Mutation analysis of the CHK2 gene in families with hereditary breast cancer

Recently CHK2 was functionally linked to the p53 pathway, and mutations in these two genes seem to result in a similar Li–Fraumeni syndrome (LFS) or Li–Fraumeni-like syndrome (LFL) multi-cancer phenotype frequently including breast cancer. As CHK2 has been found to bind and regulate BRCA1, the product of one of the 2 known major susceptibility genes to hereditary breast cancer, it also more directly makes CHK2 a suitable candidate gene for hereditary predisposition to breast cancer. Here we have screened 79 Finnish hereditary breast cancer families for germline CHK2 alterations. Twenty-one of these families also fulfilled the criteria for LFL or LFS. All families had previously been found negative for germline BRCA1 BRCA2 and TP53 mutations, together explaining about 23% of hereditary predisposition to breast cancer in our country. Only one missense-type mutation, Ile157→Thr157, was detected. The high Ile157 → Thr157mutation frequency (6.5%) observed in healthy controls and the lack of other mutations suggest that CHK2 does not contribute significantly to the hereditary breast cancer or LFL-associated breast cancer risk, at least not in the Finnish population. For Ile157 → Thr157our result deviates from what has been reported previously. © 2001 Cancer Research Campaign http://www.bjcancer.com

encodes a protein kinase required for DNA damage and replication checkpoints, is another tumour suppressor gene along with TP53 conferring predisposition to sarcoma, breast cancer and brain tumours. After DNA damage, ATM-dependent activation of both p53 and CHK2 occurs (reviewed in Prives and Hall, 1999). As CHK2 is capable of phosphorylating p53 at Ser 20 (Hirao et al, 2000), it appears to function as an intermediate kinase and thus plays a key role in connecting p53 to the response to doublestranded DNA breaks. Furthermore, CHK2 also binds to and regulates BRCA1 (Lee et al, 2000), and the phosphorylation of BRCA1 at Ser 988 is required for the release from CHK2. Wang et al (2000) suggested that BRCA1 could act as a scaffold protein that organizes different types of DNA damage sensors and then serves as an effector in response to DNA damage to coordinate repair.
Both the association to LFS/LFL and the regulatory control of BRCA1, encoded by one of the 2 known major susceptibility genes to hereditary breast cancer, makes CHK2 a good candidate gene to search for involvement in the remaining unexplained cases of genetic predisposition to this disease. The search for CHK2 mutations was performed on 79 Finnish families with indications of hereditary breast cancer, in which BRCA1, BRCA2 and TP53 mutations were previously excluded (Huusko et al, 1998(Huusko et al, , 1999Rapakko et al, 2001). The validation of observed sequence alterations was done on cohorts of suitable cancer-free and unselected breast cancer individuals.

MATERIALS AND METHODS
The search for CHK2 germline mutations included all exons and splice-site boundary regions and was performed on 79 families with hereditary breast cancer (Table 1) originating from the Oulu University Hospital area. From some of the cancer families multiple affected individuals were studied. In addition, from 3 of Mutation analysis of the CHK2 gene in families with hereditary breast cancer the families unaffected members were also analysed for a specific gene alteration. Of the total of 98 breast cancer cases, 7 (7%) were identified at or below age 35, 23 (24%) between ages 36-45, 49 (50%) between ages 46-60, and 19 (19%) at or above age 61. Fifty-eight families met the criteria for moderate-to high-risk hereditary breast cancer only, 20 families for both hereditary breast cancer and LFL, and one family for both hereditary breast cancer and LFS. The used criteria for hereditary breast cancer were one or more of the following: (1) at least 3 (2 in combination with other selection criteria) cases of breast cancer in first-or second-degree relatives; (2) early disease onset (≤35 years alone, or <45 in combination with other inclusion criteria); (3) bilateral breast cancer; or (4) multiple tumours including breast cancer in the same individual. The criteria for LFL/LFS were as in Birch et al (1994) and Eng et al (1997). Informed consent to obtain pedigree data and blood specimen for a study on cancer susceptibility gene mutations was obtained from all patients. Control DNA samples from blood were derived from 200 anonymous cancerfree donors and 259 unselected breast cancer patients. Approval to perform the study was obtained from the Ethical Board of the Northern Ostrobotnia Health Care District and the Finnish Ministry of Social Affairs and Health.
DNA extraction from blood lymphocyte specimens was performed using the standard phenol-chloroform method. The screening for CHK2 mutations was done by conformation-sensitive gel electrophoresis (CSGE) analysis (Huusko et al, 1998). Samples with a band-shift were reamplified and purified with the QIAquick PCR purification Kit (Qiagen). Sequencing analysis was performed with the Li-Cor IR 2 4200-S DNA Analysis system (Li-Cor Inc, Lincoln, USA) and using the SequiTherm EXCEL ™ II DNA Sequencing Kit-LC (Epicentre Technologies), following the protocol provided by Li-Cor. Oligonucleotides for CSGE analysis were synthesized based on available CHK2 genomic sequences (Genbank accession number AL117330). Additional oligos for CSGE and sequencing were designed by using the Primer3 software. Primer sequences and PCR conditions for CSGE and sequencing are available upon request.
Mutation frequency differencies between the tested groups were analysed in Bayesian framework (Gelman et al, 1995). Unlike the Chi-square test, this approach provides the probabilities for the presented hypothesis being both true and false. Furthermore, in the Bayesian model none of the expected values are fixed, which results in a more plausible statistical estimate. The probability model was set up assuming that the number of mutations follow poisson distribution with mean λ i = θ i N i , when the number of individuals is N i and the mutation frequency is θ i . Also, θ i was assumed to follow Beta (1, 1) = Unif (0, 1) distribution. Formally: xi|Ni,θi ~ Poisson (θiNi) θi ~ Beta (1,1) The comparisons between mutation frequencies in different groups were performed by calculating the ratio of the frequencies, R ij = θ i /θ j . Posterior distributions of the model parameters were obtained by Monte Carlo Markov Chain stimulation, which was carried out with WinBUGS 1.3 software. Also, for H 0 (estimating how well the frequency observed in one group equals that in the comparison group) traditional Chi-square test calculations were performed, using P = 0.01 as cut-off value for statistical significance.

RESULTS AND DISCUSSION
In the current study, only one missense-type mutation, Ile 157 → Thr 157 , was detected within the protein-encoding region of the CHK2 gene. This alteration was the same as that previously reported by Bell et al (1999). In addition, 2 changes in intronic sequences were found. No splice-site alterations were observed.
Ile 157 → Thr 157 was seen in 7/79 (8.9%) of breast cancer families (group 1). Four of these 7 families also met the criteria for LFL. In 2 of the mutation-positive families, the mutation segregated ambiguously with the cancer phenotype ( Figure 1). In family #5, a woman with breast cancer diagnosed at 80 carried the mutation, whereas her unaffected 47-year-old daughter did not. However, the proband's unaffected 63-year-old niece was found to be a mutation carrier. In family #7, a mother and daughter diagnosed with breast cancer at ages 64 and 49, respectively, were both mutation carriers, but the other daughter who had breast cancer at 40 was not. In addition, Ile 157 → Thr 157 was found in 13/200 (6.5%) of anonymous cancer-free blood donors (group 2), and 10/259 (3.9%) of unselected breast cancer cases (group 3). Using the Bayesian model, none of the probabilities for the mutation frequencies being higher among hereditary breast cancer patients reached 0.99, the minimum value to prove that the observed incidence is higher than expected. The obtained probabilities were 0.78 (group 1 vs 2), 0.11 (group 2 vs 3) and 0.96 (group 1 vs 3). To estimate how well the frequency observed in one group equals that in a comparison group, traditional Chisquare test calculations were made. The obtained values were 0.72 (P = 0.395), 2.96 (P = 0.085) and 5.53 (P = 0.019), respectively, and thus statistically insignificant.
As implied by the performed statistical analysis, our observation for group 2 is in contrast to the previous finding of Bell et al (1999), who did not detect the Ile 157 → Thr 157 missense mutation among any of the 50 healthy individuals used as controls, but only in one LFL individual with 3 primary tumours (breast, melanoma and lung) and no other reported family history of cancer. Although Ile 157 → Thr 157 is located within the forkhead-associated (FHA) domain, which is a highly conserved 60-amino acid proteininteraction domain essential for activation of the CHK2 yeast homolog Rad53 in response to DNA damage (Sun et al, 1998)  high mutation frequency (6.5%) now observed in healthy Finnish controls suggests that Ile 157 → Thr 157 is not, at least alone, a mutation resulting in predisposition to cancer. The statistical analysis also shows that Ile 157 → Thr 157 is not significantly enriched among breast cancer patients having hereditary disease background (including LFL). Furthermore, the ambiguous segregation in the studied informative cancer families suggests that this alteration is rather a polymorphism than a deleterious mutation. This notion is also supported by the recent observation of Wu et al (2000), who found that CHK2 protein carrying the Ile 157 → Thr 157 change has similar kinase activity, expression levels and subcellular localization as endogenous CHK2. Also, like wildtype CHK2, the mutant protein is activated following gamma radiation. However, it is still unclear whether Ile 157 → Thr 157 has other effects on cellular phenotype, or possibly acts as a genetic modifier on a breast cancer predisposing background. Bell and coworkers (1999) screened 4 LFS and 18 LFL cases, and detected CHK2 mutations in 3 of the studied families (13.6%). Therefore, a similar incidence of CHK2 mutations was initially expected also among the 21 LFL and LFS families studied by us. Together with the recent results of Sodha et al (2000) it now appears that only 1 of 3 CHK2 mutations originally reported by Bell et al (1999) is a true disease-causing change, and thus the expected frequency of CHK2 mutations in LFS and LFL families would be lower than was initially assumed.
Due to the duplications of the 3′ genomic sequences of CHK2 reported by Sodha et al (2000), atypical banding in CSGE was observed while analysing the terminal exons 10-14, encoding most of the protein kinase domain (data not shown). CSGE analysis is based on homo-and heteroduplex formation between wild-type and mutated alleles, leading to altered mobility of different types of DNA duplexes on a denaturating polyacrylamide gel. Körkkö et al (1998) showed that it is possible to detect more than one kind of mismatch in the same PCR product, by the appearance of new heteroduplex bands in CSGE. Therefore, instead of a single band (e.g. homoduplex) indicating the lack of mutation, genomic loci coamplified in PCR with a tested segment of CHK2 exon 10-14 would in CSGE analysis result in additional bands (e.g. one or more heteroduplexes). For that reason, we concluded that screening for samples displaying a different banding pattern in CSGE could at least provide a rough idea whether the analysed exons contain alterations or not. The banding patterns for exons 10-14 in our study, however, were similar in all screened DNA samples (data not shown). To conclusively exclude the presence of mutations in exons 10-14, this negative result should be confirmed by using allele-specific PCR amplification.
Unfortunately, in the current study fresh sample material or breast cancer cell lines to perform this type of analysis were not available.
As no other mutations besides Ile 157 → Thr 157 were detected within the protein-encoding region of the CHK2 gene, our results suggest that CHK2 does not play a significant role as predisposing factor for hereditary breast cancer, or LFL showing excessive cases of breast cancer, at least in the Finnish population. Larger studies will be needed to more carefully evaluate the significance of CHK2 alterations in predisposition to cancers related to LFS, as well as to estimate the possible effects of founder mutations in different populations.