Introduction

Carriers of germline mutations in BRCA1 or BRCA2 have a hereditary predisposition for developing breast and/or ovarian cancer. Several studies have indicated that the genetic makeup of BRCA1/2 related breast cancer is different from that of sporadic breast cancer. These differences comprise gains and losses of specific parts of chromosomes as well as differences in gene expression [1–6]. In line with this, the morphological and immunohistochemical phenotype of BRCA1 related breast cancer is also different [7, 8]. They often concern well demarcated medullary and poorly differentiated ductal cancers with conspicuous lymphocytoplasmic infiltrates [9, 10] that are of high-grade [11] and show high proliferation [12]. In addition, they do not express estrogen (ER), progesterone (PR) or HER-2/neu receptors [13], often lack p27Kip1 [14], but do accumulate p53 [15], and overexpress cyclin E [16], cytokeratins (CK) 5/6 and 14 [17, 18], and EGFR [19–21]. These observations point to a carcinogenetic pathway of BRCA1 related breast cancers different from that in sporadic cancers.

Hypoxia is a hallmark of many sporadic cancers [22]. Hypoxia inducible factor-1 (HIF-1) is the key regulator of the hypoxia response. HIF-1 consists of 2 subunits, HIF-1α and HIF-1β. While HIF-1β is constitutively expressed, the HIF-1α protein is continuously degraded under normoxia by the ubiquitin-proteasome pathway [23, 24]. Under hypoxia, HIF-1α protein degradation is inhibited resulting in its overexpression and subsequent binding to HIF-1β [24]. This HIF-1 complex then regulates the expression of its target genes through binding with hypoxia responsive elements in the promoter regions of these genes [25].

The overexpression of HIF-1α has been demonstrated in several types of cancer, with a negative impact on therapy response and prognosis [26–28]. In sporadic breast cancer, previous studies have demonstrated that HIF-1α overexpression plays a role in breast carcinogenesis [29–32] and is correlated with a poor prognosis in invasive breast cancer [31, 33, 34].

Little is known of the putative role of HIF-1α in hereditary breast carcinogenesis. A recent study suggested that BRCA1 plays a role in the hypoxic response by regulating HIF-1α stability and by modulating expression of vascular endothelial growth factor, a major downstream target of HIF-1α [35]. The aim of this study was therefore to examine the expression of HIF-1α in BRCA1 related breast cancer to find clues for its putative role in the BRCA1 carcinogenesis.

Materials and methods

Patients

The study group comprised 30 invasive breast cancer cases from 17 patients with a proven BRCA1 germline mutation and 13 patients with invasive breast cancer who were not screened for mutations themselves, but were known to have a BRCA1 mutation in their family. All these patients were derived from the Familial Cancer Clinic of the VU University Medical Centre, Amsterdam. Use of anonymous or coded left over material for scientific purposes is part of the standard treatment contract with patients [36].

As sporadic controls, data from our previous study [33] on invasive breast cancers from patients unselected for family history were used.

Histopathology

Tumor size was measured in the fresh resection specimens, and tumor samples were subsequently fixed in neutral buffered formaldehyde, and processed to paraffin blocks according to standard procedures. A tissue array block was made as previously described [37].

About 4 μm thick sections were cut and stained with H&E for histopathology. Tumor type was assessed according to the WHO, and tumors were graded according to the Nottingham grading system. Mitoses counting was performed as previously described [38]. Presence of necrosis was noted. Scoring was performed by one observer (PJvD) who was blinded to the origin of the tumors.

Immunohistochemistry

After deparaffination and rehydration, target retrieval solution (DAKO) was used for antigen retrieval with all slides placed in a water bath for 45 min at 97°C. A cooling off period of 20 min preceded the incubation of the HIF-1α mouse monoclonal (BD Biosciences, Pharmingen, Lexington, USA), at a dilution of 1:500. The catalysed signal amplification system (DAKO) was used to detect HIF-1α as before [25]. For ER, PR, HER-2/neu, EGFR, Ki67, p53, p27 and p21 antigen retrieval was performed in an autoclave with the slides placed in a citrate buffer (pH 6). For Glut-1, CAIX and P300 antigen retrieval was performed in citrate buffer, pH = 6.0, for 20 min at 100°C and for CK5/6 and CK14 an EDTA buffer (pH 9) was used. A cooling off period of 30 min preceded the incubation (60 min at room temperature) with the primary antibodies. Mouse monoclonal antibodies used were: ER (1:50, DAKO), PR (1:50, Novocastra, Newcastle upon Tyne, United Kingdom), HER-2/neu (1:10,000, Prof. M. van der Vijver, Dutch Cancer Institute, Amsterdam, The Netherlands), EGFR (1:10, Novocastra), CK5/6 (1:3000, Chemicon, Temecula, USA), CK14 (1:400, Neomarkers, Lab Vision Corp, Fremont, CA, USA), Ki67 (1:40, MIB-1, Immunotech, Marseille Cedex, France), p53 (1:500, DAKO), p27 (1:1000, BD Biosciences Transduction laboratories, Lexington, USA), p21 (1:50, BD Biosciences, Pharmingen).

Polyclonal primary antibodies used were: Glut-1 (1:200, DAKO), CAIX (1:1000, Abcam, Cambridge Science Park, Cambridge, UK), P300 (1:200, clone N15, Santa Cruz, CA, USA) For detection of the primary antibodies against CK5/6, CK14, CAIX and p300, a poly HRP anti Mouse/Rabbit/Rat IgG (ready to use, ImmunoLogic, ImmunoVision Technologies, Brisbane, USA) was used for the other primary antibodies a biotinylated rabbit anti-mouse antibody (DAKO) or a biotinylated swine anti-rabbit antibody was used. The signal was amplified by avidin-biotin complex formation. All slides were developed with diaminobenzidine followed by haematoxylin counterstaining. Before the slides were mounted all sections were dehydrated in alcohol and xylene.

Scoring was performed by one observer (PJvD). HIF-1α, ER, PR, Ki67, p53, p27 and p21 staining was usually confined to the nucleus. Diffuse cytoplasmic staining was sometimes seen but ignored, estimating the percentage of positively stained nuclei. p300 nuclear staining intensity was in accordance with our previous study (39) scored as negative, 1+, 2+ or 3+ and for further statistical analysis grouped as negative (neg, 1+) or positive (2+, 3+). HIF-1α was regarded overexpressed when >1% of nuclei were positive as described before [27], and the expression pattern (perinecrotic or diffuse) was noted [32]. HER-2/neu, EGFR, and CAIX stainings were scored positive when a clear membrane staining pattern was seen. Glut-1 expression was scored positive if a clear membrane or a distinct cytoplasmic staining was seen, and Ck5/6, Ck14 were scored positive in case of cytoplasmic staining.

Results

High levels of HIF-1α expression were detectable in 27/30 (90%) of the hereditary breast cancer cases, compared to 88/200 (44%) of the sporadic controls (P < 0.0001) (Table 1). Necrosis was present in 19/30 (63%) of the hereditary cases compared to 38/200 (19%) of controls (P < 0.0001). In 19/27 of the hereditary cases that showed HIF-1α expression, a perinecrotic staining pattern was observed and in 8/27 a diffuse pattern was seen, compared to 27/88 (31%) and 61/88 (69%) of sporadic cases, respectively (P = 0.0002). Glut-1 expression was detected in 27/30 hereditary cases and CAIX in 15/21 cases, and both were correlated with HIF-1α overexpression (P-value < 0.001 for both). In the sporadic cases the expression of Glut-1 and CAIX was 29% (54/183) and 13% (24/183) respectively. p300 levels were significantly higher in hereditary cases 95% (20/21) compared to hereditary cases 50% (91/183), (P = 0.0001). Furthermore, high levels of HIF-1α expression in these hereditary breast cancers were associated with a poor histological grade (P = 0.061) and EGFR expression (P = 0.099) (Table 2). HIF-1α correlated significantly negatively with the presence of ER (P = 0.033), PR (P = 0.001), and HER-2/neu (P = 0.001). For the remaining markers no significant correlations with HIF-1α expression were found.

Table 1 Expression of HIF-1α in hereditary and sporadic breast cancers in relation to various clinicopathologic features and HIF-1α downstream genes
Full size table
Table 2 Expression of HIF-1α in hereditary breast cancers in relation to various other immunophenotypic markers
Full size table

About 21/30 (70%) cases were both HIF-1α and EGFR positive. All of these HIF-1α and EGFR positive cases were Glut-1 positive, 14/21 of these cases showed a perinecrotic HIF-1α expression pattern and the remaining cases showed a diffuse HIF-1α expression pattern. In nine of the perinecrotic HIF-1α cases CAIX staining was also present. In the seven diffuse HIF-1α cases four cases were CAIX positive.

Discussion

The aim of this study was to examine the expression of HIF-1α in BRCA1 related breast cancers to establish whether the HIF-1α pathway plays a role in the BRCA-1 carcinogenesis and progression. About 90% of BRCA1 related breast cancers showed expression of HIF-1α, a percentage significantly higher than in sporadic controls. Mostly, this concerned the perinecrotic type of HIF-1α expression. Necrosis, likely caused by the well known rapid tumor cell proliferation of these hereditary cancers while the vasculature is lagging behind, was clearly more present than in the sporadic cancers. Likewise, a perinecrotic pattern of overexpression of HIF-1α was more frequent in BRCA1-related than in sporadic breast cancers. The perinecrotic type of HIF-1α expression was accompanied by overexpression of the HIF-1α downstream genes Glut-1 and CAIX, pointing towards functional HIF-1α. This type of HIF-1α overexpression is thought to be caused by (severe) hypoxia, whereas diffuse HIF-1α overexpression at (relative) normoxia is thought to be induced by growth factors like HER2 [39], HIF-1α gene amplifications [40] or mutations [41], or by other oncogenes or loss of tumour suppressor genes. We have previously shown that, compared to a diffuse staining pattern, perinecrotic HIF-1α overexpression is associated with the worst survival of sporadic breast cancer patients [33].

The present results are in contrast with a recent in vitro study where increased levels of BRCA1 were seen to increase the response of the VEGF promoter to hypoxia in a HIF-1α dependent fashion [35]. In that study, reduced levels of BRCA1 protein reduced the ability of hypoxia to induce VEGF. We, however, observed marked upregulation of HIF-1α in human BRCA1 related breast cancers. In view of the frequent presence of necrosis and perinecrotic HIF-1α expression, we hypothesize that in breast cancers in BRCA1 germline mutation carriers, hypoxia overrides the potential negative effect of BRCA1 expression loss on HIF-1α expression, yet leading to frequent perinecrotic HIF-1α expression and subsequently to activation of HIF-1 downstream genes. In line with this, p300 expression levels, a prerequisite for HIF-1 downstream activation, were high in hereditary cancers.

Whilst we previously reported frequent overexpression of EGFR in hereditary breast cancers [19, 42], we now find concomitant expression of HIF-1α and EGFR in 70% of BRCA1 related breast cancers. Furthermore, all of these cases had evidence of HIF1 downstream activation, suggesting that EGFR enhances the hypoxic response.

Several previous studies have elucidated in vitro the role of both the PI3K and the MAPK pathway in the induction of HIF-1α, including its upregulation by HER-2/neu. In addition, the upregulation of EGFR has been related to elevated levels of downstream targets of HIF-1α, like VEGF and survivin [43]. This suggests a role for specific oncogenes in the (normoxic) induction of HIF-1α. The association between HIF-1α and EGFR might be explained by the EGFR induced activation of the PI3K/PTEN/AKT/FRAP pathway, through which HER-2/neu also acts on HIF-1α [44, 45]. Further studies will have to elucidate the role of EGFR in the carcinogenesis of BRCA1 related breast cancer. Recent in vitro studies on breast basal-like cell lines showed that these cell lines are more sensitive for EGFR inhibitors and for carboplatin with a synergistic effect when these are combined [46]. This might lead to new therapy strategies for BRCA1 related breast cancer patients.

In contrast to EGFR, an association between HIF-1α and HER-2/neu as observed in previous studies [27, 39] could not be confirmed in this study. 26/27 (96%) of HIF-1α positive cases were HER-2/neu negative, in which the usual HER-2/neu negativity of hereditary breast cancers likely plays a role.

We conclude that the BRCA1 germline mutation related breast cancers show a high frequency of HIF-1α overexpression. In view of the predominantly perinecrotic staining pattern, overexpression of HIF-1α in hereditary breast cancer seems to be caused by hypoxia rather than by activation of oncogenes or inactivation of tumor suppressor genes. However, the frequent overexpression of EGFR and concomitant expression of EGFR and HIF-1α may open up news ways of treatment of BRCA1 related breast cancer by targeting EGFR.