Association of common ATM polymorphism with bilateral breast cancer



The ATM kinase has an essential role in maintaining genomic integrity. Loss of both ATM alleles results in ataxia-telangiectasia (A-T), a rare autosomal recessive neuroimmunologic disorder associated with cancer susceptibility. Individuals heterozygous for germline ATM mutations have been reported to have an increased risk for malignancy, in particular, female breast cancer. In the current study, a full mutation analysis of the ATM gene was carried out in patients from 121 breast or breast-ovarian cancer families. We discovered that the combination of 5557G→A in cis position with IVS38-8 T→C was associated with bilateral breast cancer (OR = 10.2; 95% CI = 3.1–33.8; p = 0.001). As the 5557G→A change has been reported to affect an exonic splicing enhancer, we hypothesized that the observed composite allele could have some effect on the correct splicing of exon 39. However, no aberrant transcripts were detected, but ATM expression levels of lymphoblast cell lines from heterozygous carriers of this combination allele were lower than from noncarriers (p = 0.09). Lowered gene expression levels may have direct influence on the activities in DNA damage recognition and response pathways, as well as other genome integrity maintenance functions. Based on the results, we propose a cancer risk-modifying effect for the ATM 5557G→A, IVS38-8T→C composite allele. © 2005 Wiley-Liss, Inc.

The ATM protein kinase is essential for the maintenance of genome integrity. It is a key activator of the cellular responses to DNA double-strand breaks through subsequent phosphorylation of central players in various branches of the DNA damage response pathways, including p53, MDM2, CHK2, BRCA1, SMC1 and NBS1.1 In addition, ATM has an important role in the maintenance of telomeres.2 The loss of both functional ATM alleles results in a severe hereditary disorder, ataxia-telangiectasia (A-T). Patients with A-T typically show progressive neuronal degeneration, immunodeficiency, chromosomal instability, radio- sensitivity and an increased risk of cancer.3 Although A-T is an autosomal recessive disease, individuals heterozygous for ATM mutations have been reported to have certain phenotypic effects, including an increased risk for female breast cancer.4, 5, 6 Yet the evidence regarding the role of ATM as a breast cancer susceptibility gene has been contradictory due to the low frequency of ATM mutations observed in breast cancer patients in the general population. There are two possible explanations for this controversy. First, the analysis of cases unselected for family history of breast cancer might be an inefficient way to detect ATM mutations. Second, it may be that only certain ATM mutations predispose to breast cancer.7

We have previously analyzed cases of 162 families displaying signs of hereditary susceptibility to breast cancer for the occurrence of ATM germline mutations originally identified in Finnish A-T families.8, 9 Of 8 different A-T-related mutations, only 6903insA (leading to stop at codon 2372) and 7570G→C (Ala2524Pro) were associated with increased breast cancer risk. However, the overall frequency of these 2 mutations was low, as 6903insA was observed no more than in 1 and 7570G→C in 2 families. Therefore, we wanted to investigate whether mutations apart from those identified in A-T patients could explain an additional fraction of hereditary breast cancer. For this purpose, a full mutation analysis of the ATM gene was carried out in 121 breast and breast-ovarian cancer families.

Material and methods


A total of 185 breast or ovarian cancer patients belonging to 121 families originating from northern Finland were chosen for a full screening of mutations in the ATM gene. Of the studied families, 92 were associated with breast cancer and 29 with breast-ovarian cancer. Inclusion criteria for the 71 high-risk families were the following: 3 or more cases of breast and/or ovarian cancer in first- or second-degree relatives (48 families), or 2 cases of breast and/or ovarian cancer in first- or second-degree relatives, of which at least 1 with early disease onset (≤ 35 years), bilateral disease, or multiple primary tumors (23 families). All of the high-risk families had previously been screened for germline mutations in BRCA1 and BRCA2, and 10 families showed aberrations in these 2 genes.10 The remaining 50 families with moderate disease susceptibility displayed 2 cases of breast and/or ovarian cancer in first- or second-degree relatives. All patients have given their informed consent for obtaining pedigree data and blood specimens for a study on cancer susceptibility gene mutations. Three hundred and six anonymous age-matched cancer-free female Finnish Red Cross blood donors originating from the same geographical region served as controls. Approval to perform the study was obtained from the Ethical Board of the Northern Ostrobothnia Health Care District and the Finnish Ministry of Social Affairs and Health.

Mutation analysis

Using genomic DNA, the coding regions and exon-intron boundaries of the ATM gene were screened for mutations by conformation-sensitive gel electrophoresis (CSGE).11, 12 All findings were confirmed by reamplification of the original DNA sample and direct sequencing with the Li-Cor IR2 4200-S DNA Analysis system (Li-Cor, Lincoln, NE) using the SequiTherm EXEL II DNA Sequencing Kit-LC (Epicentre Technologies, Madison, WI). Oligonucleotide sequences and PCR conditions are available on request.

The cis orientation of the observed 5557G→A, IVS38-8T→C composite allele was determined from the performed CSGE analysis based on the migration patterns of the different genotype combinations. CSGE data were supported by analysis of microsatellite DNA markers (D11S1778, D11S2179) of 9 mutation carriers, 1 homozygous for 5557A allele, from 6 families (data not shown).

mRNA isolation and mutation analysis

The effect of nucleotide changes at the mRNA level was evaluated for the 5557G→A, IVS38-8T→C combination allele. mRNA isolation from lymphoblast cell lines established from the studied patients was done using the FastTrack 2.0 Kit (Invitrogen, Carlsbad, CA), and cDNA was synthesized from 0.5 μg of mRNA with the RevertAid First Strand cDNA Synthesis Kit (Fermentas, Hanover, MD). Primers used for the cDNA-specific amplification of exon 39 were (F) 5′-CAGTGGAGGCACAAAATGTGA-3′ (located in exon 38) and (R) 5′-CTGGCTTCCTTCTTCAAATGC-3′ (located in exon 42).13 Size of PCR products was verified by electrophoresis in 1% agarose gel, and correct splicing by direct sequencing (Li-Cor).

Quantitative RT-PCR

Glyceraldehyde-3-phosphate dehydrogenase (GADPH) and ATM mRNA levels for heterozygous carriers of the 5557G→A, IVS38-8T→C composite allele (n = 7; 2 patients 5557A/A, 5 5557G/A) and noncarriers (n = 6; 3 patients 5557A/A, 3 5557G/G) were measured by real-time quantitative RT-PCR analysis using TaqMan chemistry on an ABI 7700 Sequence Detection System (Applied Biosystems, Foster City, CA) as described previously.14 The sequences of forward (F) and reverse (R) primers and dual-labeled fluorogenic probes (P) for RNA detection were as follows: GADPH (F) 5′-TGTTCGACAGTCAGCCGC-3′, (R) 5′-GGTGTCTGAGCGATGTGGC-3′, (P) 5′-Fam-TCTTCTTTTGCGTCGCCAGCCG-Tamra-3′ and ATM (F) 5′-GTGGAGTTATTGATGACGTTACATGA-3′, (R) 5′-CCCCTGAAAAGTCACAGAGGTC-3′, (P) 5′-Fam-CCAGCAAATTCTAGTGCCAGTCAGAGCA-Tamra-3′. The ATM expression level results were normalized to that of GADPH, a constitutively expressed gene quantified from the same set of samples. The cDNA synthesis and the expression level measurements were performed twice.

Statistical analyses

Statistical comparison of the 2 performed expression level measurements was assessed by paired-sample t-test and the ATM expression levels between the 2 subgroups categorized according to their genetic status were done by independent-sample t-test. Two-sided Fisher's exact or chi-square test was used to determine the statistical significance of other variables (SPSS version 12.0 for Windows; SPSS, Chicago, IL).

Results and discussion

The performed mutation analysis revealed several changes both in the exon and intron regions of ATM (Tables I and II). Only 2 novel nucleotide substitutions were found, of which 1, the 8071C→T transition in exon 57, led to amino acid substitution. This Arg2691Cys substitution is located between the FAT (FRAP/ATM/TRRAP) and catalytic kinase domains. Arg2691Cys was found in one patient diagnosed with breast cancer at age 56, but because of lack of DNA samples from additional family members, cosegregation with the disease phenotype could not be confirmed. All the other amino acid substitutions have been described previously. Although a few of the previous studies have proposed a role for some of these mutations in susceptibility to breast cancer,15, 16 our study did not show any evidence for the cosegregation of these mutations with the cancer phenotype, and the comparison of allele frequencies between cases and controls showed no statistical difference (Table I).

Table I. Sequence Variation Observed in the Exon Regions of ATM
ExonNucleotide changeEffect on proteinAllele frequency1p
  • 1


  • 2

    All positive patients carried both Phe858Leu and Pro1054Arg.

  • 3

    Novel nucleotide change.

Ex9735C→TVal245Val6.6% (8/121)8.8% (27/306)0.453
Ex152119T→CSer707Pro1.7% (2/121)(0/306)0.080
Ex192572T→CPhe858Leu2.5% (32/121)1.6% (5/306)0.693
Ex233150C→TLeu1050Leu0.8% (1/121)ND 
Ex243161C→GPro1054Arg2.5% (32/121)2.6% (8/306)1.00
Ex314258C→TLeu1420Phe1.7% (2/121)2.0% (6/306)1.00
Ex324578C→TPro1526Pro6.6% (8/121)ND 
Ex395557G→AAsp1853Asn36.4% (44/121)35.6% (109/306)0.885
   G homozygotes 56.2% (68/121)G homozygotes 56.9% (174/306) 
   A homozygotes 7.4% (9/121)A homozygotes 7.5% (23/306) 
Ex395558A→TAsp1853Val0.8% (1/121)0.3% (1/306)0.487
Ex578071C→T3Arg2691Cys0.8% (1/121)(0/306)0.283
Table II. Sequence Variation Observed in the Intron Regions of ATM
LocationNucleotide changeAllele frequency in cases1
  • IVS, intervening sequence.

  • 1


  • 2

    All positive individual carried both IVS14−55 and IVS44−61.

  • 3

    Observed always in cis position with 5557A.

  • 4

    Novel nucleotide change.

IVS1a−10A→G48.8% (59/121)
IVS11−16DelT64.5% (78/121)
IVS14−55T→G8.3% (102/121)
IVS15−68T→C4.1% (5/121)
IVS17−56G→A47.1% (57/121)
IVS24−9DelT38.0% (46/121)
IVS25−12InsA62.0% (75/121)
IVS27+40G→A0.8% (1/121)
IVS38−8T→C6.6% (83/121)
IVS38−15G→C7.4% (9/121)
IVS44−61C→G8.3% (102/121)
IVS49−434A→G0.8% (1/121)
IVS62−55T→C50.4% (61/121)
IVS62+60G→A50.4% (61/121)

Some of the previous studies have, however, proposed a phenotypic effect for the common ATM missense alteration 5557G→A (1853Asp→Asn) in exon 39. The 5557A allele has been suggested to modulate the penetrance of hereditary nonpolyposis colorectal cancer in carriers of germline MLH1 and MSH2 mutations.17 Additionally, 5557G→A was the most frequently occurring ATM polymorphism in individuals with idiopathic or radiation-induced ocular telangiectasia of either choroidal or retinal origin, without family history of A-T.18 In the current study, the frequency of 5557G→A in 185 patients of 121 families (5557G allele 0.74; 5557A allele 0.26) did not differ from that in controls (5557G allele 0.75; 5557A allele 0.25). However, the phenotypic features of breast cancer patients heterozygous for 5557G→A in cis position with the IVS38-8 T→C transition deserve some further attention. Both 5557G→A and IVS38-8T→C are known polymorphisms reported by several studies.19, 20, 21 Although in the current investigation IVS38-8T→C was always found in cis position with 5557A (Table I), it has been indicated that IVS38-8T→C occurs also in combination with 5557G.19

The 5557G→A, IVS38-8T→C composite allele displayed only slightly higher frequency in the studied families (6.6%) compared to the controls (4.2%; data not shown), but we observed a statistically significant association with bilateral breast cancer. Among 176 breast cancer patients studied, 16 had a bilateral disease, 5 of which (31.3%) were heterozygous for the 5557G→A, IVS38-8T→C composite allele. The frequency was markedly higher compared to 7 carriers in 160 (4.4%) patients with unilateral breast cancer (OR = 9.9; 95% CI = 2.7–36.5; p = 0.002) and to controls (OR = 10.2; 95% CI = 3.1–33.8; p = 0.001). Of the remaining 11 bilateral cases, 1 was found positive for the CHEK2 1100delC germline mutation, which has been proposed to associate with bilateral breast cancer,22 and 2 cases carried either the previously identified BRCA2 999del5 or TP53 Asn235Ser germline mutation.10, 23 In one of the studied pedigrees, 3 sisters (2 unilateral cases at the age of 47 and 59, and 1 bilateral case at the age of 69) and their maternal cousin (bilateral disease at the age of 52/55) were diagnosed with breast cancer. Both of the bilateral cases and the unilateral case diagnosed at age 59 were found to be 5557G→A, IVS38-8T→C composite allele carriers. The mother of the 3 sisters was an obligate, though disease-free, carrier. In addition to frequent occurrence in bilateral cases, the composite allele cosegregated with the disease phenotype in one family, where all 3 sisters diagnosed with unilateral breast cancer were found to be carriers.

As a previous study by Thorstenson et al.16 has shown that this particular 5557G→A change affects an exonic splicing enhancer (ESE), we hypothesize here that the observed 5557G→A, IVS38-8T→C composite allele could have some effect on the correct splicing of exon 39. It is known that many ATM exons have poor agreement to the splicing consensus sequences and are also small in size, which predicts that efficient splicing is likely to require enhancers to promote splicing at unfavorable sites, or silencers to repress more favorable splice sites nearby.16, 24, 25 In fact, ESEs recognized by serine/arginine-rich proteins, a family of essential splicing factors regulating also alternative splicing, have been identified in every exon of the ATM gene.16, 24 Disruption of these ESEs may alter the correct splicing,26 as might be the result of 5557G→A transition, and the intronic change IVS38-8T→C in relatively close proximity of the intron/exon boundary might strengthen this effect.

The effect of 5557G→A, IVS38-8T→C nucleotide changes on correct splicing was assessed by cDNA-specific amplification of a fragment ranging from exon 38 to exon 42. However, no splicing aberrations were detected, indicating either that the composite allele has no effect on splicing, or that only very small amounts of aberrant products arise, perhaps combined with high instability of these transcripts. In the latter case, the level of aberrant transcripts stays below the sensitivity of our detection method. When comparing the ATM expression levels of lymphoblast cell lines from heterozygous 5557G→A, IVS38-8T→C composite allele carriers (n = 7) with those from noncarriers (n = 6) by quantitative RT-PCR analysis, a difference was observed: the mean ATM expression level normalized to GADPH expression for carriers was 1.54 (95% CI = 0.9039–2.1662), and for noncarriers 2.38 (95% CI = 1.3808–3.3725). The measurements were performed twice, and paired-sample correlations were high (0.971; p < 0.001). Yet the difference in expression levels between the carriers and noncarriers did not reach statistical significance (p = 0.09), which is, at least partially, due to the low number of cell lines available for analysis.

Taking together the results of the current and our previous study,9 we have analyzed the entire ATM protein-encoding region in patients from 121 breast and breast-ovarian cancer families. The mutations associated with increased breast cancer susceptibility, and which were not observed among healthy controls, seemed to be restricted to those reported previously in Finnish A-T patients. However, the overall frequency of these A-T-related mutations seems to be low. For the common 5557G→A ATM variant, we propose a cancer risk-modifying effect when occurring in cis position with the IVS38-8T→C change, as this combination appears to associate with bilateral breast cancer. Our observations suggest that this composite allele results in lowered ATM expression levels, possibly due to an increase in incorrect splicing of exon 39 and the instability of these transcripts. This dosage variation may have direct influence on the activities in DNA damage recognition and response pathways, as well as other genome integrity maintenance functions. Possibly in combination with certain genetic background and/or environmental factors, it could modify the cancer risk by increasing genetic instability or by altering the effect of the normal DNA damage response. The present observations need to be confirmed by additional studies.


The authors thank Drs. Guillermo Blanco, Ulla Puistola, Aki Mustonen and Jaakko Ignatius and Nurse Outi Kajula for their help in patient contacts, and Ms. Kati Outila for performing part of the mutation screening work. The kind participation of all patients has been of utmost importance in performing this study.