Genetic polymorphisms have been recognized to play an important role for health and disease. A recent example of a single nucleotide polymorphism (SNP) relevant to breast cancer risk is rs2981582 which is located at the fibroblast growth factor receptor 2 (FGFR2) and of which the homozygous minor AA genotype is significantly associated with an increased risk to develop breast cancer.1 With respect to breast tumors, this polymorphism is associated with ER positive but not ER negative status, thus linking it with breast cancer prognosis and prediction.2 Whether genetic variations of regulatory genes may affect other breast cancer prognostic and treatment relevant tumor characteristics such as human epidermal growth factor receptor 2 (HER2) overexpression is as of yet unknown.
HER2 overexpression of breast tumors is associated with rapid tumor growth, increased risk of recurrence after surgery, poor response to conventional chemotherapy and shortened survival.3 However, patients with HER2 positive breast tumors benefit from treatment with trastuzumab and show a 50% reduction in the risk of recurrence.4–6 HER2 is overexpressed in 10–34% of breast tumors3 and therefore, it is of interest to understand the regulatory mechanisms of HER2 expression. The molecular basis of HER2 overexpression in breast cancer is gene amplification and transcriptional up-regulation.7, 8 Transcription factor families AP-2 and Ets have been identified to contribute to HER2 overexpression in breast cancer cells9–11 and it has been recognized that cell cycle deregulation via the impaired control of regulatory elements plays an important role for the increased proliferative capacity of tumor cells.12 In addition to their roles as oncogenes or tumor suppressors, cell cycle regulators impact tumor characteristics with respect to expression of proteins.13–18 There is a possibility that cell cycle regulators may be important prognostic and predictive markers for breast cancer and represent potential drug targets, a reason why they deserve special attention. Because cyclin D1 (CCND1), cyclin D3 (CCND3) and the E2F transcription factor 2 (E2F2) cooperate in cell cycle regulation, they represent a set of regulators that may impact breast cancer characteristics either alone or in combination. CCND1 and CCND3 control G1/S phase transition and E2F2 is mainly involved in the cell-cycle progression from G1 to S phase via transcriptional regulation of relevant genes, DNA synthesis, checkpoint control, apoptosis, DNA repair and development.19 Importantly, E2F2 is a potent activator of the CCND3 promoter.20 Based on known common polymorphisms E2F2_-5368_A>G (rs760607), CCND1_870_A>G (rs603965) and CCND3_-677_C>T (rs3218086) with frequencies of minor alleles as being 40%, 48% and 18%, respectively, these cell cycle regulators are attractive candidates for the investigation of an influence on HER2 expression in breast cancer.
We analyzed the German population-based breast cancer case-control collection GENICA21–24 and show that all 3 polymorphisms were significantly associated with HER2 expression in breast tumors.
Material and methods
The GENICA study participants of the population-based breast cancer case-control study from the Greater Bonn Region, Germany, were recruited between August 2000 and September 2004 as described previously.21, 22, 24 In brief, there are 1,143 incident breast cancer cases and 1,155 population controls matched in 5-year classes. Cases and controls were eligible if they were of Caucasian ethnicity, current residents of the study region and below 80 years of age. Information on known and potential risk factors was collected for all participants via in-person interviews. Response rates were 88% for cases and 67% for controls. Potential breast cancer risk factors included age at diagnosis (<50, ≥50 years), menopausal status (premenopausal, postmenopausal), family history of breast cancer (yes, no), use of oral contraceptives (never, >0–<5, 5–<10, ≥10 years), use of hormone therapy (never, >0–<10, ≥10 years), body mass index (<20, 20–<25, 25–<30, ≥30 kg/m2) and smoking status (never, former, current) (Table I). Information on clinical and histopathological tumor characteristics was available for 1,011 (89%) cases. Data included HER2 status (positive, negative), ER status (positive, negative), PR status (positive, negative), grading (G1, G2, G3), tumor size (cis, T1, T2, T3, T4), histology (ductal, lobular, ductolobular) and nodal status (N0, N ≥ 1) (Table I). The GENICA study was approved by the Ethićs Committee of the University of Bonn. All study participants gave written informed consent.
Table I. Baseline Information on Breast Cancer Cases and Population Controls of the Genica Study and E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T Genotype Frequencies
Genomic DNA was extracted from heparinized blood samples (Puregene™, Gentra Systems, Mineapolis) as previously described.21 DNA samples were available for 1,021 (89%) cases and 1,015 (88%) controls. All 2,036 DNA samples were genotyped at E2F2_-5368_A>G (rs760607), CCND1_870_A>G (rs603965) and CCND3_-677_C>T (rs3218086). Polymorphisms were selected on the basis of a minimum minor allele frequency of 15% in Caucasians and a known or predictable functional effect. Genotyping was performed by matrix assisted laser desorption/ionization time-of-flight mass spectrometry (MALDI-TOF MS) as described previously.21, 25 A Bruker Autoflex and a Sequenom Compact MALDI-TOF MS were used for data acquisitions from the SpectroCHIP. Genotyping calls were made with MassARRAY RT software v 18.104.22.168 (Sequenom, San Diego, CA). For quality control repeated analyses were performed for 10% of randomly selected samples. Primers were synthesized by Metabion International AG, Martinsried, Germany.
Breast tumor specimens were fixed in 4% buffered formaldehyde and embedded in paraffin. Immunohistochemical stainings were performed on 4 μm sections. HER2 status of breast tumor samples was determined by protein expression as described previously.26 HER2 status which was not mandatory at the time of recruitment is available for 67% of breast cancers. Staining intensity of the cellular membrane was evaluated following the guidelines for scoring HercepTest, DAKO (Glostrop, Denmark). Expression of hormonal receptors ER and PR was determined by multiplying the number of positive cells by the intensity of nuclear staining, evaluation followed the German Immuno Reactive Score according to St. Gallen consensus 2003.27
In silico search for putative binding sites of transcription factors
Polymorphisms rs760607 and rs3218086 are located in the promoter region of E2F2 and CCND3, respectively. Because of the possibility of the location of these polymorphisms in binding sites of transcription factors modifying gene expression, we used the TFSEARCH software (www.cbrc.jp/research/db/TFSEARCH.html) to identify such putative transcription factor binding sites in silico.
E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T genotype frequencies were tested for Hardy-Weinberg equilibrium. Associations between genotypes and breast cancer risks were analyzed by logistic regression conditional on age-adjusted for 6 epidemiological breast cancer risk factors (menopausal status, family history of breast cancer, use of oral contraceptives, use of hormone therapy, body mass index and smoking). Subgroup analyses were performed for these 6 variables.
In breast cancer cases, associations between E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T genotypes and 7 clinical and histopathological tumor characteristics (HER2-, ER- and PR-status, grading, tumor size, histology as well as nodal status) were analyzed. All statistical tests were two-sided. To correct for multiple testing we divided the significance level of 0.05 by the number of tested variables. In case of epidemiological variables, 0.05 was divided by 6 with p values ≤0.008 being considered significant. In the case of tumor characteristics, the significance level of 0.05 was divided by 7 with p values ≤0.007 being considered significant. Risk estimates were given as odds ratios (OR) and 95% confidence interval (CI).
Since E2F2_-5368_A>G, CCND1_870_A>G and CCND3_ -677_C>T polymorphisms are located in genes encoding factors cooperating in the same functional network, we performed pairwise interaction analysis to test whether these 3 polymorphisms might have interactive effects.
All statistical analyses were done using SAS v 9.1.3 (SAS Institute, Cary, NC).
We analyzed E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T polymorphisms within the GENICA study population. Genotype call rates were >98%, concordance among repeated samples was 100% and all genotype frequencies in cases and controls met Hardy-Weinberg equilibrium. None of the polymorphisms showed differences with respect to genotype frequencies between breast cancer cases and population controls (Table I). Moreover, in subgroup analyses we observed no differences between breast cancer cases and controls with respect to menopausal status, family history of breast cancer, use of oral contraceptives, use of hormone therapy, body mass index and smoking (data not shown).
After we had excluded that any of the 3 polymorphisms was associated with breast cancer risk, we analyzed the breast cancer case group for potential associations between genotypes of E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T and clinical as well as histopathological tumor characteristics and observed strong associations with the HER2 status.
For E2F2_-5368_A>G, we observed significant associations with HER2 status for the heterozygous AG genotype an OR of 0.59 (95% CI: 0.40–0.87, p = 0.007, Table II) and for the combined AG/GG genotypes an OR of 0.60 (95% CI: 0.42–0.85, p = 0.004, Table II). For CCND1_870_A>G, we observed significant associations with HER2 status for the heterozygous AG genotype an OR of 0.64 (95% CI: 0.43–0.96, p = 0.030, Table II) and for the combined AG/GG genotypes an OR of 0.66 (95% CI: 0.45–0.96, p = 0.029, Table II). For both polymorphisms E2F2_ -5368_A>G and CCND1_870_A>G we observed no significant risk association for the rare homozygous genotype. For CCND3_ -677_C>T, we observed an increased risk for HER2 positive breast tumors for the heterozygous CT genotype with an OR of 1.79 (95% CI: 1.24–2.60, p = 0.002, Table II) and for the combined CT/TT genotypes an OR of 1.72 (95% CI: 1.20–2.49, p = 0.004, Table II). For the rare homozygous genotype TT, we observed no significant risk association.
Table II. Association of E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T Polymorphisms and HER2 Status of Breast Tumors
Following correction for multiple testing the effect of CCND1_870_A>G on HER2 status vanished, since only p values ≤0.007 were considered significant. Effects of E2F2_-5368_A>G and CCND3_-677_C>T on HER2 status remained significant.
No associations between ER and PR status, grading, tumor size, histology or nodal status and the analyzed polymorphisms were observed (data not shown).
To evaluate potential interactive effects of E2F2_-5368_A>G, CCND1_870_A>G and CCND3_-677_C>T on HER2 status we performed interaction analysis. We observed no interactions among the 3 polymorphism, therefore the observed effects on HER2 status appear to be independent (data not shown).
In silico analyses of promoter polymorphisms E2F2_ -5368_A>G and CCND3_-677_C>T led to the identification of putative binding sites for transcription factors. The E2F2_ -5368_A>G leads to the formation of a putative SRY transcription factor binding site and CCND3_-677_C>T leads to the loss of a potential HSR1 binding site. These findings facilitate a functional discussion of the observed effects (Fig. 1).
We investigated the impact of polymorphisms E2F2_ -5368_A>G, CCND1_870_A>G and CCND3_-677_C>T on clinical and histopathological characteristics of breast tumors within a collection of incident breast cancers of the GENICA breast cancer case-control study. None of the polymorphisms showed an association with breast cancer risk which is in line with published data.28–33 In the absence of any primary breast cancer risk associations, we observed significant associations with the HER2 status of breast tumors for all 3 polymorphisms with strongest effects observed for the E2F2 and CCND3 polymorphisms. These effects were independent of each other and showed no interactive effect.
In the case of E2F2_-5368_A>G (rs760607), the observed effect was protective in those patients with the minor G allele had a lower likelihood to overexpress HER2 in their breast tumors. The polymorphism is located within the promoter region of the E2F2 gene, however, any functional consequences are unknown. While the exchange from A to G leads to the formation of a putative SRY transcription factor binding site, there is a possibility that this novel SRY site may contribute to an increase of E2F2 expression and that in turn, higher E2F2 levels may provoke higher levels of factors known to down-regulate HER2. For example, in breast carcinoma cells E2F has been found to regulate integrin alpha6beta4 and to promote invasion.34 In colon cancer cells, integrin alpha5/beta1 was found to interact with HER2 and to control HER2 expression.35 Furthermore, HER2 (ErbB2) is a target of STUB1 (CHIP), a chaperone-dependent E3 ubiquitin ligase that mediates HER2 stability.36, 37
Cyclins D are overexpressed in human cancers and commonly deregulated during oncogenesis.38, 39 The CCND1_870_A>G (rs603965) polymorphism affects the exon 4/intron 4 splicing donor site resulting in a variant lacking exon 5 and prolonged half-life of the transcript.40 Since CCND1 interacts with several transcription factors to regulate their activity,41 there may be a possibility that increased levels of CCND1 are involved in the up-regulation of factors involved in down-regulation or degradation of HER2 leading to HER2 negative breast tumors.
The CCND3_-677_C>T (rs3218086) polymorphism is located in the promoter region of CCND3. The exchange from C to T leads to the loss of a potential HSR1 binding site which may cause lower expression of CCND3. Since CCND3 is able to interact with transcription factors, e.g., hATF5,42 there may be a possibility that lower levels of CCND3 decreases levels of factors down-regulating HER2 and therefore may lead to higher expression of HER2 in breast tumors.
For all the 3 polymorphisms, we observed similar effects of potentially increased expression of E2F2, CCND1 and CCND3 with lower expression of HER2 in breast tumors and vice versa lower expression of the analyzed cell cycle regulators with increased expression of HER2 (Fig. 1). We conclude that the analysis of the E2F2_-5368_A>G, CCND1_870_A>G and CCND3_ -677_C>T polymorphisms on the constitutional DNA level may hold the potential to provide molecular marker information for the prediction of the HER2 status of breast tumors. Therefore, they might give further insight into underlying mechanisms of HER2 expression in tumor tissues and potentially support the discovery of novel prevention strategies or therapy targets while exploring associated pathways.
Our findings are based on a collection of incident breast cancer cases recently recruited in the Greater Bonn area within a Germany population-based case-control study. HER2 assessment was available for more than 660 cases performed at a single diagnostic centre and by expert pathologists using standard HER2 immunohistochemistry. Yet, according to current standards developed within international association studies our findings must be considered preliminary because of the limited study size. From the point of view of multiple testing, our findings of E2F2 and CCND3 polymorphisms being involved in the modulation of HER2 breast tumor levels were robust. It is therefore our intention to encourage larger collaborative studies in order to validate our findings, which may be particularly important in the light of potential clinical implications.
We are indebted to all women participating in the GENICA study. We gratefully acknowledge support by interviewers as well as physicians and pathologists of the study region.