Introduction
Breast cancer is one of the main causes of cancer-related deaths among women worldwide, with 5% to 10% of cases being due to hereditary risk. However, mutations in the two major genes,
BRCA1 and
BRCA2, are found in only 15% to 20% of hereditary breast cancer (HBC) families [
1]. Several studies have reported evidence that germline mutations in other susceptibility genes, such as
ATM, PABL2, BRIP1 and
CHEK2, might be the predisposing factor in some HBC families [
2‐
5]. In addition, the lower penetrance of these mutations suggests that they might act in concert with other hereditary factors [
6‐
10].
CHEK2 is the human homolog of
Rad53 (
Saccharomyces cerevisiae) and
Cds1 (
Schizosaccharomyces pombe). This family of kinases is characterized by several domains: a SQ/TQ cluster domain, a Forkhead-associated (FHA) domain and a Ser/Thr kinase domain [
11]. In response to DNA double-strand breaks or replicative stress, CHEK2 is activated by the kinases ATM and ATR [
12]. These proteins catalyze the phosphorylation of threonine 68 of CHEK2, causing its transient dimerization via the FHA domain. This leads to CHEK2
trans-autophosphorylation and its full activation [
13]. Activated CHEK2 monomers phosphorylate, in turn, numerous downstream substrates, including the P53 tumor suppressor, CDC25 family proteins and serine 988 of BRCA1, activating cell-cycle checkpoints and increasing DNA repair efficiency [
14‐
17]. These interactions suggest that CHEK2 may also play a role in breast cancer [
14].
Germline
CHEK2 mutations are associated with breast cancer in different populations. For example, heterozygosity for the well-studied c.1100delC mutation, present in 1.4% of the Finnish population and in 0.2% of the Polish population, confers a relative risk for developing breast tumors of about 2 for women and 10 for men [
18,
19] Likewise, the variant Ile157Thr, present in 5.3% of the Finnish population and in 4.8% of the Polish population, confers a relative risk of breast cancer of 1.5 [
20,
21].
However, very few groups have studied the entire
CHEK2 gene in HBC [
22‐
25]. It is essential to establish a causal link between sequence variants and CHEK2 function. Little is known about the impact of missense mutations on protein function, although substitutions in the FHA domain and the kinase domain have been shown to abolish activity [
22,
26,
27]. In this study, we screened the whole
CHEK2 coding sequence for mutations in non-
BRCA HBC families and a control population without any family history of breast cancer. Point mutations were evaluated by
in silico analyses and an
in vitro kinase activity test.
Discussion
We found strong evidence of an association between CHEK2 variants and HBC, with an OR of 5.18. Of 16 different mutations, 9 were unreferenced variants. This demonstrates that, in populations without founder mutations, an aggregate of rare variants makes CHEK2 an appreciable breast cancer risk gene.
The functional consequences of missense variants can be difficult to establish, and in estimating associated risks it is important to separate deleterious from neutral variants. We were unfortunately unable to complement the functional data presented here with a study of the cosegregation of these variants with cancer, because only the index case was available for analysis in the majority of families.
Missense variant Ser39Phe was predicted as probably deleterious by two of the three scores (SIFT and Align-GVGD), but exhibited WT-like kinase activity. This discordance may suggest that not all deleterious changes in the CHEK2 protein can be revealed by the in vitro kinase activity test, most notably for changes outside the catalytic domain. Changes affecting interactions with upstream activators such as ATM, for example, may not be detectable by our measure. In contrast, Lys224Glu was predicted to be a tolerable change by the three scores, but exhibited null kinase activity, demonstrating the complementarity of those two approaches.
Further functional tests, such as expression in eukaryotic cells, followed by measures of activation by DNA strand breaks, protein stability and interaction with cellular partners may be necessary to appreciate all effects of these mutations, especially for those where the in silico and in vitro conclusions differ. We thus retain this variant as potentially deleterious, unlike Lys244Arg, which was characterized as benign by all measures.
The association between the
CHEK2 gene and breast cancer risk has been supported mainly by case-control studies of founder mutations such as 1100delC, I157T (frequent in northern and eastern Europe) or the Polish founder mutation IVS2+1 G > A (c.444+1G > A) [
19,
20,
24,
43,
44]. In our population, only one of these founder mutations was observed, accounting for one-third of deleterious mutations. Analysis of the entire coding sequence was necessary to capture the majority of the different mutations present. This might be the case for other populations where the frequency of the
CHEK2 founder mutations is low.
In Table
4, to give an overview of
CHEK2 contribution to breast cancer, we summarize the results of 36 different case-control studies from different countries where the presence of variants was assessed by allele-specific sequencing or DNA sequencing of the entire gene. The ORs of breast cancer from the different studies of c.1100delC are similar, regardless of the selection of cases, with a combined OR of 2.77. We also found comparable results for the other protein-truncating mutation c.444+1G > A, which is less frequent but has an OR similar to that for c.1100delC. No positive association with HBC was observed, possible due to the very low frequency of the variant in both cases and controls. The frequent variant I157T was associated with lower ORs than null mutations. Although this variant has been associated with breast cancer risk in early-onset or unselected cases, in our study it did not exhibit a significant association with HBC. Although the frequency of these deleterious mutations was different among populations, the ORs associated with breast cancer were consistent for the two null mutations and lower for the missense mutation. These data were collected using allele-specific sequencing, suggesting that testing for
CHEK2 founder mutations is cost-effective in some populations because the variants are sufficiently common and the test is relatively inexpensive. Consequently, however, these techniques exclude mutations present elsewhere in the gene.
Table 4
Odds ratio for breast cancer among women with CHEK2 variants
c.1100delC | | Mixed populations | All studies | 0.33% (559/166,596) | 0.90% (861/94,076) | 2.77 (2.49 to 3.08)** |
| | AJ, AUS, B, CDN, CZ, DK, E, FIN, G, IRL, KP, NL, S, UK, USA | UBC | 0.31% (2,502/80,168) | 0.80% (464/58,290) | 2.56 (2.19 to 2.99)** |
| | B, CDN, CZ, D, FIN, NL, S, UK, USA | HBC | 0.39% (111/28,402) | 1.27% (215/17,000) | 3.29 (2.61 to 4.14)** |
| | D, USA, AUS, CDN | EOBC | 0.14% (14/9,846) | 0.46% (21/4,588) | 2.77 (1.23 to 6.26)** |
| | AJ, DK, G, FIN, NL, RUS, UK, USA | BBC | 0.43% (130/29,936) | 1.35% (142/10,496) | 3.17 (2.49 to 4.03)** |
c.444+1G > A (IVS2+1G > A) | | Mixed populations | All studies | 0.19% (82/42,266) | 0.66% (217/33,142) | 3.45 (2.67 to 4.46)** |
| | PL | EOBC | 0.19% (49/25,426) | 0.59% (91/15,338) | 3.10 (2.19 to 4.39)** |
| | D, BY | HBC | 0.12% (3/2,586) | 0.26% (4/1,536) | 2.25 (0.5 to 10.08) (NS) |
| | D, BY, PL | UBC | 0.20% (28/14,254) | 0.61% (113/18,604) | 3.12 (2.06 to 4.72)** |
I157T | | Mixed populations | All studies | 2.14% (1,031/48,268) | 3.11% (1062/34,128) | 1.56 (1.43 to 1.70)** |
| | PL, USA, AUS, CDN | EOBC | 1.46% (270/18,432) | 1.99% (215/10,780) | 1.59 (1.32 to 1.91)** |
| | D, BY, FIN, NL, USA | HBC | 1.77% (116/6,544) | 1.51% (47/3,118) | 0.89 (0.60 to 1.20) (NS) |
| | By, CZ, D, FIN, PL, USA | UBC | 2.04% (644/31,600) | 2.95% (807/27,346) | 1.48 (1.33 to 1.65)** |
Whole-gene studies | | USA | UBC | 0.57% (24/4,210) | 0.50% (4/800) | 0.88 (0.3 to 2.55) (NS) |
| | D | HBC | 0.50% (18/3,630) | 3.00% (31/1,032) | 6.38 (3.54 to 11.5)** |
| | CZ | HBC | 1.39% (19/1,366) | 2.01% (27/1,346) | 1.46 (0.80 to 2.65) (NS) |
| | F | HBC | 0.39% (4/1,026) | 1.58% (16/1,014) | 4.15 (1.38 to 12.50)** |
| | USA, CDN, AUS | EOBC | 1.84% (40/2,178) | 4.91% (6/2,484) | 2.76 (1.65 to 4.60)** |
Because the c.1100delC allele does not seem to be present in southern Europeans or in most non-Caucasian populations [
45‐
47], other research groups have used full-gene sequencing to determine whether other variants contribute to breast cancer risk. There is a positive association between
CHEK2 variants and HBC in the Australian, Canadian, North American, German and now French, but not Czech Republic, populations [
22‐
25]. This suggests that
CHEK2 analysis in populations where the common founder mutations are rare requires screening of the entire sequence.
Narod's [
48] recent review supports the view that testing non-
BRCA HBC families for mutations in
CHEK2 can provide useful information to evaluate the risk of breast cancer and suggests that the relatively high cost of sequencing makes only the targeted search of frequent mutations cost-effective. In certain populations, one or a few mutations do indeed capture the majority of
CHEK2 variants associated with cancer risk. In most regions, however, this allele-specific approach is inadequate and a full-resequencing strategy should be considered. The rapidly falling cost of resequencing, as well as alternate techniques, should make this possible.
Conclusions
The usefulness of the information gained from genetic analysis of CHEK2 is currently a matter of debate. As we have discussed, the risk of breast cancer for a woman with a null mutation in this gene is increased two- to fivefold. Increased breast surveillance may be proposed for carriers, but when counseling a family with many breast cancer cases, only some of whom carry the CHEK2 mutation, it is unclear what advice may be given to noncarriers. Collecting research information on CHEK2 mutations, however, serves to advance our understanding of the contribution of this gene to hereditary cancer risk.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AD contributed to the sequencing of CHEK2, designed and performed the kinase activity, participated in the in silico analyses and drafted the manuscript. YB provided expert technical advice, and helped to draft the manuscript. NU designed the study, participated in the sequencing of CHEK2 and provided expertise for the in silico analyses. YJB supervised the study and helped to draft the manuscript. All authors read and approved the final manuscript.