Elevated Levels of Somatic Mutation in a Manifesting BRCA1 Mutation Carrier

Abstract

Homozygous loss of activity at the breast cancer-predisposing genes BRCA1 and BRCA2 (FANCD1) confers increased susceptibility to DNA double strand breaks, but this genotype occurs only in the tumor itself, following loss of heterozygosity at one of these loci. Thus, if these genes play a role in tumor etiology as opposed to tumor progression, they must manifest a heterozygous phenotype at the cellular level. To investigate the potential consequences of somatic heterozygosity for a BRCA1 mutation demonstrably associated with breast carcinogenesis on background somatic mutational burden, we applied the two standard assays of in vivo human somatic mutation to blood samples from a manifesting carrier of the Q1200X mutation in BRCA1 whose tumor was uniquely ascertained through an MRI screening study. The patient had an allele-loss mutation frequency of 19.4 × 10⁻⁶ at the autosomal GPA locus in erythrocytes and 17.1 × 10⁻⁶ at the X-linked HPRT locus in lymphocytes. Both of these mutation frequencies are significantly higher than expected from age-matched disease-free controls (P < 0.05). Mutation at the HPRT locus was similarly elevated in lymphoblastoid cell lines established from three other BRCA1 mutation carriers with breast cancer. Our patient’s GPA mutation frequency is below the level established for diagnosis of homozygous Fanconi anemia patients, but consistent with data from obligate heterozygotes. The increased HPRT mutation frequency is more reminiscent of data from patients with xeroderma pigmentosum, a disease characterized by UV sensitivity and deficiency in the nucleotide excision pathway of DNA repair. Therefore, this BRCA1-associated breast cancer patient manifests a unique phenotype of increased background mutagenesis that likely contributed to the development of her disease independent of loss of heterozygosity at the susceptibility locus.

Introduction

One of the greatest advances in our understanding of cancer etiology has been the identification of subsets of patients exhibiting hereditary susceptibility to the disease. Cancer susceptibility occurs in two distinct forms. The first involves very rare, usually recessive disorders with very high cancer incidence that often have a spectrum of developmental symptoms detectable at birth or in early childhood. Examples include such syndromes as ataxia telangiectasia (AT), Fanconi anemia (FA) and xeroderma pigmentosum (XP), diseases that have been known for decades. Cloning and characterization of the underlying genetic defects in these diseases has shown that they are associated with deficiencies in the various processes of DNA repair. The second type of cancer susceptibility is more subtle, usually associated with cancer incidence in early adulthood, with few or no other phenotypic manifestations. These diseases are considered to be genetically dominant and, as they occur in patients with no other symptoms, their tumors are often considered to be sporadic in the absence of a considerable family history or molecular analysis. Examples of these disorders include Li-Fraumeni syndrome and Lynch syndrome II or hereditary non-polyposis colorectal cancer (HNPCC). The underlying defect in these diseases has been identified as heterozygosity for inactivation of recessive oncogenes known as tumor suppressors. Thus, although the tumor-promoting phenotype is recessive at the cellular level, the high incidence of somatic segregation of heterozygous alleles causes it to be effectively dominant at the organismal level. For many years it has been speculated that there might be some connection between these two types of cancer susceptibility.

The breast cancer susceptibility syndromes associated with mutational inactivation of the BRCA1 and BRCA2 genes are considered to be of the second type of cancer-prone diseases, almost by definition establishing these genes as tumor suppressors. They have also been implicated in DNA repair processes, however, suggesting that they may, in fact, be “mutator” genes like those responsible for the recessive cancer syndromes.¹ This situation became clearer when it was recently discovered that homozygosity for mutations in the BRCA2 gene was responsible for FA patients of complementation group D, identifying it also as the FANCD1 gene.² The tumor suppressor genes that are the genetic basis of HNPCC are known to be involved in post-replicative repair of base mismatches in DNA³ and the p53 gene, the basis of Li-Fraumeni syndrome, is involved in repressing gene amplification,⁴ so both of these syndromes could be considered as “mutator” gene syndromes as well.

The possibility that BRCA1 heterozygotes manifest a cellular phenotype promoting carcinogenesis has been investigated with conflicting results; most studies showing no hypermutability in these cells or individuals,^5,6 while others have reported high spontaneous frequencies of mutation⁷ and/or hyperinducibility.⁸

There are two widely applied methods of measuring somatic mutation in humans, both blood cell-based.⁹ The clonogenic HPRT assay involves chemical selection for inactivation of the X-linked housekeeping gene hypoxanthine-guanine phosphoribosyltransferase in cultured lymphocytes. The flow cytometric glycophorin A (GPA) assay is based on detection and quantitation of somatic “allele loss” mutations at the glycophorin A locus on chromosome 4. The GPA assay is potentially sensitive to a broad spectrum of mutational mechanismus, including point mutation, small insertions and/or deletions, chromosomal aneuploidy, epigenetic gene inactivation, homologous or non-homologous recombination.¹⁰ The HPRT assay is sensitive to point mutations, small insertions and deletions, but large deletions and chromosome loss are inviable, and recombination appears to be suppressed, probably due to X-inactivation.

Previous studies have shown an association between human in vivo somatic mutation level and elevated risk of cancer,^11–14 particularly in the so-called “DNA repair deficiency” diseases AT, FA and Bloom syndrome, which show 10-, 50- and 100-fold increases in GPA mutation frequencies, respectively.¹⁵ Epidemiological evidence also suggests that the otherwise asymptomatic heterozygotes for the radiation-sensitivity syndrome AT are predisposed to breast cancer.^16,17 We recently reported the detection of a mammographically undetectable breast tumor in an MRI pilot screening program of young, high-risk women.¹⁸ In the present study, mutation frequencies (M_f) at the GPA and HPRT reporter loci are measured in this breast cancer patient, who is now known to be a carrier for an inactivating mutation of the BRCA1 gene.

Materials and Methods

The patient was a 36-year-old woman with strong family history of breast cancer (two affected first degree relatives) recruited into a clinical trial of MRI screening for young woman at high risk for breast cancer with dense breast tissue.¹⁸ She was subsequently found to carry a Q1200X premature termination mutation in the BRCA1 locus.¹⁹ Despite recent negative mammography, gadolinium enhancement images revealed a small, 1 cm lesion in the upper-outer quadrant of the left breast, identified pathologically as an infiltrating ductal carcinoma. The patient underwent a modified radical mastectomy of the left breast and chose to also undergo a contralateral prophylactic total mastectomy. Blood was obtained for analysis with consent under Magee-Womens Hospital/University of Pittsburgh IRB # MWH-94-108.

HPRT and GPA somatic mutation analyses

The HPRT and GPA somatic mutation assays were performed on blood samples obtained from the patient and concurrent disease-free controls. The clonogenic HPRT assay was performed by the method of O’Neill et al.²⁰ T lymphocytes were stimulated to proliferate in culture and enzyme-deficient mutant clones selected in the presence of the toxic nucleoside analog 6-thioguanine. Results for the concurrently analyzed normal control, an M_f of 5.0 × 10⁻⁶, were consistent with our previous experience with this individual (P = 0.45, z test), who has been analyzed 6 other times in our laboratory, yielding an average M_f of 8.6 ± 3.3 × 10⁻⁶ (mean ± standard deviation).

Briefly, the GPA assay involves labeling of the two polymorphic forms of the erythrocyte cell-surface protein glycophorin A with monoclonal antibodies followed by flow cytometric analysis to quantify cells with variant phenotypes associated with bone marrow somatic mutation.²¹ The patient had the most informative genotype for this analysis, heterozygosity for the MN blood group. Results for the concurrently analyzed normal control (total GPA M_f, 12.6 × 10⁻⁶, allele loss 6.0 × 10⁻⁶, loss and duplication 6.6 × 10⁻⁶) were consistent with our previous experience with this individual (P = 0.17, 0.28 and 0.17 for total GPA M_f, allele loss frequency and loss and duplication frequency, respectively, z tests), who has been analyzed 11 other times in our laboratory, yielding a total GPA M_f of 9.5 ± 1.3 × 10⁻⁶, an allele loss frequency of 5.1 ± 0.8 × 10⁻⁶ and a loss and duplication frequency of 4.4 ± 0.8 × 10⁻⁶.

Controls

Normal GPA M_f for comparison were derived from our control database, which is under continuous revision and supplementation, and contains historical as well as contemporaneous data. Several subsets of these data have been published.^22–24 Normal HPRT M_f for comparison are derived from a number of published reports^25–31 as well as our own contemporaneous data. GPA M_f from homozygous and heterozygous FA patients were taken from the report of Sala-Trepat et al.³² and our own ongoing analyses^33,34 of clinical samples and samples provided from the International Fanconi Anemia Registry (IFAR). HPRT M_f from homozygous and heterozygous FA patients were derived from Vijayalaxmi et al.³⁵ and Sala-Trepat et al.³² GPA allele loss M_f from XP homozygotes were reported in Langlois et al.¹⁵ HPRT M_f in XP homozygotes and heterozygotes were compiled from a number of reports.^12,36–42

Lymphoblastoid cell lines

Three lymphoblastoid cell lines from breast cancer patients confirmed as heterozygous for inactivating mutations in the BRCA1 gene were obtained from the Coriell Cell Repositories (Camden, NJ): GM13708, GM13709 and GM13712, as well as 9 control cell lines: AG09393, AG09980, AG10111, GM00946, GM01814, GM03797, GM05380, GM14448 and GM14820.

Statistical analysis

Analysis of individual results in the context of control populations was done with the z test at ∝ = 0.05 on ln-transformed data. Comparisons between populations were performed with the t test assuming unequal variances on similarly transformed data or the nonparametric Mann Whitney U test at the same level of significance.

Results

Somatic mutation at the autosomal GPA locus

The patient’s total GPA M_f was found to be 28.2 × 10⁻⁶. M_f at this locus are known to be significantly age-dependent,^22,43 so the patient’s result is compared with normal women ± 2 years from her age in Fig. 1, and against the M_f of our entire normals database, age-adjusted, in Table 1. Although she appears to have a somewhat elevated total GPA M_f (Fig. 1), in the 91^st centile for the entire control database, it is not significantly higher than that expected for her age (Table 1). The GPA assay also allows for the straightforward characterization of mutations into two broad classes; those arising by simple allele loss and those arising by allele loss accompanied by duplication of the remaining allele.^9,10,21 The patient’s allele loss M_f was 19.4 × 10⁻⁶, and her loss and duplication M_f was 8.8 × 10⁻⁶ (note that the two add up to her total GPA M_f). It can now be seen that the elevation in the patient’s M_f is confined to the allele loss class of variation, where it is unambiguously significant (Fig. 1 and Table 2).

Figure 1.

GPA M_f of the hereditary breast cancer patient and 18 age-matched controls (35–39 years). (a) Total GPA: bar indicates average M_f of control population, 14.0 ± 5.6 × 10⁻⁶. The patient’s result is significantly higher than the M_f of this limited control population (P = 0.005). (b) Allele loss: bar indicates average M_f of control population, 6.6 ± 3.4 × 10⁻⁶. The patient’s M_f is significantly higher than that of the controls (P = 0.0001). (c) Loss and duplication: bar indicates average M_f of the control population, 7.4 ± 2.8 × 10⁻⁶. The patient’s result is not significantly different from the M_f of these controls (P = 0.30).

Table 1.

GPA mutation frequencies for normal and DNA repair-deficient patients and populations

Population	GPA Mf (× 10−6)			Patient
	N	Mean ± SD	Range	Percentile	P1
a) all mutant classes
Patient		28.2
Normal controls	644	16.6 ± 16.32	0.19–79.3	0.91	0.22
FA patients	63	455.8 ± 751.63	49.4–4257.4	0.02	0.022
FA carriers	36	31.9 ± 36.74	6.2–90.2	0.66	0.42
b) allele loss mutants only
Patient		19.4
Normal controls	644	6.7 ± 4.35	0.01–34.5	0.97	0.001
FA patients	63	379.9 ± 737.96	24.6–4084.0	0.02	0.047
FA carriers	36	10.4 ± 8.07	1.0–37.4	0.90	0.11
XP patients	7	4.6 ± 4.3	1.0–2.2	1.00	0.028
c) allele loss and duplication mutants only
Patient		8.8
Normal controls	644	9.2 ± 11.88	0.01–152.3	0.67	0.47
FA patients	63	74.0 ± 56.39	6.2–270.4	0.03	0.016
FA carriers	36	20.5 ± 31.710	4.0–152.8	0.43	0.31

¹z-test for patient data

²excluding outlier with total GPA M_f of 6.7 × 10⁻³

³excluding outliers with total GPA M_f of 7.4, 8.0 and 9.9 × 10⁻³

⁴excluding outlier with total GPA M_f of 2.7 × 10⁻⁴

⁵excluding 6 outliers with GPA allele loss M_f of 4.5, 4.8, 7.1, 7.7, 7.8, and 16.2 × 10⁻⁵

⁶excluding 3 outliers with GPA allele loss M_f of 7.0, 7.9 and 9.8 × 10⁻³

⁷excluding outlier with GPA allele loss M_f of 4.4 × 10⁻⁵

⁸excluding outlier with GPA loss and duplication M_f of 8.7 × 10⁻⁴

⁹excluding 2 outliers with GPA loss and duplication M_f of 3.2 and 3.5 × 10⁻⁴

¹⁰excluding outlier with GPA loss and duplication M_f of 2.6 × 10⁻⁴

Table 2.

HPRT mutation frequencies for normal and DNA repair-deficient patients and populations

Population	HPRT Mf (× 10−6)			Patient
	N	Mean ± SD	Range	Percentile	P1
Patient		17.1
Normal controls	482	7.5 ± 4.92	0.32–6.4	0.95	0.049
FA patients	28	5.8 ± 4.63	0.20–5.4	0.97	0.083
FA carriers	27	12.0 ± 5.04	5.0–5.0	0.90	0.13
XP patients	16	15.0 ± 10.7	1.8–6.0	0.65	0.29
XP carriers	7	11.9 ± 8.7	3.4–0.4	0.88	0.19

¹z-test for patient data

²excluding 10 outliers with HPRT M_f of > 3.1 × 10⁻⁵

³excluding outlier with HPRT M_f of 2.6 × 10⁻⁵

⁴excluding outlier with HPRT M_f of 6.8 × 10⁻⁵

Somatic mutation at the X-linked HPRT locus

The patient’s HPRT M_f was found to be 17.1 × 10⁻⁶. Once again, in vivo M_f at the HPRT locus are known to be age-dependent,^25,44,45 so the patient’s result is compared to normal women ± 2 years from her age in Fig. 2, and against the M_f of our entire normals database, age-adjust ed, in Table 2. Compared to the subset of age-matched controls, the patient’s M_f does not appear to be unusually high (Fig. 2), but in comparison with the entire database it does just reach significance (Table 2). To confirm this result, we obtained lymphoblastoid cell lines from 9 disease-free controls and 3 additional breast cancer patients with known inactivating mutations in the BRCA1 gene (all nonsense mutations resulting in premature truncation of the protein product). The HPRT M_f for the normal cell lines was 8.8 ± 6.8 × 10⁻⁶, not significantly different from our in vivo controls (P = 0.69). The mean HPRT M_f for the 3 heterozygous BRCA1 patient cell lines (Fig. 2) was 16.7 ± 1.2 × 10⁻⁶, not significantly different from the in vivo result from our patient (P = 0.63), but signification higher than that of the in vitro controls (P = 0.009). This effect is accentuated by combining the in vivo and in vitro data (P < 0.0001).

Figure 2.

HPRT M_f of the hereditary breast cancer patient, 18 age-matched controls (35–39 years), 3 lymphoblastoid cell lines derived from hereditary breast cancer patients and 9 lymphoblastoid cell lines derived from disease-free controls. The solid bars represent the average M_f for the control population (9.7 ± 5.8 × 10⁻⁶), heterozygous BRCA1^+/− cell lines (16.7 ± 1.2 × 10⁻⁶), and control cell lines (8.8 ± 6.8 × 10⁻⁶), respectively. The patient’s result itself is not significantly higher than that of this restricted set of controls (P = 0.092), but the average M_f of the 3 patient-derived lymphoblastoid cell lines is, both alone (P = 0.048) and with the addition of the patient’s lymphocyte-derived data (P = 0.022). The lighter line represents the average M_f of the entire control database age-adjusted to that of the patient (Table 2).

Discussion

In vivo somatic M_f in a breast cancer patient heterozygous for an inactivating mutation in the BRCA1 gene was significantly elevated over disease-free controls; 1.7-fold in the case of GPA M_f and 2.3-fold in the case of in vivo HPRT M_f. The latter result was confirmed in a study of lymphoblastoid cell lines derived from breast cancer patients, which exhibited a 1.9-fold increase in HPRT M_f over similar controls. These increases are similar to those observed in analyses of M_f in blood samples from sporadic cancer patients: 1.7-fold increases in both HPRT¹² and GPA M_f.¹⁴ In breast cancer patients these increases in M_f are 1.3-fold for HPRT⁴⁶ and 1.5-fold for GPA.⁴⁷ In our hereditary breast cancer patient, this increase becomes 2.9-fold if only the “allele loss” class of GPA mutation is considered, a pattern that is consistent with induced mutation by most genotoxic exposures, including ionizing radiation,⁹ and the elevated spontaneous M_f observed in patients with AT.²⁴ This pattern suggests that haploinsufficiency for BRCA1 does not affect the incidence of mitotic recombination (including gene conversion) or chromosome missegregation, as they result in allele loss and duplication. Possible mechanisms associated with the observed increase in simple allele loss still range from point mutations and small, intragenic insertions and deletions to large-scale, cytogenetically detectable chromosome deletion and even whole chromosome loss,^10,21 however, since we observed a comparable increase in mutation frequency at the X-linked HPRT locus, and major deletions and chromosome loss are inviable at this locus,⁴⁸ it would appear that BRCA1 heterozygosity is associated with gene-specific rather than regional mutational susceptibility.

In patients with FA, we have found a consistent increase in allele loss M_f, but a variable increase in the frequency of loss and duplication.^33,34 Since the BRCA1 gene has been functionally linked to the FA repair system,¹ deficiency for this gene might be expected to produce a similar phenotype. As can be seen from Table 1, however, the GPA M_f of our BRCA1 heterozygote was significantly lower than those seen in FA patients, both in total and in each mutational subclass. A more valid comparison, however, is to compare the patient’s results to those of FA heterozygotes, who do seem to have a subtle mutational susceptibility phenotype.^32–34 While the overall increase in GPA M_f in our patient is certainly consistent with that observed in FA heterozygotes, the pattern does not match, with the patient’s increase occurring in the allele loss class and the FA heterozygotes primarily increased in loss and duplication M_f.

Somewhat surprisingly, FA patients do not exhibit an increased M_f at the HPRT locus,³² although they have a significant shift in their HPRT mutational spectrum towards gene deletions.^49,50 These two apparently contradictory findings are rationalized by invoking the well-established limited viability of these cells, especially in the presence of genotoxic agents.^51,52 Indeed, in the present analysis FA patients actually exhibit a significantly lower HPRT M_f than controls (P = 0.037) (Table 2). FA heterozygotes, however, exhibit an increased HPRT M_f, which, although not significant in either of the original reports,^32,35 is highly significant in our pooled analysis (P < 0.001). This increase in HPRT M_f is consistent with those observed in our BRCA1 carriers, both in vivo (Table 2) and in vitro (P = 0.10).

Another possible explanation for the increased mutation frequency we have observed in BRCA1 heterozygotes is an effect on the nucleotide excision DNA repair pathway (NER). This pathway repairs DNA damage that causes a distortion in the DNA helix, notably intrastrand linkages caused by UV irradiation and bulky adducts such as those caused by exposure to polycyclic aromatic hydrocarbons, and deficiency in this DNA repair system is associated with the cancer-predisposing disease XP.⁵³ Overexpression of BRCA1 has been shown to enhance NER capacity,⁵⁴ probably through transcriptional regulation of NER genes.⁵⁵ Several studies have suggested that breast cancer patients have lower NER capacity than the normal population,^56–58 and we have shown that early stage breast tumors are consistently deficient in this type of DNA repair.⁴⁷

The phenotype of XP patients seems to be the opposite of that of FA patients: they exhibit significantly increased HPRT M_f,^12,36–42 but there is no evidence of an effect on GPA M_f¹⁵ (Tables 1 and 2). In our pooled analysis, the increase in HPRT M_f is highly significant (P = 0.008), and is consistent with that of our BRCA1 carrier patient (Table 2) and data from the heterozygous BRCA1 lymphoblastoid cell lines (P = 0.23). XP heterozygotes also exhibit a slightly increased HPRT M_f,^37–39 although it does not reach statistical significance (P = 0.16). Since XP patients have been identified with as much as 50% residual NER activity,⁵³ it is reasonable that heterozygotes should express a detectable phenotype, but, since the HPRT M_f increase in XP homozygotes is only on the order of 2-fold it would require analysis of a much larger population of heterozygotes to achieve statistical significance. Again, however, the results from our patient (Table 2), and the BRCA1 carrier cell lines (P = 0.22) are not significantly different from those of the population of XP carriers.

BRCA1 carriers therefore manifest a unique phenotype with respect to mutational susceptibility that may be related to their tumor incidence. A more significant deficiency in BRCA1 activity arises during tumorigenesis when the locus undergoes loss of heterozygosity, but this is likely to play a greater role in tumor progression.

Abbreviations/acronyms

AT

ataxia telangiectasia

FA

Fanconi anemia

GPA

glycophorin A

HNPCC

hereditary non-polyposis colorectal cancer

HPRT

hypoxanthine-guanine phosphoribosyltransferase

M_f

mutation frequency

MRI

magnetic resonance imaging

NER

nucleotide excision repair

XP

xeroderma pigmentosum