Thr476Met and Tyr380Ser did not have any kinase activity. Nine of the eleven mutations showed kinase activity consistent with the in silico analysis, demonstrating the good but incomplete correlation of those two approaches (Table 3).

Both in silico and in vitro analyses suggested that the variant Lys244Arg, present in cases and controls, can be considered benign. This variant was thus removed from the pool of potentially deleterious CHEK2 variants that contribute to HBC. As a result, the mutation frequency was reduced to 1.48% for cases and 0.29% for controls (Table 2). This difference remained significant (P = 0.0042), and the OR associated with the presence of a deleterious mutation was increased to 5.18 (95% CI: 1.49 to 18.00).