ATBF1-A Messenger RNA Expression Is Correlated with Better Prognosis in Breast Cancer

Tissue-specific expression of the human α-fetoprotein (AFP) gene is strongly stimulated by an enhancer element located 3.3 to 4.9 kb upstream of the transcription initiation site (1, 2). One of the enhancer elements contains an AT-rich core sequence (called the AT motif; ref. (3). To determine the nuclear factor in hepatoma cell lines that interacts with the human AFP enhancer AT motif, Morinaga et al. (4) screened a hepatoma cDNA expression library with an AFP enhancer fragment that bore the AT motif. They succeeded in isolating a cDNA that coded for an AT motif-binding factor, termed ATBF1. This was the largest DNA-binding protein identified to that time and the first protein shown to contain multiple homeodomains and multiple zinc finger motifs. The protein had a predicted mass of 306 kDa and contained four homeodomains and 17 zinc finger motifs. In 1995, a second ATBF1 cDNA was isolated (5). It was called ATBF1-A, contained 23 zinc finger motifs and was 3.3 kb longer than the previously reported clone, now identified by the term ATBF1-B. Analysis of the expression and function of this ATBF1-A gene has subsequently shown that it plays a role in the differentiation of a variety of cell types (5, 6). It was found that AFP expression is inhibited by ATBF1-A and that the absence of ATBF1-A is a distinct feature of AFP-producing gastric cancer cells, which are characterized by extremely high malignancy (7). ATBF1-A protein has also been reported to bind to c-Myb and repress its transcription activity, which may result in changes in cell growth and differentiation (8). Recently, it was revealed that in AFP-producing gastric cancer cells treatment with mitomycin can induce ATBF1 gene expression, which enhances the promoter activity of the p21 (Waf1/Cip1) gene and the activation of p21 (Waf1/Cip1) promoter would lead to the arrest of AFP-producing gastric cancer cell growth (9). These results together may imply that altered gene expression or biological activity may be associated with malignant transformation (10). The human ATBF1 transcription factor gene has been assigned to chromosome 16q22.3-q23.1 (11). Our previous and others studies showed that a loss of heterozygosity on the long arm of chromosome 16 is one of the most frequent genetic events in breast cancer, which implies the presence of a tumor suppressor gene (12, 13). This led us to hypothesize that there was a link between levels of ATBF1 expression and the metastatic potential of breast cancer and also, therefore, the prognosis of these patients. The regulation of AFP gene by ATBF1-A as shown in AFP-producing gastric cancer also gives us a hint that ATBF1-A gene also plays some role in other malignant tumors such as breast cancer. It is well known that AFP production decrease rapidly after birth and the protein is rarely found in adult tissue, and elevated levels of AFP in adults, which occur following the reactivation of the gene, are indicative of several disease conditions, most notably malignant liver cancer. Therefore, understanding the system of AFP gene regulation might lead us to a greater appreciation of the process involved in carcinogenesis and also the general principles concerned in cell differentiation.

At present, the expression of ATBF1-A and its clinicopathologic importance remains unclear in breast cancer. Through the use of quantitative real-time reverse transcriptase-PCR, done using LightCycler, and inmmunohistochemistry we report here on a correlation between levels of ATBF1-A expression and several clinicopathologic factors, in samples taken from 153 female patients with invasive carcinoma of the breast. To our knowledge, this is the first report of the quantitative expression of the ATBF1-A gene and protein in breast cancer.

Patients and Tumor Samples. Primary invasive breast carcinoma specimens were obtained, by surgical excision, from 153 female patients at the Department of Breast and Endocrine Surgery, Nagoya City University Hospital, Nagoya, Japan between 1992 and 2000. Informed consent was obtained from all patients before surgery. The ethics committee of Nagoya City University Graduate School of Medicine, Nagoya, approved the study protocol. The median age of the patients was 53 years (range, 34-88 years). The patients' tumors were classified using the International Union Against Cancer staging system as follows: 40 cases were classified as stage I, 95 cases as stage II, 16 cases as stage III, and 2 cases as stage IV. As postoperative adjuvant treatment, tamoxifen was given to patients with estrogen receptor-α (ERα) and/or progesterone receptor (PR)–positive tumors. Depending on tumor stage, the following chemotherapy regimens were given: oral 5-fluorouracil, CMF (100 mg p.o. cyclophosphamide, days 1-14; 40 mg i.v. methotrexate, days 1 and 8; 500 mg i.v. 5-fluorouracil, days 1 and 8), or CEF (500 mg cyclophosphamide; 60 mg epirubicin; 500 mg 5-fluorouracil; every 3 or 4 weeks). Since 1995, postoperative treatment has been done with reference to the recommendation of St. Gullen (14). After recurrence, patients with ERα- and PR-negative tumors were treated with CMF, CEF, and taxanes. Patients with hormone receptor–positive tumors and nonvisceral metastases were treated with endocrine therapy, such as antiestrogens, aromatase inhibitors, and medroxyprogesterone acetate. Patients were followed postoperatively, every 3 months. The median follow-up period was 61 months (range, 48-144 months).

Patients were graded histopathologically according to the modified Bloom and Richardson method proposed by Elston and Ellis (15). Samples were snap frozen in liquid nitrogen and stored at −80°C until RNA extraction.

Isolation of Total RNA and Reverse Transcription. Total RNA from homogeneous breast cancer tissue, which was microscopically confirmed, was isolated from ∼500 mg of frozen specimen. Total RNA was also isolated from one flask of the human hepatoma HepG2 cell line for use as a positive control (7) and to generate standard curves, which was kindly provided by Dr. N. Harada (Division of Molecular Genetics, Institute for Comprehensive Medical Science, Fujita Health University, Toyoake, Aichi, Japan; ref. (16). mRNA was isolated using the TRIZOL reagent (Life Technologies, Inc., Tokyo, Japan) according to the manufacturer's instructions. Reverse transcription reactions were done as previously described (17).

Primers and Probes. We conducted Blast searches (Genbank) to confirm the specificity of the nucleotide sequences chosen for the primers and probes and to confirm the absence of DNA polymorphism. To avoid detection of contaminating genomic DNA, the primers were located at exon 2 and exon 3. The specific oligonucleotide primers were synthesized according to published information on the ATBF1-A gene as follows: sense primer, 5′-GGACGCCCAGTTCATGATG-3′ (3319-3337) and antisense primer, 5′-TCGCCCAGGTTCATCAGCT-3′ (3468-3450). The PCR product size is 150 bp. The donor probe 5′-GCCATGACGCCTGCTCTAGTGGG-3′ has a fluorescein label at its 3′ end. The acceptor probe 5′-GGTGAGATCCCCCCTAGACATGCG-3′ has LC Red 640 at its 5′ end.

To ensure the fidelity of mRNA extraction and reverse transcription, all samples were subjected to PCR amplification with oligonucleotide primers and probes specific for the constitutively expressed gene glyceraldehyde-3-phosphate dehydrogenase (GAPDH) and normalized. GAPDH primers were as follows: forward primer, 5′-AAATCAAGTGGGGCGATGC TG-3′ and reverse primer, 5′-GCAGAGATGATGACCCTTTTG-3′. The sequences of the GAPDH probes used for real-time LightCycler PCR were 5′-AGAAGGCTGGGGCTCATTTGCAGGG-3′ and 5′-GTCCACTGGCGTCTTCACCACCATG-3′. All primers and probes were purchased from the Japanese Gene Institute (Saitama, Japan).

Real-Time Reverse Transcriptase-PCR. Real-time reverse transcription-PCR was done using a LightCycler (Roche Molecular Biochemicals, Mannheim, Germany) as previously reported (18). The PCR reaction was carried out in a final volume of 20 μL containing 2.4 μL of 25 mmol/L MgCl₂, 0.5μL of 20 pmol/μL sense primer and antisense primer, 0.4μL of 10 pmol/μL donor and acceptor probe, 2 μL PCR master mix, 1.5 μL cDNA, made up to 20 μL with water. After an initial denaturation step at 95°C for 60 seconds, temperature cycling was initiated. Each cycle consisted of denaturation at 95°C for 0 second, hybridization at 56°C for 5seconds, and elongation at 72°C for 6 seconds. The fluorescence signal was acquired at the end of the hybridization step. A total of 55 cycles were done. Cycling conditions for GAPDH were as follows: initial denaturation at 95°C for 60 seconds, followed by 50 cycles at 95°C for 0 second, 60°C for 5 seconds, and 72°C for 8 seconds.

Standard Curves and Presentation of Results. For each PCR run, a standard curve was constructed using serial dilutions of cDNA obtained from the HepG2 cell line. The level of expression of ATBF1-A mRNA was given as relative copy numbers normalized against GAPDH mRNA and shown as mean ± SD. Relative ATBF1 mRNA expression was calculated by the formula: (ATBF1-A/GAPDH) × 1,000.

A nontemplate negative control was included in each experiment. All of the nontemplate negative controls, the standard cDNA dilutions from the HepG2 cell line, and the tumor samples were assayed in duplicate. All of the patient samples with a coefficient of variation for gene mRNA copy number data >10% were retested using the method of Bieche et al. (19).

Immunohistochemical Staining of ATBF1-A Protein. Immunohistochemical staining of ATBF1-A protein from available 90 cases of paired tissues was done as previously reported (20). In brief, tissue sections were deparaffinized and rehydrated and boiled in target retrieval solution (DAKO, Carpinteria, CA) to improve staining. All sections were incubated with methanol containing 0.3% H₂O₂ 1.0% sodium azide to block endogenous peroxidase activity and were incubated with rabbit antiserum for ATBF1-A (at 1:10,000 dilution) overnight at 4°C. Immunoreactive products were detected using DAKO Envision+, then visualized after adding 3,3′-diaminobenzidine as the chromogen.

Immunohistochemical Staining of ERα and PR. Immunostaining of ERα and PR was done as previously described (21, 22). Briefly, the slides were incubated at a dilution of 1:100, with either anti-ERα primary antibody (ER1D5; DAKO, Kyoto, Japan) or anti-PR primary antibody (PgR636; DAKO, Kyoto, Japan), using the streptavidin-biotin system (SAB-PO kit, Nichirei, Tokyo, Japan) according to the manufacturer's instructions. The immunostaining of ATBF1-A, ERα, and PR was subjectively assessed by two independent investigators (Z.Z. and H.I.), and discordant results were resolved by consultation with a third investigator (H.Y.). Immunohistochemical scoring of ATBF1-A was scored by immunohistochemical score as previously described (21). The expression of ERα and PR was scored by assigning a proportion score and an intensity score according to Allred's procedure (23).

Statistical Analysis. The nonparametric Mann-Whitney Utest was used for the statistical analysis of the association between ATBF1-A expression and clinicopathologic factors. Disease-free survival (DFS) and overall survival curves were generated by the Kaplan-Meier method and verified by the log-rank (Cox-Mantel) and Breslow-Gehan-Wilcoxon tests. Cox's proportional hazard model was used for univariate and multivariate analyses of prognostic values. Differences were considered significant when a P < 5% was obtained.

Patient Demographics and Tumor Characteristics. Clinical characteristics are summarized in Table 1. The amount of ATBF1-A mRNA expressed in the samples from the 153patients ranged from 0 to 1975 relative copy numbers (mean, 438).

The Level of Expression of ATBF1-A mRNA Correlated with Axillary Lymph Node Metastasis and Tumor Size. The level of ATBF1-A mRNA expression in patients lacking axillary lymph node metastases (508 ± 464) was significantly higher than the level in the group with axillary lymph node metastases (317 ± 375; P = 0.0007). Additionally, it was found that the level of ATBF1-A mRNA in the group of patients with tumors measuring <2 cm (494 ± 386) was higher than in the group with tumors >2 cm (418 ± 460; P = 0.022). However, the level ofATBF1-A mRNA expression was not found to be significantly correlated with any other clinicopathologic factors, such as the patient's age, menopausal status, or histologic grade of the tumor (Table 1).

The Level of Expression of ATBF1-A mRNA Correlated with ERα Protein Expression but not with PR Protein Expression. A higher level of ATBF1-A mRNA expression was also found in the ERα-positive group of patients (486 ± 478) than that in ERα-negative group (337 ± 350; P = 0.042). However, there was no difference in the level of ATBF1-A mRNA expression between the PR-positive and PR-negative groups of patients (Table 1).

Patients whose Tumor Expressed Higher Levels of ATBF1-A mRNA Tended to Have Better Disease-Free and Overall Survival. To identify a clinically meaningful cutoff point for a level of ATBF1-A mRNA expression that could be used in disease prognosis analysis, various levels of ATBF1-A mRNA expression were tested using the Kaplan-Meier method and verified by the log-rank (Cox-Mantel) and Breslow-Gehan-Wilcoxon tests. When analyzing DFS the cutoff point for the level of ATBF1-A mRNA was set at 396: patients with a high level of ATBF1-A mRNA expression (912 ± 445, n = 53) tended to have a better prognosis than those with a low expression (187 ± 107, n = 100); log-rank (Cox-Mantel) test, P= 0.005; Breslow-Gehan-Wilcoxon test, P = 0.006 (Fig. 1A). When analyzing overall survival, the same method was adopted. When the cutoff point for the level of ATBF1-A mRNA was setat396: patients with a high level of ATBF1-A mRNA expression (912 ± 445, n = 53) tended to have a better prognosis than those with a low expression (187 ± 107, n = 100); log-rank (Cox-Mantel) test, P = 0.02; Breslow-Gehan-Wilcoxon test, P=0.04 (Fig. 1B).

Univariate and Multivariate Prognostic Analysis of ATBF1-A mRNA Expression in Breast Cancer for DFS and Overall Survival. From the limited 153 cases of breast cancer patients, it was found that ATBF1-A mRNA expression is an independent prognostic factor for DFS (Table 2) but not for overall survival (Table 3).

ATBF1-A Protein Expression in Breast Cancer and its Correlation with Clinicopathologic Factors. Specific ATBF1-A protein staining was found in the nucleus of some normal epithelial cells (Fig. 2A) and in the cytoplasm of the majority of breast cancer cells (Fig. 2B). We did not find any relationship between ATBF1-A protein expression and the clinicopathologic factors and we also did not find the correlation between ATBF1-A mRNA expression and protein expression (data not shown).

ATBF1-A and ATBF1-B transcripts are generated by alternative promoter usage combined with alternative splicing. ATBF1-A is a large transcription factor containing four homeodomains and 23 zinc finger motifs; the deduced protein is 404 kDa. It differs from ATBF1-B by an extra 3.3-kb sequence at the 5′ end and a 920-amino-acid extension at the NH₂ terminus (5). Previous studies on the functions of ATBF1-A showed that the absence of ATBF1-A is responsible for AFP gene expression in gastric cancer, and the absence of ATBF1-A is a distinct characteristic of AFP gene expression gastric cancer and might be important for its highly malignant nature (7). These previous studies also showed that ATBF1-A plays a role in the maintenance of the undifferentiated state of myoblasts, and its down-regulation is a prerequisite for the initiation of the terminal differentiation of murine C2C12 myoblasts (6). It also showed that ATBF1-A is expressed in the crypts and the bases of the villi of the small intestine and negatively regulates transcription of brush-border enzyme gene, aminopeptidase-N and proposed that ATBF1-A regulating aminopeptidase-N gene expression in the crypt-villus axis of the small intestine is a landmark of enterocyte differentiation and maturation (24). Taken together, all these observations helped formulate the proposal that altered expression or biological activity of ATBF1-A may be associated with malignant transformation (10).

However, the expression of AFBF1-A and its clinicopathologic function remains unclear in breast cancer. In the present study, it was found that ATBF1-A mRNA was expressed at significantly higher levels in breast cancer patients with no axillary lymph node involvement, with small tumors measuring <2 cm, and with ERα-positive tumors. However, no relationship was found between ATBF1-A mRNA expression and any of the other clinicopathologic factors under study (i.e., the patient's age, menopausal status, histologic grade of the tumor, or PR status). Patients with high levels of expression of ATBF1-A mRNA tended to have a better prognosis than those with low expression. Furthermore, the highly significant difference in ATBF1-A expression between the lymph node positive and lymph node negative groups (P = 0.0007) suggests that, in breast cancer, ATBF1-A levels may serve as a marker or indicator of lymph node metastasis. However, this needs to be confirmed at the protein level in a larger group of patients with a more homogeneous stage because it is the protein not the mRNA is predictive. The significantly higher level of ATBF1-A expression seen in ERα-positive tumors compared with ERα-negative tumors (P = 0.042) suggests that levels of this transcription factor might have potential as an indicator of responsiveness to endocrine therapy. However, this needs to be verified in a large cohort of endocrine therapy responders and nonresponders patients. In addition, ATBF1-A mRNA expression could have implication as a prognostic indicator for breast cancer progression as shown in this study by univariate and multivariate analyses. The correlation shown in this study, between increased ATBF1-A mRNA expression and better prognosis for a patient's DFS and overall survival, indicates that there are similarities in the nature of the ATBF1-A gene seen in breast and AFP-positive gastric cancer. This possible similarity is deduced from the fact that in AFP-positive gastric cancers, as discussed above, negative ATBF1-A expression is associated with extremely high malignancy. The reason for the significant correlation between increased ATBF1-A mRNA expression and better prognostic clinicopathologic factors in this study remains unknown. The human ATBF1 transcription factor gene has been assigned to chromosome 16q22.3-q23.1 (11). In previous studies, we and others have shown that loss of heterozygosity on the long arm of chromosome 16 is one of the most frequent genetic events in breast cancer, which implies the presence of a tumor suppressor gene (12, 13). Further study to examine loss of heterozygosity on the ATBF1-A gene is needed to clarify its relationships with the clinicopathologic factors as shown in the present study.

It is also of intriguing but highly preliminary to notice that the exporting of ATBF1-A protein, as shown in the present study by immunohistochemical, from nucleus of normal epithelial cells to the cytoplasm of breast cancer cells. The modulating of ATBF1-A protein shuttling between nucleus and cytoplasm compartment remains unknown and the function of ATBF1-A protein presented in the cytoplasm of breast cancer cells also remains unclear. We postulate that the positive staining of ATBF1-A in the cytoplasm of breast tumor cells may means that the proliferation status of tumor cells, which was supported by our unpublished observation that in rat breast feeding stage the positive staining is in the normal epithelium cytoplasm and after stopping of feeding for 10 days the cytoplasm staining decrease greatly.⁴

In conclusions, in breast cancer, levels of ATBF1-A may serve as a predictive indicator of lymph node metastasis. The results of this study also imply that ATBF1-A gene expression may have potential as both a marker of endocrine responsiveness and also as a prognostic indicator for breast cancer progression.

We thank Mariko Nishio for her excellent technical support.