A series of 55 primary tumors from breast cancer patients diagnosed between 2002 and 2004 at the University Hospital of Liege, Belgium, was analyzed. The mean age of the patients was 61.9 years and the median was 59.0 years (range: 38.0 to 88.0 years). The clinicopathological data of the patients are summarized in Table 1. The tumor samples were fixed in 10% buffered formalin and embedded in paraffin. The histological diagnosis was confirmed by reviewing the original sections of the primary tumors. All of the tumors were simultaneously evaluated for histological type and grade by senior pathologists. The most representative blocks were selected and cut into new 5-μm-thick sections for immunohistochemical analyses. The study was approved by the local ethics committee at the Liege University Hospital.

Sections of breast biopsy specimens underwent immunoperoxidase staining using antibodies directed against AP-2α (1:100) (#39001; Active Motif, Carlsbad, CA, USA), against YY1 (1:50) (H-10; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA), or against ERBB2 (1:250) (A0485; Dako A/S, Glostrup, Denmark). The sections were deparaffinized in xylene and rehydrated in methanol. Endogenous peroxidases were blocked by 5% H₂O₂ treatment. For better antigen retrieval, the samples were boiled either in a microwave oven for 3 × 5 minutes in citrate buffer (AP-2α and YY1) or in a water bath at 99°C in EDTA (ethylenediaminetetraacetic acid) buffer for 40 minutes (ERBB2). Samples were then washed with phosphate-buffered saline-Tween (pH 7.2; 1.5%) and incubated with the primary antibody at room temperature for 30 minutes (AP-2α and YY1) or 1 hour (ERBB2). After washings, the revelation was performed with the use of appropriate secondary antibodies and the LSAB2 system (AP-2α and YY1; Dako A/S) or the EnVision kit (ERBB2; Dako A/S) according to the supplier's recommendations. Immunoreactivity was visualized by a treatment with diaminobenzidine (Sigma-Aldrich, St. Louis, MO, USA), and the slides were counterstained with Mayer's hematoxylin.

For statistical analyses of AP-2α and YY1 immunoreactivity, the percentage distribution of stained tumor cell nuclei in the sample was divided into low (<80%) or high (≥ 80%) expression groups according to Pellikainen and colleagues 26. ERBB2 scoring was performed according to the recently proposed guidelines from the American Society of Clinical Oncology (ASCO) and the College of American Pathologists (CAP) 27. Pathological ERBB2 overexpression (3+) was detected in 27% of tumors. There was a significant statistical association between ERBB2 overexpression and Ki67 immunostaining, and an inverse relationship was demonstrated with estrogen receptor and progesterone receptor status (Table 1). Furthermore, a trend toward a direct link between ERBB2 overexpression and both p53 expression and histological grade III was observed.

Fluorescent in situ hybridization (FISH) was performed with the INFORM HER-2/neu probe (approved by the U.S. Food and Drug Administration) and the BenchMark XT automated system (Ventana Medical Systems, Inc., Tucson, AZ, USA) according to the supplier's recommendations. A minimum of 50 cell nuclei were counted, and gene amplification was considered as present when an average of more than six ERBB2 gene copies per cell was observed 27.

The statistical analyses were carried out with a χ² test for categorical variables at a p value of less than 0.05 for significance by using Statistica software (StatSoft, Inc., Tulsa, OK, USA).

The BT-474 human mammary carcinoma cells were purchased from the American Type Culture Collection (Manassas, VA, USA) and cultured in the RPMI 1640 medium supplemented with 10% (vol/vol) fetal bovine serum, 2 mM glutamine, and 100 μg/mL penicillin/streptomycin (all from Cambrex Bio Science Verviers S.p.r.l., Verviers, Belgium).

Cells were transfected (a) on days 0 and 2 by 150 nM siRNA directed against AP-2α and/or AP-2γ transcripts as indicated or (b) on day 0 by 30 nM siRNA against YY1, or 100 nM combined siRNAs against AP-2α and AP-2γ transcripts, or both as indicated. As control, cells were transfected by either an siRNA against luciferase mRNA 28 or a negative control siRNA OR-0030-neg05 from Eurogentec S.A. (Seraing, Belgium). Total RNA was extracted after 2 to 4 days of treatment. Real-time reverse transcription-polymerase chain reaction (RT-PCR) for AP-2α, AP-2γ, ERBB2, and β2-microglobulin (standard gene) transcripts were performed on 1 μg of total extracted RNA. The standardized transcript levels were reported to the values obtained in cells transfected with the control siRNA. The RT-PCR analysis was performed on an ABI Prism 7000 apparatus (Applied Biosystems, Foster City, CA, USA) using standard protocol. Western blot analysis was performed on proteins extracted after 1 or 3 days of treatment as indicated. The antibodies used for Western blot were 3B5 for AP-2α, 6E4/4 for AP-2γ, H-10 for YY1, and C-19 for Ku70 (all purchased from Santa Cruz Biotechnology, Inc.) and a rabbit antibody for ERBB2 (06–562; Upstate, now part of Millipore Corporation, Billerica, MA, USA). The sequences of the siRNAs and the RT-PCR primers (all purchased from Eurogentec S.A.) are presented in Table 2.

We first detected levels of ERBB2, AP-2α, and YY1 proteins by immunohistochemistry (IHC) in tumor specimens from 55 cases of breast carcinomas (Tables 1 and 3). Representative examples of the staining patterns obtained for ERBB2 receptor and the transcription factors AP-2α and YY1 are shown in Figure 1. The AP-2α and YY1 proteins were detected mainly in the nuclear compartment, while cytoplasmic staining was rare and weak (Figure 1a–d). We scored the AP-2α and YY1 levels as low or high regarding the percentage of stained nuclei according to Pellikainen and colleagues 26 (Figure 1). High AP-2α and YY1 protein levels were seen in 42% and 45% of breast carcinomas, respectively. For ERBB2, we considered only membranous staining (Figure 1e–h). Scoring was carried out according to the ASCO/CAP guidelines 27.

We then analyzed the correlations between the levels of ERBB2, AP-2α, and YY1 proteins in the tumors. Our statistical analyses showed that ERBB2 expression was significantly associated with a high AP-2α transcription factor level (p = 0.003) (Table 3). Accordingly, 83% of the tumors with high AP-2α level had a 2+ or 3+ IHC score for ERBB2 protein and only 17% had a low ERBB2 expression (0 to 1+). On the other hand, 83% of the tumors with ERBB2 low expression had a low level of AP-2α protein. In contrast, no association between ERBB2 and YY1 expression was observed (Table 3). However, there was a significant association of ERBB2 protein level with combined overexpression of AP-2α and YY1 transcription factors (p ≤ 0.02) (Table 3). Indeed, among the 19 cases presenting high levels of both AP-2α and YY1, 84% had an ERBB2 2+ or 3+ score (Table 3).

FISH demonstrated ERBB2 gene amplification in 11 cases of breast cancer. All of these had an ERBB2 3+ score in IHC. Moreover, out of the 15 ERBB2 3+ cases, 11 were FISH-positive, 2 showed no ERBB2 amplification, 1 was borderline with an average of 4 copies of ERBB2 gene per cell, and 1 was undetermined due to lack of material for FISH testing. Interestingly, the 2 cases of ERBB2 3+ immunostaining without gene amplification and the borderline case showed high AP-2α and YY1 levels. On the other hand, all the ERBB2 3+ cases with low levels of both AP-2α and YY1 showed ERBB2 gene amplification. Furthermore, when considering the entire ERBB2 expressing group (1+, 2+, 3+), we observed a significant inverse correlation between ERBB2 FISH status on one hand and AP-2α and YY1 levels on the other hand (p = 0.017 and 0.029, respectively) (Table 4). In particular, 80% of the cases with high levels of both AP-2α and YY1 proteins did not present ERBB2 gene amplification (p = 0.006) (Table 4). These results suggest that when the ERBB2 gene is not amplified, the ERBB2 expression may rely partially on AP-2α and YY1 transcription factor levels. Accordingly, considering only the FISH-negative cases (n = 34), the correlation between ERBB2 and AP-2α levels was higher than in the whole group (r = 0.67, p < 0.001 compared with r = 0.31, p = 0.022 in the whole group). Interestingly, the percentage of ERBB2 2+3+/AP-2α-low cases decreased (12% in the FISH-negative group compared with 24% in the entire group) (Table 5). Similarly, the percentage of ERBB2-positive/AP-2α-low/YY1-low cases diminished (from 22% in the whole group to 9% in the FISH-negative group) (Table 5). These results indicate that high ERBB2 expression may result either from gene amplification or from increased transcription factor levels.

After these correlation results in breast cancer specimens, we sought to study these relationships on the functional side. Although the AP-2 family is known to activate the ERBB2 promoter in reporter vectors, the effect of these transcription factors on the expression of the endogenous ERBB2 gene has not been clearly established. To find out whether AP-2 factors do contribute functionally to ERBB2 overexpression in vivo, we measured ERBB2 mRNA levels in breast cancer cells in which the expressions of AP-2α and AP-2γ were downregulated by siRNAs. BT-474 breast cancer cells overexpressing ERBB2, both through amplification and enhanced transcription, were transfected with AP-2α and AP-2γ siRNAs, both independently and in combination for several days. Two different siRNAs against AP-2α and three distinct siRNAs against AP-2γ were first tested (data not shown). The best one for each target, as tested by Western blotting, was further used in the study. First, AP-2α and AP-2γ transcript levels were quantified by real-time RT-PCR in the cells transfected by the AP-2 siRNAs and reported to transfection of a control siRNA. Transfection of AP-2α siRNA alone inhibited AP-2α expression (Figure 2a) but did not modify the AP-2γ transcript level (Figure 2b). Comparable results were obtained in cells transfected with AP-2γ siRNA alone (Figure 2a,b). Moreover, transfection of both AP-2α and AP-2γ siRNAs induced a decrease in both mRNA levels (Figure 2a,b). Similar results were obtained at the protein level (Figure 2c, Western blotting). Markedly, 3 days after treatment with the combination of AP-2α and AP-2γ siRNAs, both factors were undetectable (Figure 2c, lane 4), demonstrating that the siRNAs specifically inhibited the expression of their targets. We then quantified ERBB2 transcript level by real-time RT-PCR in those cells transfected by the siRNAs. Either the AP-2α siRNA or the AP-2γ siRNA alone produced a small transient downregulation of the ERBB2 transcript level (Figure 2d). Interestingly, transfection of both AP-2α and AP-2γ siRNAs induced a significant decrease in the endogenous ERBB2 mRNA level (Figure 2d). This result was also obtained with another set of siRNAs against AP-2α and AP-2γ (data not shown). Moreover, the inhibition was seen at the protein level (Figure 2e, lane 2). Indeed, we observed a reduction of ERBB2 protein level to 57% of control upon transfection of both siRNAs directed against AP-2α and AP-2γ. Going deeper, we further added an siRNA directed against YY1 and observed that the ERBB2 protein level could be decreased even further to 22% of control by this combination of siRNAs (Figure 2e, lane 4). These results are strong evidence that AP-2 and YY1 transcription factors effectively participate in ERBB2 expression in breast cancer cells.

Surprisingly, we also observed a strong association between AP-2α and YY1 expression levels in the primary breast tumors (p < 0.001) (Table 6). Indeed, the majority of the tumors contained either low levels (26 cases) or high amounts (19 cases) of both proteins. In contrast, there were only 10 cases (18%) with a high level of either AP-2α or YY1 alone (Table 6). Next, we analyzed the relationship between these factors in a breast cancer cell line. We observed that transfection of an siRNA directed against YY1 reduced both AP-2α and AP-2γ protein levels (Figure 2e, lane 3). Additionally, the combination of siRNAs directed against both AP-2α and AP-2γ also diminished the YY1 protein level (Figure 2e, lane 2). These results suggest that, besides the cooperation between AP-2 and YY1 transcription factors on the ERBB2 promoter, there is an intricate relationship between the expressions of these proteins.