Background
Breast cancer is the third most common form of cancer, with almost 1.5 million women in the world diagnosed with the disease in 2010 alone [
1,
2]. The extensive use of mammography has resulted in a large proportion of breast cancer cases being detected at an earlier stage, resulting in increased survival and outcome [
3]. However, approximately 3-6% of patients continue to present with metastatic disease at diagnosis throughout the US and Europe [
4,
5]. As a significant number of cases present with metastatic disease when the primary tumor is not locally advanced [
6], one can hypothesize that there is heterogeneity in tumor biology between patients, versus a failure of screening. Despite the presence of targeted therapeutics for hormone receptor sensitive and HER2 over-expressing breast cancers, treatment of metastatic disease remains incurable. Therefore, identification of women with a predisposition to develop tumors that are more likely to metastasize is critical to help develop improved prevention and screening strategies for those individuals.
The Breast Cancer 1, early onset gene (
BRCA1) located on chromosome 17q21.31 [
7,
8] encodes a tumor suppressor that plays a critical role in the DNA damage response and repair pathways [
9,
10]. Germline variants in the open-reading-frame of
BRCA1 confer a mean risk of 54% and 39% for developing hereditary breast and ovarian cancer (respectively) by age 70 [
11‐
14]. However,
BRCA1 open-reading-frame variants only account for a small portion of hereditary breast cancer cases that occur primarily in young, premenopausal patients [
15]. Therefore, the search for additional germline variants, outside of the
BRCA1 open-reading-frame predicting increased breast cancer risk has been undertaken. Such variants in the
BRCA1 3’UTR have recently been identified and were first implicated in breast and ovarian cancer susceptibility in high-risk families [
16]. Two variants
5711 + 421 T/T and
5711 + 1286 T/T (located in the
BRCA1 3’UTR) are associated with cancer risk in Thai women from breast and ovarian cancer families (OR = 3.0). Independent evaluation of the
5711 + 421 T/T variant (referred to here as rs8176318 or the
BRCA1-3’UTR-variant) revealed significant variation in baseline frequency by ethnicity, with a documented minor allele frequency in Irish populations of approximately 0.28 [
17]. Homozygous G > T variants were found to be associated with increased risk of breast cancer in African American women (OR = 9.48, 95% CI 1.01-88.80), and were specifically associated with the development of triple negative breast cancer (OR = 12.19, 95% CI 1.29-115.21) [
17]. This data suggests that the
BRCA1-3’UTR-variant not only confers an increased risk of developing breast cancer, but may also be associated with tumor biology, since the propensity to develop triple negative breast cancer is higher than that of the other subtypes. One could hypothesize from these findings that the
BRCA1-3’UTR-variant functions similarly to that of canonical
BRCA1 open-reading-frame variants, which are more commonly associated with development of triple negative breast cancer as opposed to the other subtypes [
18‐
20].
Open reading frame variants in
BRCA1 have not clearly been associated with unique tumor biology, but only have been predictive of response to therapeutic agents that take advantage of their inherent DNA repair defects [
21]. In contrast, 3’UTR variants in cancer- associated genes have been shown to predict both altered response to specific therapies, as well as inherent differences in tumor biology. This is likely due to the fact that these variants are in regions of regulatory elements that control the nature and timing of gene expression, and their effects are only manifest under particular extracellular and/or intracellular stimuli (for review see ([
22]). One mechanism for regulation of these variants is by trans-acting factors such as miRNAs, which are rapidly altered by external factors such as genotoxic stress [
23] and estrogen receptor signaling [
24].
Based on evidence of the biological function of other 3’UTR variants in cancer, and the association of the
BRCA1-3’UTR-variant with breast cancer risk in two previous studies [
16,
17], we sought to investigate the impact of this variant on
BRCA1 expression and its association with tumor biology as seen in clinical presentation in a clinically well-annotated breast cancer population.
Methods
Luciferase reporter assay
Luciferase reporters containing either the rs8176318 G-allele or T-allele were generated by PCR amplification of the BRCA1 3’UTR loci from HMEC genomic DNA (heterozygous for the BRCA1-3’UTR-variant) using the following DNA oligonucleotides:
BRCA1 forward 5’ ATGACTCGAGCTGCAGCCAGCCACAGGTACAGAGCCACAG 3’
BRCA1 reverse 5’ ATGAGCGGCCGCGTGTTTGCTACCAAGTTTATTTGCAGTG 3’
PCR amplicons were subcloned into the XhoI and NotI sites (underlined) of the psiCHECK2 dual luciferase vector (Progema). Constructs were sequence verified to confirm that the only difference in the BRCA1 3’UTR was the rs8176318 variant.
MCF-7, MDA-MB-231, MDA-MB-361, MDA-MB-468, Hs 578 T and BT-20 cells were purchased from the ATCC and grown at 37°C and 5% CO2 according to the manufacturer’s protocol. MCF-7 and BT-20 cells were cultured using MEM (GIBCO) supplemented with 10% fetal bovine serum (GIBCO) and 100 ug/ml penicillin, 100 U streptomycin. MDA-MB-231, MDA-MB-361 and MDA-MB-468 cells were cultured using Leibovitz’s L-15 (GIBCO) supplemented with 10% fetal bovine serum and 100 ug/ml penicillin, 100 U streptomycin. Hs5788T cells were cultured in DMEM (GIBCO) supplemented with 10% fetal bovine serum and 100 ug/ml penicillin, 100 U streptomycin.
Cells in log-growth phase were transfected with either the G-allele or T-allele expressing luciferase reporters (100 ng) using Lipfectamine 2000 (Invitrogen) according to the manufacturer’s protocol. Following a 16-hour incubation the cells were lysed and analyzed for dual luciferase activities by quantitative titration using the dual luciferase assay kit (Promega). Renilla luciferase was normalized to firefly luciferase. Graphed is the mean ± standard deviation (SD) of 3 independent experiments. Statistical significance was determined by student’s t-test (1-tailed, paired t-test). A p-value of less than 0.05 was considered statistically significant.
Immunofluorescence staining of BRCA1in tumor tissue
BRCA1 protein expression was analyzed from tumor tissue derived from the triple negative breast cancer cohort subset with corresponding
BRCA1-3’UTR-variant genotype information, using an immunofluorescent platform, AQUA™, on tissue microarrays (TMAs) of tumor cores.
BRCA1 protein was assessed using monoclonal MS110 Ab-1 anti-
BRCA1 (Calbiochem) [
25‐
27] and rabbit polyclonal anticytokeratin (DAKO), at dilutions of 1:100 and 1:200 respectively in 0.3% BSA/TBS buffer for 1 h at 37°C. AQUA has been described previously [
28,
29].
Estrogen withdrawal assay
MCF-7 cells cultured in phenol-red free MEM (GIBCO) containing 5% fetal bovine serum and 100 ug/ml penicillin, 100 U streptomycin, were treated with either 100 nM Fuvestrant (Sigma I4409) or β-Estradiol (Sigma E8875). Following a 48-hour incubation, the cells were transfected with luciferase reporters (100 ng) harboring either the BRCA1 G-allele or T-allele 3’UTR using Lipofectamine 2000. After a 16-hour incubation the cells were lysed and analyzed for dual luciferase activities by quantitative titration. Renilla luciferase was normalized to firefly luciferase. Graphed is the mean ± SD of 3 independent experiments, preformed in triplicate. Statistical significance was determined by student’s t-test (1-tailed, paired t-test). A p-value of less than 0.05 was considered statistically significant.
Total RNA was isolated from cell lysates by Trizol extraction as previously described [
30]. cDNA was generated using iScript cDNA Synthesis Kit (Bio-Rad). Target mRNA was amplified by qPCR using iTaq SYBR Green Supermix with ROX (Bio-Rad) on a 7900HT Fast Real-Time PCR System (Applied Biosystems) using the following DNA oligonucleotide primers:
Actin forward 5’ AGAAAATCTGGCACCACACC 3’
Actin reverse 5’ AGAGGCGTACAGGGATAGCA 3’
GREB1 forward 5’ GTGGTAGCCGAGTGGACAAT 3’
GREB1 reverse 5’ TGTGCATTACGGACCAGGTA 3’
TFF1 forward 5’ CACCATGGAGAACAAGGTGA 3’
TFF1 reverse 5’ CCGAGCTCTGGGACTAATCA 3’
mRNA levels were calculated by the delta-delta C
T method [
31]. Samples were run in triplicate and standard deviation (SD) is the average of 3 independent experiments.
Study populations
All women with a biopsy confirming breast cancer at Galway Hospital and its affiliates are approached to enroll in the breast cancer study including DNA collection. Informed consent, a detailed family history of breast and/or ovarian cancer and a peripheral venous blood sample are obtained from cases and controls. Controls were women from the west of Ireland, primarily over 60 years of age, without a personal history of cancer of any type and without a first-degree family member with breast or ovarian cancer. These controls were accrued primarily from Active Retirement association meetings and from Nursing home residents. All cases and controls were recruited following appropriate ethical approval from the Galway University Ethics Committee. 728 cases and 387 controls were included from this cohort.
The Irish patient cohort consisted of 728 women with invasive, primary operable breast cancer diagnosed between June 1980 and August 2007, with complete receptor status (outlined in Additional file
1). Receptor status was determined using established histopathological methods and immunohistochemistry, followed by fluorescence in-situ hybridisation (FISH) to confirm HER2/
neu positivity in samples that scored a 2+ on Hercept test. The samples were then grouped into Luminal A, Luminal B, HER2 and triple negative subtypes based on receptor status but in the absence of gene expression analysis. Patient demographics and tumor characteristics were recorded and outcome/survival data was prospectively maintained using hospital medical records. Disease free survival (DFS) was defined as time in months, from breast cancer diagnosis to point of loco/regional recurrence or distant disease progression, progression free survival (PFS) was defined as time in months from the point of diagnosis of Stage IV cancer to disease progression and overall survival (OS) was defined as the time from breast cancer diagnosis to the end of follow-up or death (months).
The CT Triple Negative Breast Cancer (TNBC) Cohort has been previously described [
32], but briefly, FFPE tissue was obtained from 134 TNBC patients, who underwent surgery at the Yale University New Haven Hospital or the Hospital of Bridgeport, Connecticut, between 1985 and 2007. Patient sample collection was performed through a Yale HIC approved tissue collection protocol. Tissue of 120 patients was used for TMA construction and the follow up time for these patients ranged between 3 months and 19 years with a mean follow up of 4.4 years. Patient age at diagnosis ranged from 30 to 90 years, with a mean age at diagnosis of 53 years. Sixty-two patients were diagnosed as node negative and 40 patients as node positive. There were 65 patients who were Caucasian in this cohort who were used for this analysis. Treatment was known in 86% of patients, out of those 63% received chemotherapy. Gene expression in TMAs was analyzed by AQUA technology [
28,
29], and results were reviewed and confirmed by two independent pathologists.
The High Risk Breast Program from Vermont is a database that is IRB approved and was established at the University of Vermont in 2003. Eligible women include those with a strong family history of breast cancer (55.2% of participants), a prior breast biopsy showing atypical ductal hyperplasia or lobular neoplasia (14.7%), a known germline abnormality of BRCA1 or 2 (7.3%, but excluded from this study), or a prior history of receiving chemo-radiotherapy for Hodgkin’s disease (1.3%). At study entry, unaffected high-risk women provide anthropometric measurements, medical/family history, physical activity and diet information, mammography reports, health behavior information and provide a blood sample for storage that may be used for future research. 536 women have been enrolled into this database since 2003 with follow-up visits, questionnaire completion and blood draws occurring at 4 and 8 years after study entry. Status of enrolled women (i.e., new cancer diagnosis) is updated on an ongoing annual basis. For this study, 367 women were genotyped for the BRCA1-variant.
BRCA1-3’UTR-variant genotyping
1–3 mL of whole blood was drawn from the Irish cases and controls and DNA was isolated. DNA was isolated from FFPE tissue for genotyping for the TNBC Cohort. DNA was supplied from the Vermont cohort. From blood, DNA was isolated using a DNA extraction kit (Gentra Puregene) or Ambion according to the manufacturer’s protocol. Genotyping was performed using a custom TaqMan genotyping assay (Applied Biosysytems) that was specific for rs8176318. Each reaction was performed in a 20 μl volume using 10 μl of 2× TaqMan Genotyping MaterMix, 1 μl of the 20× variant assay, approximately 40 ng of DNA and nuclease free water in a 96-well plate. The reactions were run on the Applied Biosystems 7900HT Fast Real-Time PCR System in a two-stage process incorporating PCR amplification and allelic discrimination. Genotypes were analyzed using the Applied Biosystems SDS 2.3 genotyping software and automatic calls were verified by observing the spectral contributions of the dye corresponding to the sequence specific probe on the Multicomponent Data Plot. Internal quality control was maintained using established positive and negative controls to ensure genotyping accuracy and 6% percent of DNA samples were genotyped in duplicate with 100% consistency of results. Two DNA samples of the 728 cases failed to amplify and were excluded from further analyses. All Caucasian cases from the TNBC cohort amplified and were included in the analysis. All BRCA coding sequence non-mutant patients from the Vermont cohort were included.
Statistical analysis
The genetic distribution of the breast cases and controls were tested for Hardy-Weinberg equilibrium and were found to be in equilibrium. In order to evaluate the distribution of patient demographics in cases and controls as well as tumor features among the cases, categorical variables were analyzed using the χ2 test and continuous variables were analyzed using t-tests. Binary logistic regression was used to evaluate the association of each genotype with cancer. Case–control analysis comparing genotypes in different models was performed using a χ2 test to obtain odds ratios (OR), 95% Confidence Interval (CI) and p-values. Based on the preceding statistical findings, the dominant model was used for all further analyses.
Prevalence of the variant across cancer subtypes, and comparison of the respective subtypes against controls were evaluated using χ2 analyses. The Luminal A cases were stratified according to menopausal status and the observed genotype distribution compared with controls using χ2 test. Association of the variant with ER/PR status controlling for other patient and tumor variables was analyzed using binary logistic regression.
Binary logistic regression was used to evaluate the independent effect of metastasis and disease stage in predicting variant positivity in all cancer cases and Luminal A cases specifically. Logistic regression analyses for all cases and Luminal A cases with a binary outcome variable coded as rs8176318 positive (TT or GT genotypes) or negative (GG genotype) included variables such as age at diagnosis, menopausal status, tumor grade, ER/PR status and stage.
Cox Proportional Hazards models were fitted to evaluate the effect of the variant on disease free survival, progression free survival and overall survival in all cancer cases and according cancer stage.
Fisher’s Exact Test was used to examine the statistical significance of the association between mammographic density and the presence or absence of the BRCA1-3’UTR-variant in both the entire population, as well as in a variety of subsets (BMI categories, pre- or post-menopausal women, and age at menarche categories).
Discussion
Here we show for the first time that the rs8176318 G > T 3’UTR variant (the BRCA1-3’UTR-variant) is associated with decreased BRCA1 expression both in vitro and in vivo, and is impacted by cellular exposure to estrogen. More importantly, we show that this variant predicts aggressive breast cancer biology and stage IV disease, as well as modest increased breast cancer risk in a homogeneous well-characterized west-Irish population. In addition, studying a collection of women at high risk for breast cancer, we found that this variant is associated with features usually considered to improve the ability of mammograms to detect disease (lower mammographic density). These findings suggest that presentation with stage IV disease of BRCA1-3’UTR-variant patients is unlikely to be due to the inability to detect disease early, but instead suggests that this variant predicts biologically aggressive disease. These are hypothesis deserving further investigation.
While the findings of increased cancer risk are in agreement with prior reports [
16,
17], this is the first study evaluating biologic function and clinical associations of the
BRCA1-3’UTR-variant with the patients that are carriers and develop cancer. While the search for germ-line variants in
BRCA1 have predominantly focused on open-reading-frame variants, increasing evidence is showing that alterations in non-coding regions of genes (such as this variant) explain a proportion of cancer susceptibility, and more importantly play a role in tumor biology and can act as prognostic biomarkers. While the exact biological mechanism leading to altered
BRCA1 expression in
BRCA1-3’UTR-variant associated tumors is unknown, it is predicted to be a miRNA binding site of miR-20a-3p and miR-5001-3p by target prediction programs including MirSNP and PolymiRTS, and was shown previously to be impacted by miRNA targeting [
16]. We hypothesize that this may be more complex, with this region potentially being a landing dock for other RNA binding proteins, and is work that is ongoing but outside of the scope of this proposal.
Diminished expression of
BRCA1 has previously been shown to increase the growth rate of benign and malignant breast tissue [
38,
39]. In another study, loss of nuclear
BRCA1 expression (using IHC) was significantly associated with high histological grade (p < 0.025) (p < 0.05) [
40]. Both of these findings could help explain the association of the
BRCA1-3’UTR-variant with tumor progression and aggressive phenotype. Interestingly, low
BRCA1 mRNA expression identified in sporadic breast cancer specimens has been associated with development of distant metastasis (p = 0.019) and a shorter disease free interval (p = 0.015) [
41]. Additionally, Japanese women whose tumors stained negative for
BRCA1 expression had worse disease free survival than similar patients whose tumors were positive for
BRCA1 staining [
42]. Overall, these findings are in agreement with our findings regarding the
BRCA1-3’UTR-variant, that reduced
BRCA1 expression in the absence of germ-line protein coding sequence variants may be associated with aggressive tumor biology.
Although the
BRCA1-3’UTR-variant has now been shown to predict a significant increased risk of breast cancer risk in three independent well-characterized cohorts, it is notable that this variant has not been reported from GWAS analyses. We hypothesize that this may be partly due to the association of the
BRCA1-3’UTR-variant with advanced disease presentation, as patients with Stage IV cancer are generally underrepresented in cohorts that are not comprehensive sequential patient collections, such as the one used in this study, as well as in the Pelletier triple negative cohort study [
17]. Another possibility is that because this variant, similar to other identified 3’UTR variants, is altered by “context”, in this case estrogen, which will be altered by body habitus as well as the societal acceptance of hormone replacement therapy, it would make it more likely to be missed in mixed populations such as those used in GWAS studies. For this new class of mutation, 3’UTR variants, the homogeneity and appropriate characterization of the study sample is likely to be much more important than simple sample size.
Our findings suggest a hypothesis where in women with the BRCA1-3’UTR-variant, if progressing to an estrogen independent phenotype, their BRCA1 becomes even less functional, possibly allowing more DNA damage, and perhaps selection for a more aggressive breast cancer genotype. These findings could also indicate that the BRCA1-3’UTR-variant becomes the greatest risk for cancer development at the time of estrogen withdrawal, or menopause. While the steps required to lead to breast tumorigenesis in these patients will require studies with in vitro and in vivo models, this work represents a significant step forward in generating hypotheses about this variant, as well as understanding the role of this variant, and other such variants, in cancer biology.
Competing interest
JBW is the co-founder of a company that has licensed IP regarding the rs8176318 polymorphism from Yale University.
Authors’ contributions
JD carried out the genotyping and participated in writing the manuscript. DWS participated in the study design, carried out the luciferase reporter assays and participated in writing the manuscript. CP carried out luciferase reporter assays. RL preformed the AQUA analysis. CS, CC, KLG, TM, LH participated in patient sample and database curation. DW and JN carried out the statistical analysis. MK, NM participated in developing the study design. TR and MW analyzed the Vermont samples and weighed in on the interpretation. JBW participated in developing the study design, coordination of collaborations and patient sample acquisition and helped write the manuscript. All authors read and approved the final manuscript.