A sensitive test for the detection of specific DSB repair defects in primary cells from breast cancer specimens

Authors


Abstract

Increasing evidence indicates that breast cancer pathogenesis is linked with DNA double-strand break (DSB) repair dysfunction. This conclusion is based on advances in the study of functions of breast cancer susceptibility genes such as BRCA1 and BRCA2, on the identification of breast cancer-associated changes regarding the genetics, expression, and localization of multiple DSB repair factors, and on observations indicating enhanced radiation-induced chromosomal damage in cells from predisposed individuals and sporadic breast cancer patients. In this pilot study, we describe a sensitive method for the analysis of DSB repair functions in mammary carcinomas. Using this method we firstly document alterations in pathway-specific DSB repair activities in primary cells originating from familial as well as sporadic breast cancer. In particular, we identified increases in the mutagenic nonhomologous end joining and single-strand annealing mechanisms in sporadic breast cancers with wild-type BRCA1 and BRCA2, and, thus, similar phenotypes to tumors with mutant alleles of BRCA1 and BRCA2. This suggests that detection of error-prone DSB repair activities may be useful to extend the limits of genotypic characterization of high-risk susceptibility genes. This method may, therefore, serve as a marker for breast cancer risk assessment and, even more importantly, for the prediction of responsiveness to targeted therapies such as to inhibitors of poly(ADP-ribose)polymerase (PARP1). © 2008 Wiley-Liss, Inc.

Germline loss-of-function mutations affecting one allele of the BRCA1, BRCA2, TP53, or PTEN genes predispose to breast cancer with high, i.e. up to 10-fold, risk. The more recently established susceptibility genes ATM, CHEK2, NBS1, RAD50, BRIP1, and PALB2 confer a 2-fold increase in breast cancer risk.1, 2 Despite these advances in the identification of breast cancer susceptibility genes, no more than 30–50% of hereditary breast cancer can be explained by mutations in known genes.1, 3 Still, current approaches to predict breast cancer risk primarily rely on genotyping single predisposing genes.4 Thus, direct nucleotide sequencing has remained the gold standard technique for breast cancer risk assessment involving BRCA1 and BRCA2 mutations, even though mutations in the regulatory portion of the genes and large genomic rearrangements that produce deletions of whole exons escape this detection method. Moreover, roughly one third of the sequence alterations represent unclassified variants, i.e. missense mutations with unclear pathogenicity. Most importantly, lack of association between BRCA1 or BRCA2 (BRCA) gene alterations and breast cancer predisposition can be explained by the presence of mutations in known or as yet undiscovered genes that interact with the BRCA1 and/or BRCA2 pathway. Indeed the majority of familial cases are caused by one or more unknown high-penetrance susceptibility genes, and/or a combination of multiple low-penetrance modifier alleles.4, 5 All in all, a magnitude of mechanisms of inactivation for an increasing number of genes are underlying susceptibility to breast cancer. Strikingly however, the 10 established breast cancer susceptibility genes all directly or indirectly play a role in DSB repair.1, 2, 5

In mammalian cells, two major pathways exist that require sequence homologies for the repair of a DSB. Homologous recombination (HR) is a comparatively nonmutagenic homology-directed DSB repair (HDR) mechanism, and also bypasses DNA damage that remains unrepaired until the replication fork encounters the DNA lesion. RAD51 represents the central HR recombinase, which, in a concerted action with several co-factors including BRCA2, catalyzes the strand exchange reaction between the damaged recipient and the homologous donor DNA. The second HDR mechanism, single-strand annealing (SSA), involves the hybridization of complementary single-strands formed after processing of a DSB. SSA requires the pairing activity of RAD52, but is independent of RAD51-mediated strand invasion. In contrast with HR, SSA is nonconservative, because it deletes intervening sequences between repeats.6, 7 In the absence of extended sequence homologies, nonhomologous end joining (NHEJ) can reseal broken DNA ends. NHEJ represents a highly error-prone DSB repair pathway, particularly, when involving mechanisms alternative to the canonical pathway based on the DNA end binding protein complex DNA-PK. NHEJ allows low-fidelity rejoining of broken DNA ends by use of short microhomologies in the vicinity of the DSB, followed by mutagenic deletion of the terminal nucleotides.6–8

NHEJ is considered the major mechanism responsible for oncogenic genome rearrangements such as chromosomal translocations. At least in cell-based model systems, SSA also mediates translocations.9 When deregulated, even HR can cause destabilizing rearrangements involving imperfectly homologous repeat sequences within the genome.10 The high-risk breast cancer susceptibility gene products BRCA1, BRCA2, and p53 are central components of the surveillance system that ensures high-fidelity DSB repair, and, therefore, suppresses genomic instabilities. BRCA1 as a scaffolding factor and BRCA2 as a chaperone for RAD51 filament assembly promote the safest pathway, HR.11 p53 counteracts error-prone HR between divergent sequences.12–14 Even PTEN, which is well-known as inhibitory component of the phosphatidyl inositol 3-kinase (PI3K)/Akt pathway, has recently been connected with HR through regulation of RAD51 transcription.15 BRIP1/BACH1 and PALB2 are binding partners of BRCA1 and BRCA2, respectively, and contribute to their DNA repair functions.16, 17 NBS1 and RAD50 together with MRE11 form a complex, which plays a role in DSB detection and subsequent checkpoint signaling via ATM, tethering of the two DNA substrates, and DSB processing.6, 18 Through post-translational phosphorylation, ATM and CHK2 regulate the activities of various DSB repair proteins including BRCA1 and p53.5

The above listed links between breast cancer susceptibility genes and DSB repair suggest that use of function-based assays has the potential to significantly improve discrimination of high- and low-risk groups. In support of this idea, differences were observed between predisposed individuals carrying a BRCA1 or BRCA2 mutation and controls, when scoring radiation-induced chromosome aberrations or the appearance of micronuclei in peripheral lymphocytes. For comparison, use of the comet assay did not discriminate between women with and without a BRCA1 mutation, possibly because this assay monitors changes in the speed rather than the fidelity of DSB repair. It should also be noted that micronucleus induction was not seen in lymphoblastoid cell lines (LCLs) with a BRCA1 mutation and that there was no strict correlation between the results from the chromosome aberration and micronucleus test.19, 20 These limitations suggest that for superior approaches in risk assessment the development of alternative functional assays is needed. In further, support of a functional approach, there is a growing list of reports which describe inactivation of DSB repair-related genes in sporadic mammary tumors, through multiple mechanisms including gene silencing due to epigenetic phenomena.5 In the light of these and recent data indicating that the DSB repair capacity in the tumor determines responsiveness to conventional and newly emerging tumor therapies,21 functional testing of tumor samples may additionally be useful to direct therapeutic interventions.

To develop an experimental system that allows detection of specific DSB repair defects in breast cancer, we isolated epithelial cells from breast tumor specimens and applied an EGFP-based DSB repair assay system. Different from classical systems relying on the survival and clonal outgrowth of cells under selection pressure, detection of cellular fluorescence enables DSB repair measurements with a fast readout and, therefore, in primary cells. The availability of multiple reporters with different repair (recipient and donor) substrates allows comparative analysis of NHEJ, SSA, and HR activities under identical experimental conditions. Our analyses showed that functional changes with respect to specific DSB repair mechanisms are detectable in these cells. Therefore, this method represents a promising tool for the establishment of a novel biomarker for breast cancer susceptibility and a predictor for targeted therapies.

Abbreviations

BIC, breast cancer information core; DAB, diaminobenzidine; DSB, DNA double-strand break; EMA, human epithelial membrane antigen; EMT, epithelial to mesenchymal cell type; ER, estrogen receptor; FA, Fanconi anemia; HDR, homology-directed DSB repair; HR, homologous recombination; HRP, horse radish peroxidase; LCL, lymphoblastoid cell line; LOH, loss of heterozygosity; NHEJ, nonhomologous end joining; PARP1, Poly(ADP-ribose)polymerase 1; PCR, Polymerase chain reaction; PI3K, phosphatidyl inositol 3-kinase; PR, progesterone receptor; SSA, single-strand annealing.

Material and methods

Tissue specimens, establishment of primary cell cultures and purification of epithelial cells

Carcinoma tissues from primary breast cancer patients were surgically resected in the Department of Surgical Disciplines, All India Institute of Medical Sciences, New Delhi, with prior informed written consent of the patients. Clinical staging followed the classification of American Joint Committee on Cancer revised 6th edition.22 Breast carcinoma tissues were collected and processed for setting up the primary cell cultures as detailed in Bagadi et al.23 Enrichment of epithelial cells was achieved by differential trypsinization as described previously.23, 24 Thereafter fibroblast overgrowth was reduced by selective chemical elimination using geneticin (G418 sulfate, 100 μg/ml; GIBCO BRL, Grand Island, NY) as described elsewhere.25 Finally, epithelial cells were purified using MACS epithelial columns (Miltenyi Biotech GmbH, Bergisch Gladbach, Germany).24 Eluted, purified epithelial cells as well as MDA-MB-157 control cells were plated in tissue culture flasks containing DMEM/F12 medium supplemented with 10% FBS (Gibco Invitrogen Cell Culture, Grand Island, NY), crystallized bovine insulin (5 μg/ml) (Sigma-Aldrich, Bangalore, India), epidermal growth factor (10 ng/ml) (Gibco Invitrogen Cell Culture), antibiotics (1% penicillin-streptomycin) and propagated by periodic subculturing.

Cell lines and cultivation

LCLs were maintained in RPMI 1640 supplemented with 12% FCS, 3 mM L-glutamine (Biochrom, Berlin, Germany). HCC1937 BL was established from a breast cancer patient heterozygous for the BRCA1 germline mutation 5382insC,26 GM13023A from a Fanconi anemia patient, complementation group D1 (BRCA2); GM02253F cells are mutated in the XPD gene (Coriell Cell Repositories). 416MI cells were derived from a normal, healthy individual.

Growth rate and colony formation in soft agar

Population doubling times were established by counting cell numbers at different time periods during tissue culture immediately before repair measurements. To assess the growth potential in anchorage free environment, primary breast cancer cells were grown under substratum-independent conditions. For this purpose single-cell suspensions of purified epithelial cancer cells were dispersed in epithelial cell culture medium containing bacto-agar and cultivated for 2 weeks. The formation of progressively enlarging colonies was analyzed by light microscopy. The human breast cancer cell line MCF7, which forms colonies in soft agar, was used as positive control, the human nonmalignant breast epithelial cell line MCF10A, which does not form any colonies in soft agar, served as negative control.

Immunocytochemistry

Cells grown on coverslips were fixed with acetone and processed for immunocytochemical analysis as described in Bagadi et al.23 Briefly, cells were immunolabeled with primary antibodies (1 μg/ml) followed by biotinylated secondary antibody, (0.8 μg/ml) (DAKO, Produktionsvej, Glostrup, Denmark). Primary antibodies were directed against estrogen and progesterone receptor (PR), namely sc-8002 and sc-810 (Santa Cruz Biotechnology, CA), Human Epithelial Membrane Antigen (EMA), clone E29 (DakoCytomation, Carpinteria, UA) and vimentin, V 6630, Monoclonal Anti-Vimentin antibody, V9 (Sigma Immunochemicals, St. Louis, MO). Cells were then incubated with streptavidin conjugated horse radish peroxidase and the color was developed using diaminobenzidine as chromogen. Slides were counterstained with Mayer's hematoxylin, mounted with DPX mountant and examined under a light microscope.

Mutational analysis of TP53, BRCA1 and BRCA2

Genomic DNA was isolated from breast cancer cells either by the standard phenol-chloroform method or using the Promega Wizard SV Genomic DNA isolation kit. Polymerase chain reaction (PCR) amplified products encompassing exons 5, 6, 7, 8, and 9 of the TP53 gene were analyzed for mutations by direct automated sequencing. The 24 BRCA1 and 27 BRCA2 exons were sequenced completely by use of an Applied Biosystem ABI 377 sequencer (Foster City).

DSB repair assay

To investigate pathway-specific DSB repair activities in primary breast cancer cell cultures, the fluorescence-based test system described in Akyüz et al.27 was applied. Different plasmid mixtures, containing the meganuclease expression plasmid (pCMV-I-SceI) together with one of the recombination substrates (EJ-EGFP, Δ-EGFP/3EGFP, Δ-EGFP/5EGFP) and pBS control plasmid (for DSB repair measurements) or wild-type EGFP plasmid (for determination of transfection efficiencies) were introduced by nucleofection (Amaxa, Cologne, Germany). The cells were, then, cultivated for one generation time (20–43 hr) at 37°C. To analyze different DSB repair pathways in LCLs the corresponding plasmid mixtures were introduced via electroporation27 followed by cultivation for 24 hr. Subsequently, cellular fluorescence was quantified flow cytometrically by use of a FACS®Calibur FACSscan (Becton & Dickinson, Heidelberg, Germany). Recombination frequencies were calculated from the fraction of EGFP positive cells and corrected for the individual transfection efficiencies, determined in parallel each. Transfection efficiencies ranged from 30 to 60% in the breast cancer cell types analyzed, and 13 to 41% in LCLs. The statistical significance of differences was determined using Student's t test for unpaired samples.

Results

Establishment of primary breast cancer cell cultures

Accumulating evidence indicates that changes in the status of a large variety of DSB repair factors contribute to the development of familial as well as sporadic breast cancer.1, 5 In this work, we wanted to understand whether functional testing with respect to defined DSB repair activities can be applied to breast cancer specimens. To this end, five samples were successfully processed for long-term (>3 weeks) primary epithelial cell cultures, enabling purification and characterization of growth characteristics (see below). One of the cultures, Bca11, was established from a 64-year-old patient with ductal carcinoma in situ and a family history of breast cancer. Details of the sporadic tumors are summarized in Table I.

Table I. Clinical and pathological features of sporadic breast tumors used for initiation of cell cultures
Cell culturePatient age (yr)Cancer typeFamily historyTNM stageStage grouping
Bca954Invasive ductalSporadicT2N2M0IIIA
Bca1040Invasive ductalSporadicT2N2M0IIIA
Bca1464Invasive ductalSporadicT2N2M0IIIA
Bca1646Invasive ductalSporadicT3N2M0IIIA

To obtain pure epithelial cell populations we developed a strategy, which combines differential trypsinization,23, 24 selective chemical elimination of fibroblasts,25 and immunopurification using MACS epithelial columns. In the resulting cultures, all the cells displayed the typical morphology of the epithelial lineage and expressed EMA, as is representatively shown for culture Bca10 in Figures 1a and 1b. Bca9, Bca10, and Bca16 were additionally positive for the mesenchymal cell marker vimentin. All the cells grew in multilayers, which indicates loss of contact inhibition and, therefore, malignant potential. When we additionally tested Bca9, Bca10, and Bca11 cultures for anchorage-independent growth, all the cells formed progressively enlarging colonies in soft agar, which is another characteristic feature of transformed cells (Fig. 1c). Further, immunocytochemical analysis revealed that the cells had undergone molecular alterations characteristic of aggressive cancer,28 such as lack of ER and PR expression (Table II). In summary, we established primary breast cancer cell cultures which are likely to reflect the features of the original tumor, which is different from existing breast cancer derived cell lines that had mostly been obtained from secondary tumors or pleural effusions of patients with advanced stages of breast cancer.29

Figure 1.

Epithelial and transformed nature of breast cancer cells. (a) Cellular morphology. MACS column purified Bca10 cells display the typical epithelial cell shape (original magnification ×200). (b) Immunocytochemical analysis. Bca10 cells express EMA (original magnification ×200). (c) Anchorage-independent growth. Bca10 cells form colonies after plating in soft agar.

Table II. Cell Culture Characteristics
Cell cultureProtein marker expression
ERPRp53BRCA1
  • 1

    (C): Cytoplasmic.

Bca9NegativeNegativeNegativePositive
Bca10NegativeNegativeNegativeNegative
Bca11NegativeNegativeNegativeNegative (C)1
Bca14NegativeNegativeNegativeNegative (C)
Bca16NegativeNegativeNegativeNegative (C)

Analysis of DSB repair activities in primary breast cancer cells

Having successfully isolated tumor epithelial cells from breast cancer tissues, we applied the fluorescence-based test system27 on these cultures for comparative functional testing of DSB repair activities. This short-time assay enables analysis of the different DSB repair mechanisms through the use of specifically designed repair substrates (Fig. 2a). These substrates carry differently mutated EGFP genes and I-Sce I meganuclease recognition sites such that targeted cleavage by I-Sce I triggers microhomology-mediated NHEJ on plasmid EJ-EGFP, HDR pathways on Δ-EGFP/3′EGFP (predominantly SSA27), and HR on Δ-EGFP/5′EGFP (mostly gene conversion). Successful DSB repair events restore wild-type EGFP and are, therefore, monitored by FACS analysis as the fraction of green fluorescing cells within the population of nonfluorescent cells.

Figure 2.

Analysis of pathway-specific DSB repair activities in breast cancer cells. (a) Repair substrate design. The EGFP-based DSB repair assay27 monitors recombinative DNA-exchange processes which involve differently mutated EGFP genes (EJ-EGFP, Δ-EGFP, 3′EGFP, 5′EGFP) and restore wild-type EGFP. EJ-EGFP and Δ-EGFP are internally mutated EGFP genes that contain a recognition sequence for the rare-cutting I-Sce I meganuclease. Within the EJ-EGFP plasmid, 5 bp homologies were additionally generated 5′ and 3′ to the I-Sce I recognition sequence such that microhomology-mediated NHEJ can reconstitute the wild-type EGFP gene after targeted cleavage. 5EGFP and 3EGFP represent 3′ and 5′ truncated versions of EGFP, respectively. Plasmid Δ-EGFP/3′EGFP enables analysis of non-conservative HDR pathways, plasmid Δ-EGFP/5′EGFP detects conservative HR. DSB repair events were initiated through the introduction of a targeted DSB after I-SceI meganuclease expression and scored by FACS-analysis of green fluorescent cells. (b) NHEJ analysis. Cells from the control line MDA-MB-157 (MM157; grey columns) and primary epithelial cell cultures from sporadic (Bca9, Bca10, Bca14, Bca16; black columns) and familial breast cancer (Bca11; black column) were transfected with pCMV-I-SceI together with plasmid EJ-EGFP. DSB repair frequencies were determined after one generation time. For better comparison, mean frequencies in MDA-MB-157 cells were set as 100% (absolute value: 0.45 × 10−3). Columns, mean values from two to six measurements; bars, SEM (sporadic)/SD (familial). (c) HDR analysis. Breast cancer cells were processed as in (b), except for the use of substrate Δ-EGFP/3′EGFP (absolute frequency for MDA-MB-157: 3.06 × 10−3). (d) HR analysis. Breast cancer cells were processed as in (b), except for the use of substrate Δ-EGFP/5′EGFP (absolute frequency for MDA-MB-157: 0.34 × 10−3).

Primary breast cancer cultures at passage four to eight and the ER-negative breast cancer cell line MDA-MB-157 were subjected to nucleofection with DSB repair substrate (EJ-EGFP, Δ-EGFP/3′EGFP, or Δ-EGFP/5′EGFP) and I-Sce I expression plasmid (pCMV-I-SceI). Preceding experiments had indicated that transfection efficacy of the recently available method of nucleofection was superior to lipofection with established transfection reagents such as FuGENE, and concomitantly caused negligibly low cytotoxicities. Here, transfection efficiencies of 30–60% were reached for primary cultures and control cells. Following transfection, cells were cultivated for one generation time (20–43 hr) before cellular fluorescence was quantified flow cytometrically. To rule out possible indirect effects on DSB repair related to differences in growth, cell lethality, transcriptional, and translational activities, we performed identical co-transfections including wild-type EGFP control plasmid in parallel to each single measurement. The resulting transfection efficiencies served to normalize recombination frequencies individually.

With respect to the primary cultures Bca9, Bca10, Bca14, and Bca16, originating from sporadic breast cancer, we recorded NHEJ and HR repair frequencies, that did not show statistically significant differences to the values obtained with MDA-MB-157 breast cancer cells (Figs. 2b and 2d). When measuring HDR, Bca10 cells performed repair at a 2.2-fold elevated frequency (p = 0.0202), whereas cultures Bca9, Bca14, and Bca16 again showed no significant difference compared to the control (Fig. 2c). Strikingly, with Bca11 cells from familial mammary carcinoma we noticed a 130.8-fold increase in NHEJ (p = 0.0375) and a 9.6-fold stimulation of HDR (p < 0.0001). Taken together, we observed elevated activities in specific DSB repair pathways for two primary breast cancer cultures.

Analysis of DSB repair activities in cells from predisposed individuals

Mutations in the genes BRCA1 and BRCA2 confer very high breast cancer risk.1 Over the last years functional studies revealed that the products of these genes play important roles in DSB repair.5, 11 Given that significant DSB repair changes were found in Bca10 and Bca11 cells, we next wished to compare these changes to the DSB repair pattern in cells from predisposed individuals. To this end, we applied the different DSB repair substrates and I-Sce I expression plasmid to HCC1937 BL cells, which had been established from a breast cancer patient, carrying the BRCA1 germline mutation 5382insC,30 and to GM13023A cells from a FA patient, complementation group D1, linked to a molecular defect in BRCA2.31 As controls, we used cells from a normal, healthy individual (416MI) and GM02253F cells, mutated in the XPD gene, i.e. with a defect in nucleotide excision repair.

When measuring NHEJ, we found 6.6- and 9.5-fold increased frequencies for HCC1937 BL (p = 0.0378) and GM13023A cells (p = 0.0001), respectively, as compared to 416MI (Fig. 3a). Elevated frequencies were also detected for HDR in both cell lines, however, with marginal significance for HCC1937 BL (HCC1937 BL: 2.9-fold increase, p = 0.056; GM13023A: 3.5-fold increase, p = 0.0011) (Fig. 3b). Concomitantly, HR was 11-fold reduced in GM13023A cells (p = 0.0004) (Fig. 3c). Comparison of HCC1937 BL and GM13023A cells with GM02253F gave similar results, namely a NHEJ increase for both LCLs (HCC1937 BL: 2.8-fold increase, p = 0.0094; GM13023A: 4.0-fold increase, p = 0.0007), and for GM13023A additionally 1.7-fold enhanced HDR (p = 0.0405) and 4.9-fold downregulated HR (p = 0.0132). These data confirmed that the functional changes observed were found in comparison to different control cell lines and reflected a DSB repair rather than a general defect in DNA repair.

Figure 3.

DSB repair activities in LCLs from predisposed individuals.(a) NHEJ analysis. LCLs from predisposed individuals (HCC1937 BL, GM13023A) and control lines (416MI, GM02253F) were transfected with pCMV-I-SceI together with plasmid EJ-EGFP and DSB repair frequencies determined after 24 hr. Mean frequencies in 416MI cells were set as 100% (absolute value: 0.13 × 10−3). Columns, mean values from nine measurements; bars, SEM. (b) HDR analysis. LCLs were processed as in (a), except for the use of substrate Δ-EGFP/3′EGFP (absolute frequency for 416MI: 2.33 × 10−3). (c) HR analysis. Breast cancer cells were processed as in (a), except for the use of substrate Δ-EGFP/5′EGFP (absolute frequency for 416MI: 0.23 × 10−3).

Molecular analysis of the breast cancer susceptibility genes BRCA1, BRCA2 and TP53

BRCA1 and BRCA2 germline mutations are most frequently found in familial breast cancers.32 Since the Bca11 culture was established from a patient with family history of breast cancer, Bca11 cells were screened for BRCA1 and BRCA2 mutations by sequencing all the exonic and neighboring intronic sequences. Interestingly, Bca11 does not harbor any mutation in BRCA1 or BRCA2, although polymorphisms (SNPs) were identified in both genes (Table III). Sequencing of the BRCA1 and BRCA2 genes in Bca10 cells, which also displayed an altered DSB repair pattern, again revealed no mutations but polymorphisms. All the polymorphisms identified in these cultures were heterozygous. The predicted amino acid changes within BRCA1 lie out of the domains involved in the critical interactions with BARD1, BACH1, or the RAD50/MRE11/NBS1 complex, although there is overlap with the region which is required for efficient RAD51 foci formation. Amino acid changes within BRCA2 were found out of the RAD51, PALB2, and EMSY binding sites.11, 17, 33–35 The Breast Cancer Information Core database, which contains germline BRCA sequences from breast cancer patients and from women with elevated risk (http://research.nhgri.nih.gov/bic/) was searched for these sequence alterations. All the exonic sequence alterations identified here were not found to be clinically important. Additionally, all the changes observed at the intron-exon borders are not predicted to affect the splice site sequences (http://www.uni-duesseldorf.de/rna/html/5__ss_mutation_assessment.php).

Table III. SNPs in the BRCA1 and BRCA2 Genes1
Cell cultureGeneExonSNP IDAmino acid change
  • 1

    BRCA1 and BRCA2 genomic DNA sequences were analysed within all exonic and neighbouring intronic regions.

Bca10BRCA18−34c/t 
  9+67c/t 
  112201C→T Ser694Ser
  112430T→C Leu771Leu
  112731C→T Pro871Leu
  113232A→G Glu1038Gly
  113667A→G Lys1183Arg
  134427T→C Ser1436Ser
  18−12c→t 
 BRCA22203G→A non-coding
  113624A→G Lys1132Lys
  147470A→G Ser2414Ser
Bca11BRCA1112731C→T Pro871Leu
  18−12c→t 
 BRCA24+67a/c Asn289His
  101093A→C His321His
  101342C→A Ser455Ser
  101593A→G Ans991Asp
  113199A→G 
  17−14t→c 

Next, we performed mutational analysis of TP53 to understand whether DSB repair changes observed in Bca10 and Bca11 cells are related to mutations in the third key breast cancer susceptibility gene. Sequencing revealed that these two cell cultures harbor wild-type TP53 for exons 5 to 9. Exons 5 to 9 bear the vast majority of cancer-related mutations and encode the central domain of p53, whose integrity is required for DSB repair regulatory functions.12–14, 36, 37 Further indicating the existence of wild-type TP53 in these cells, p53 protein expression was undetectable by immunocytochemistry (Table II).

Since reduced or absent expression compromises BRCA1 functions in approximately one third of mammary and ovarian carcinomas,38, 39 we additionally performed immunocytochemical protein staining. Interestingly, in Bca10 cells nuclear BRCA1 protein expression was not detectable at all (Table II). In Bca11, Bca14, and Bca16 cells, BRCA1 was observed in the cytoplasm rather than the nucleus. These observations may indicate disturbed BRCA1 expression and possibly localization, respectively, in these breast cancer cells.

Discussion

This is to our knowledge the first report that describes the analysis of distinct DSB repair activities in primary breast cancer cells. So far, tests to evaluate DNA repair activities in man relied on less specific endpoints such as DNA breakage or micronucleus formation and were applied on peripheral blood lymphocytes.5 However, breast carcinogenesis in BRCA patients is associated with loss of heterozygosity of BRCA1 and BRCA2, respectively, thereby abolishing the respective DSB repair activities in tumor cells specifically.40, 41 Overloading or loss of upstream DNA repair pathways results in lethality of these BRCA1 or BRCA2 depleted malignant but not the normal cells, which formed the basis for the development of novel targeted therapies. Thus, BRCA tumors show increased sensitivities to certain DNA damaging chemotherapeutic agents and inhibitors of the base excision repair protein PARP1.21, 26, 42 Intriguingly, the growing list of breast cancer susceptibility genes encompasses 10 genes whose products directly or indirectly influence DSB repair.1 Emerging evidence further indicates that somatically acquired gene mutations affecting DSB repair factors as well as epigenetic and protein expression/localization changes are associated with the formation of sporadic mammary carcinoma.5 Therefore, the impact of DSB repair dysfunction goes way beyond germline BRCA1 or BRCA2 mutations and may even represent the common denominator of mammary tumor formation at least with respect to certain subgroups resembling BRCA-like tumors. These findings emphasize the need for new methods enabling functional analyses of DSB repair activities in the primary tumor to accurately predict tumor-specific responsiveness to conventional as well as novel therapeutic treatments.

Epithelial cell cultures established from primary mammary carcinomas provide an important experimental system to characterize the phenotype underlying tumor initiation and progression.30, 43 However, isolation and maintenance of pure epithelial cell cultures from human primary breast tumors is difficult. In this work, through a multistep procedure, we successfully established primary epithelial cell cultures from 5 breast cancer patients. The epithelial nature of the cells was manifested from morphological characteristics and EMA marker positivity. Immunocytochemical analysis of the cell cultures, showed additional expression of the mesenchymal cell marker, vimentin, for cultures Bca9, Bca10, and Bca16 which would suggest transition of epithelial to mesenchymal cell type, implicated in the conversion of early stage tumors into high grade-malignancies.44 Cells from all the different cultures established in this study showed characteristic features of transformed cells such as loss of contact inhibition or anchorage-independent growth pattern. The cells were ER- and PR-negative and grew in multilayers, further confirming the malignant phenotype. Though our study documents the feasibility of DSB repair analysis in epithelial breast cancer cells, culture establishment still represents the bottleneck and is particularly labor intensive. This made it impractical to examine samples in a large-scale format, which is required to determine whether DSB repair analysis with this method will be an independent risk marker or treatment predictor. Therefore, further improvements of the cultivation methodology are currently under investigation in our laboratory to allow systematic comparative analysis of breast tumor samples in the future.

To determine DSB repair frequencies, we introduced DSB repair substrates and I-Sce I expression plasmid through nucleofection. Using this method, we achieved reproducibly high transfection efficiencies with all the primary cell cultures and with control breast epithelial cell lines analyzed. This allowed making use of the EGFP-based repair assay with a fast readout and multiple reporter plasmids, enabling pathway-specific measurements in primary cells. Comparison of NHEJ, nonconservative (SSA), and conservative (HR) HDR activities in primary breast cancer versus MDA-MB-157 cells revealed statistically significant differences in 2 of the 5 cultures analyzed. Differences detected in these cultures rather under- than overestimate alterations in repair activities, because we determined equivalent (NHEJ) or even lower (SSA, HR) DSB repair frequencies with immortalized, nontransformed MCF12A and MCF10A cells as compared with the breast cancer cell line MDA-MB-157 (data not shown).

In cells from the familial breast carcinoma Bca11, we found NHEJ activities, which were elevated by two orders of magnitude. Nonconservative HDR, not HR, was also significantly enhanced, although with one order of magnitude less dramatically than NHEJ. One culture among the sporadic breast cancer cell cultures (Bca10) also showed an increase in HDR, though more modestly than Bca11. Characterization of the DSB repair pattern of human LCLs from BRCA1 or FANCD1/BRCA2 mutation carriers by use of the same EGFP-based DNA substrates demonstrated an up to 9.5-fold increase in NHEJ together with a 2.9- to 3.5-fold HDR stimulation. Additionally, HR was 11-fold reduced in FANCD1/BRCA2 LCLs. These data were fairly in agreement with published results obtained with reporter assays, PCR, or restriction enzyme analysis, suggesting that the use of extrachromosomal, EGFP-based DNA substrates enables to detect functional changes in DSB repair caused by major breast cancer susceptibility genes. In these previous studies NHEJ, particularly error-prone NHEJ, was stimulated in BRCA1 mutant cell lines, SSA was increased in BRCA2 mutant cells, and HR was reduced in both mutant cell types.7, 11, 45, 46 In terms of NHEJ and SSA performance, Bca11 and Bca10 cells, thus, resemble BRCA1 and/or BRCA2 mutant cells. Arguing against this interpretation, we saw elevated rather than reduced HR frequencies with Bca11 or Bca10 cells. However, in the tumors investigated here, other DSB repair regulatory pathways might be affected in combination with potential dysfunction of the BRCA1 or BRCA2 pathways. Indeed, acquired somatic mutation in TP53 has been found more frequently in BRCA than sporadic tumors.47 In cells devoid of functional p53 versus wild-type p53 counterparts, we and others previously scored elevated HR frequencies.12–14 From this, it is conceivable that inactivation of the molecular pathway involving p53 neutralizes HR frequency reduction in cells carrying BRCA1 and BRCA2 mutations. In conclusion, we propose that NHEJ and/or SSA rather than HR changes may serve as sensitive and robust biomarker for DSB repair dysfunction in tumor samples.

To understand whether Bca11 or Bca10 cells carry mutations in the candidate breast cancer susceptibility genes emerging from our functional analysis, we performed sequence analyses. Mutations were neither revealed in the BRCA1, BRCA2, nor the TP53 gene. Even though no mutations were found in these cells, polymorphisms in BRCA1 and BRCA2 genes were observed. Resulting amino acid exchanges are unlikely to interfere with critical interactions involving DSB repair proteins. The biological significance of these polymorphic sites remains to be determined, to understand their implication, if any, on BRCA1 and BRCA2 gene function, and consequences for breast carcinogenesis. However, even in the absence of genetically altered copies of these breast cancer susceptibility genes, epigenetic mechanisms as well as altered protein interactions were shown to be responsible for BRCA1 and BRCA2 inactivation in tumors.5 In particular, 30–40% of sporadic mammary and ovarian carcinomas show reduced or absent BRCA1 expression.38, 39 Aberrant cytoplasmic overexpression of the critical binding partner of BRCA1, BARD1, may represent a potential mechanism for cytoplasmic sequestration of BRCA1 in breast and ovarian cancer.48 Here, we detected cytoplasmic localization of BRCA1 protein in Bca11 cells by immunocytochemistry. The nonavailability of the protein in the nucleus of these cells could implicate defects in DSB repair pathways. Arguing against this interpretation, cytoplasmic localization was also observed with cultures Bca14 and Bca16 without abnormal DSB repair activities. In Bca10 cells, in which BRCA1 was not detectable at all, the altered protein status more likely contributes to the DSB repair pattern observed. Alternatively, other genes within the extensive BRCA1, BRCA2, and p53 signaling pathways may have undergone genetic or epigenetic alterations, which cause BRCA- and/or mutant p53-like DSB repair changes in Bca10 and Bca11.

Taken together, our findings indicated that in two primary breast cancer cell cultures the error-prone pathways of microhomology-mediated NHEJ (Bca11) and/or SSA (Bca11, Bca10), which both cause deletions between sequence repeats,6 had become deregulated. From the dramatic effect in Bca11 cells, it is conceivable that these cells carry a germ-line mutation in an unrecognized DSB repair gene in the BRCA1 pathway, and that loss of the wild-type allele, mutations or epigenetic changes in complementary genes occurred early in tumor development. Irrespective of the molecular details of the surveillance pathway altered, increases in error-prone pathways compensate for loss of error-free repair and indicate defects in DSB repair leading to breast cancer. Therefore, the method described here could be useful in the identification of functional changes in tumor specimens that would predict responsiveness to therapeutics which target DNA repair in both familial and specific subtypes of sporadic breast cancer.

Acknowledgements

We are grateful to Dr. Silke Süsse and Ms. Sandra Herman as well as Mrs. Antje Merkle, Ulm, for expert help during recombination measurements and BRCA sequence analysis, respectively.

Ancillary