FOXA1: Growth inhibitor and a favorable prognostic factor in human breast cancer

Abstract

The transcription factor Forkhead-box A1 (Foxa1), a member of the FOX class of transcription factors, has been implicated in the pathogenesis of lung, esophageal and prostate cancers. We have recently identified transcriptional activation of p27 by FOXA1. In this study, we analyzed the activities and expression pattern of FOXA1 in breast cancer. Forced expression of FOXA1 inhibited clonal growth of breast cancer cell lines, and FOXA1 levels inversely correlated with growth stimuli. In the estrogen receptor (ER)-positive MCF-7 cells, FOXA1 increased p27 promoter activity and inhibited the ER pathway activity. Analysis of FOXA1 expression in breast tissue arrays revealed significantly higher expression in pure ductal carcinomas in situ compared to invasive ductal carcinomas (IDC); and in IDC, high expression of FOXA1 was associated with favorable prognostic factors. Yet, FOXA1 expression was noted in a subset of the ER-negative tumors. Taken together, our findings suggest a growth inhibitory role for FOXA1, and identify it as a novel, potential prognostic factor in breast cancer. © 2006 Wiley-Liss, Inc.

The FOX class of transcription factors, now counting more than 100 members, is characterized by an evolutionary conserved 110 amino-acid DNA binding domain, known as the forkhead (FH) domain.1, 2 Three FOXA proteins, FOXA1, FOXA2 and FOXA3, are currently known and each shares a conserved structure, consisting of DNA binding domain and 4 transactivating regions, 2 in the C-terminal side of the protein and 2 in its N-terminus.3, 4, 5 FOXA1 is expressed in the liver, pancreas, bladder, prostate, colon, lung as well as mammary gland and can bind to the promoters of more than 100 genes associated with metabolic processes, regulation of signaling and the cell cycle.1, 2, 6 In mice, embryos carrying a homozygous null mutation for FOXA1 develop normally to term but suffer from severe postnatal growth retardation and hypoglycemia followed by death between postnatal days 2 and 12.7, 8

High expression of FOXA1 has been reported in various tumors, including lung, esophageal and prostate cancer.9, 10 In prostate cancer, current data suggest a growth inhibitory role for FOXA1. While FOXA1 is expressed in both preneoplastic lesions and adenocarcinomas, its expression is associated with markers of differentiation, and transfection assays revealed that FOXA1 had an inhibitory effect on the androgen receptor.11 Moreover, FOXA1 null prostate shows hyperplastic lesions.11

In breast cancer, studies of global gene expression revealed high expression of FOXA1 mRNA, often in association with the expression of the estrogen receptor alpha (ERα),12, 13 but also showed FOXA1 expression in a subset of ER-negative tumors.14 Among the ER-positive tumors, expression of FOXA1 mRNA was noted in tumors that showed favorable outcomes.15 In accordance with a growth inhibitory role of FOXA1 in breast cancer, studies in MCF-7 cells suggested downregulation of FOXA1 mRNA levels following estrogen stimulation.15, 16 We have recently identified a FOXA1 consensus site within a BRCA1 responsive element of the p27 promoter.17 Using electrophoretic mobility shift assay, we demonstrated direct binding of FOXA1 to the p27 promoter; and using transient transfection reporter assays in the HCT-116 colon cell line revealed that FOXA1 activated the promoter alone and in conjunction with BRCA1. Moreover, we discovered BRCA1-induced stabilization of FOXA1. These results further support a growth suppressive role for FOXA1.17 Interestingly, recent studies identified a role for FOXA1 as a coactivator of the ER, thus suggesting a growth promoting role for the protein.18, 19

Given the significance of FOXA1 in other malignancies, the aim of this study was to elucidate the expression pattern and transcriptional activity of FOXA1 in breast cancer. FOXA1 expression was analyzed in breast tissue arrays and its expression correlated with that of the ER, as well as with other favorable prognostic factors. Forced expression of FOXA1 inhibited clonal growth of breast cancer cell lines and transient transfection reporter assays revealed inhibition of ERE transcriptional activity and enhancement of p27 promoter activity in the ER-positive MCF-7 cells. Our results suggest FOXA1 as a favorable prognostic factor and identify 2 of the mechanisms that may be involved in its growth inhibitory activity in breast cancer.

Material and methods

Chemicals and antibodies

17β-estradiol (E2), 4-hydroxytamoxifen (4HT) and G418 were obtained from Sigma, MO. The antibodies used include anti-FOXA1 (C-20), anti-ERα (F-10) and anti-p27^Kip1 (C-19) all from Santa Cruz Biotechnology, CA; anti-GAPDH (Research Diagnostic, NJ); Horseradish peroxidase-conjugated anti-mouse IgG and anti-rabbit IgG (Amersham biosciences, NJ); Horseradish peroxidase-conjugated anti-goat (sc-2020, Santa Cruz Biotechnology, CA).

Constructs

Mammalian Foxa1 expression vector was a generous gift of K. Zaret (Brown University, Providence, RI). pGL2-basic was obtained from Promega, CA. The promoter region of p27-1609 to +178 was generated by PCR and subcloned into pGL2-basic aspreviously described.17 The ERα expression vector (HEGO) was a generous gift of P. Chambon (University of Strasbourg, France). The ERE-luciferase reporter construct, kindly provided by D. Harris, (UCLA, CA), consists of 2 repeats of the upstream region of the vitellogenin ERE promoter from −331 to −289, followed by region −109 to +45 upstream of the thymidylate kinase, followed by the luciferase gene. Generation of pCDNA3.1-FOXA1 and pCDMA3.1-FOXA1 constructs: full length or parts of mammalian FOXA1 cDNA were subcloned into EcoRI and NotI sites of pCDNA 3.1 (Invitrogen, CA) using the primers:

• pCDNA-FOXA1 (bases 1–1401)-3′primer GTGGAATTCTGATGTTAGGGACTGTG,

• 5′ primer TCGAGCGGCCGCGGAAGTATTTAGCAC.

• pCDNA-FOXA1 Forkhead (FH) (bases 421–885)-3′ primer GTGGAATTCTGATGGCGTACGC

• GCCG; 5′ primer TCGAGCGGCCGCGGAAGTATTTAGCAC.

• pCDNA-FOXA1 C-terminus (bases 421–1401)-3′ primer GTGGAATTCTGATGGCGTACGC

• GCCG; 5′ primer CGAGCGGCCGCGGAAGTATTTAGCAC.

• pCDNA-FOXA1 N-terminus (bases 1–885)-3′primer GTGGAATTCTGATGTTAGGGACT

• GTG; 5′ primer TCGAGCGGCCGCGGAAGTATTTAGCAC.

• pCDNA-FOXA1 small C-terminus (bases 885–1401)-3′ primer GTGGAATTCTGAAGGACC

• CCTCA; 5′ primer TCGAGCGGCCGCGGAAGTATTTAGCAC.

Generation of GFP-FOXA1

Mammalian FOXA1 cDNA was cloned into EcoR1 sites downstream to the GFP gene of pEGFP-C1 (Clontech, CA), thus forming a GFP-FOXA1 construct in which the N-terminus of FOXA1 was modified. The primers used include 5′ primer GCTTCGAATTCTATGTTAGGGACTGTG; 3′ primer TCAGAGAATTCAGCTAGGAAGTATTTAG. All the subcloned constructs were sequenced.

Generation of FOXA1 directed siRNA

FOXA1-directed siRNA were planned and generated using the RNAi oligo retriever web site (http://katahdin.cshl.org:9331/RNAi) and the “shagging PCR protocol”. Three siRNAs primers were designed, each directed against a 29 base pair sequence in the human FOXA1 and U6 promoter reverse primer sequence. All were found to be specific for FOXA1 by BLAST search and their full sequences were:

• 1
  Against 696–725: 5′AAAAAAGAGCCCTTGCCCGGCCTGCCCAGGAAGCGCAAGCTTCCGCTCCCCGGACAAGCCGGGCAAGGGCTCGGTGTTTCGTCCTTTCCACAA
• 2
  Against 1023–1052: 5′AAAAAAGTCCTCAACCCCGAGACGCCCCCTGTCACC AAGCTTCGCGACAGGGGGCGCCTCGGAGTTGAAGACGGTGTTTCGTCCTTTCCACAA
• 3
  Against 1376–1405: 5′AAAAAAAGACGGATCCGGAATACACACCTTAGTAGC AAGCTTCCTACCAAGGTGTGTATTCCAGACCCGTCCGGTGTTTCGTCCTTTCCACAA

The primers were used together with a SP6 primer to clone the U6 promoter, generating a PCR product containing both the U6 promoter and the FOXA1-directed siRNA sequence. The PCR product was inserted into the pCR2.1 vector using a TA cloning kit (Invitrogen, CA). A scrambled siRNA was designed by the same method and was used as a control.

Tissue arrays and immunohistochemistry analysis

Breast cancer tissue arrays were created, after IRB approval, from cancers diagnosed at Cedars-Sinai Medical Center from 1991 to 1998. Clinical parameters and follow-up information were obtained from hospital records. The tissue arrays were constructed using a manual arrayer (Beecher Instruments, MD) and contained three 1-mm samples for each specimen. For immunohistochemistry analysis, 5 μm sections were cut from the tissue array. For immunohistochemistry staining, the slides were deparaffinized through xylenes and graded ethyl alcohols and then rinsed in water, followed by quenching of endogenous peroxidase activity by a 30% solution of hydrogen peroxidase in methanol for 10 min. Antigen retrieval was performed by boiling the slides in 0.01 mol/l sodium citrate buffer pH 6.0 in a microwave oven at maximum power for 1 min and at 20% power for 9 min, followed by a 20 min cool down and rinses in wash buffer. The slides were then incubated for 1 hr with the anti-FOXA1 antibody (1:500 dilution), after blocking with normal serum for 30 min. The slides were reacted with the secondary antibody for 30 min; signal amplification and chromogen development were conducted for 30 min each. The stained slides were counterstained with hematoxylin and mounted. The specificity of staining for the antibody was determined by staining cell blocks prepared from HCT-116 cells transfected with FOXA1 expressing vector and each run included appropriate positive and negative control slides. Positive staining was noted only in the nuclei of normal and tumor cells. Staining was scored by percent of positive tumor cells and staining intensity. On the basis of the combined score, negative and low expressions were categorized low, while intermediate and high expressions were categorized as high.

Cells and transfections

All cell lines were obtained from the ATCC. The breast cancer cell lines used: MCF-7 and BT-474 cells were grown in DMEM medium containing 10% FCS; MDA-MB-231 and SK-BR-3 were grown in RPMI medium containing 10% FCS. The selected cell lines have different phenotypes and different expression patterns of the ER, HER2 and p53, and thus serve as models for various subclasses of breast cancer.20 The HCT-116 colon cancer cell line was grown in medium containing 10% FCS.

All transfections used Lipofectamine 2000 (Invitrogen, CA) according to the manufacturer's instructions. For E2 and 4HT studies, cells were first cultured in the appropriate media using 10% charcoal-treated serum for 3 days and then treated with the appropriate agent.

Luciferase assays

Cells were plated in 12-well plates and transfected with the reporter vector and the various constructs. Transfection efficiency was normalized using pRL-SV40 at 1/20 of the total DNA concentration. Luciferase assay was conducted according to the manufacturer's instructions (Promega, CA). For E2 and 4HT studies, cells were cultured in the appropriate media with 10% charcoal-treated serum for 2 days and then treated with the appropriate agent.

Real-time quantitative PCR

At 2 days following transfection with the various constructs, total RNA was prepared from breast cancer cell lines, using the RNeasy kit (Qiagen, CA), according to the manufacturer's instructions. A total RNA (2.5 μg) was reverse transcribed using random hexamers and Superscript II (Invitrogen, CA). The cDNA was then used for real-time quantitative PCR using iCycler (Bio-Rad, CA) according to the manufacturer's protocol. p27-specific primers were 5′ GTGGACCCAAACACTGATCC and 5′ AGAAGAATCGTCGGTTGCAG. Equal loading was determined using 18S specific primers.

Western blot analysis

Cells were lysed for total protein extraction in a buffer containing 50 mM TrisCl pH 7.4, 150 mM NaCl and 2% NP-40 together with a protease inhibitors cocktail (Roche Diagnostics). Protein lysates were separated on polyacrylamide gels and immunoblotted with the various antibodies as indicated. Immunoactivity was detected with horseradish peroxidase-conjugated secondary antibody and visualized by Enhanced Chemiluminescence (Pierce, IL). Quantification of the results was performed using AlphaImager 2000 (Alpha Innotech, CA).

Localization studies

The pEGFP-C1 or GFP-FOXA1 plasmids were transfected into the cells. Cells were visualized by inverted fluorescence microscope (Zeiss Axioscop Microscope).

Colony assays

For colony assays, cells were plated in 6-well plates and transfected with either FOX-pCDNA3.1 construct or pCDNA3.1. Two days after transfection, G418 (400 μg/ml) was added to the culture media. G418 containing media was replaced every 3 days. The cells were fixed and stained at day 14 using Gentian Violet. Untransfected cells died within the 2 weeks of culture in the selection media (data not shown). Quantification of the results was performed using AlphaImager 2000 (Alpha Innotech, CA).

Cell cycle assays

Cells were transfected with either FOXA1-directed siRNA constructs or scrambled siRNA construct. One or 2 days after transfection, the cells were harvested, fixed in methanol and stained with propidium iodide (Abcam, MA). Flow cytometry was performed at the Flow Cytometry Core facility of Cedars-Sinai Medical Center.

Statistical analysis

Results for continuous variables were presented as mean ± standard deviation. Results for categorical variables were presented as number (%). Two-group differences in continuous variables were assessed by the t test, using the Satterthwaite adjustment to degrees of freedom when there was evidence against the homogeneity of variance assumption. Two-group differences in categorical variables were assessed by the χ² and Fisher exact tests. Tests for linear trends in proportions were assessed by the Cochran-Armitage trend test. Stepwise logistic regression was used to assess potential predictors of high FOXA1. All significance tests were 2-tailed. All calculations and statistical tests were performed using the software package SAS version 9.1 (SAS Institute, Cary, NC).

Results

Transcriptional activation of p27 by FOXA1

We have recently identified a FOXA1 binding site on the p27 promoter and shown activation of the promoter by FOXA1 in the HCT-116 colon cancer cells.17 To elucidate the transcriptional regulation of the p27 promoter by FOXA1 in breast cancer, transient transfection reporter assays were conducted in the MCF-7 and MDA-MB-231 cells. The cells were transfected with a p27-luciferase reporter construct and either a FOXA1 expression vector or control vector, and treated with either E2 or 4HT. Exposure of MCF-7 cells to E2 resulted in 36% (p < 0.05) reduction and 4HT treatment resulted in mild elevation of p27 promoter activity (Fig. 1a). Forced expression of FOXA1 resulted in about 2.5-fold increase in p27-luciferase activity, irrespective of either E2 or 4-HT treatment (p < 0.05 for control and E2 stimulation groups). We confirmed that FOXA1 did not activate a control heterologous promoter (Fig. 1b). Similar experiments were performed in MDA-MB-231 cells and revealed very low basal p27-luciferase activity, and no significant elevation following FOXA1 transfection (data not shown).

Figure 1. Transcriptional regulation of p27 by FOXA1. (a) MCF-7 cells were transiently transfected with either FOXA1 or the control vector, together with the p27 promoter-luciferase and renilla-luciferase reporters. Cells were serum-starved for 48 hr followed by treatment with either E2 (10 nM) or 4HT (1 nM) for 24 hr. The resulting luciferase activities were analyzed and normalized to renilla activities and are shown relative to the control vector. (b) MCF-7 cells were transiently transfected with either FOXA1 or the control vector, together with either the p27 promoter-luciferase or pGL2-basic and with renilla-luciferase reporters and analyzed for luciferase activity. (c) HCT-116 cells were transfected with the FOXA1 construct with either FOXA1-directed siRNA constructs or a scrambled siRNA construct (siRNAc), as indicated. Forty-eight hours after transfection, cells were lysed and immunoblotted with antibodies to FOXA1 or to GAPDH. Quantification of FOXA1 expression relative to GAPDH expression is presented. (d) MCF-7 cells were transiently transfected with either FOXA1-directed siRNA or scrambled siRNA, together with p27 promoter-luciferase and the renilla-luciferase reporters and analyzed for luciferase activities. (e) Schematic representation of the FOXA1 constructs used. (f) MCF-7 cells were transiently transfected with either FOXA1 constructs or the control vector, as indicated, together with the p27 promoter-luciferase and renilla-luciferase reporters, and analyzed for luciferase activities. (g) MCF-7 or MDA-MB-231 cells were transiently transfected with either FOXA1 constructs or a control vector. p27 mRNA levels were determined 48 hr after transfection by quantitative real time-PCR. All data are shown as mean ± S.D. of at least 3 independent experiments, each performed in triplicate. Asterisk indicates p < 0.05.

Three FOXA1 directed siRNA constructs were designed and transfected into HCT-116 cells, which normally do not express FOXA1, together with FOXA1 expression vector. The three constructs, siRNA1, siRNA2 and siRNA3, decreased FOXA1 levels by 90, 40 and 80%, respectively, compared to a scrambled siRNA (siRNAc), which had no effect on FOXA1 levels (Fig. 1c). siRNA1 was used in all further studies. Silencing of FOXA1 in MCF-7 cells resulted in a 45% inhibition of p27 promoter activity (p = 0.002, Fig. 1d).

To elucidate the role of FOXA1 domains in its transcriptional activities, 3 FOXA1 constructs were designed (Fig. 1e): (i) NT: contains the N-terminal part and FH region; (ii) FH: contains the FH motif; (iii) CT: contains the C-terminal part and the FH region. The various constructs were transfected into MCF-7 cells together with the p27 reporter construct (Fig. 1f). Full length FOXA1 elevated the p27 transcriptional activity by 5-fold (p = 0.04), while the NT construct enhanced the activity by 10-fold (p = 0.005 for either NT vs. control or NT vs. FOXA1). The FH and CT constructs also enhanced the transcriptional activity, though to a lesser degree than the full length FOXA1.

To elucidate the transcriptional activity of FOXA1 on endogenous p27, p27 mRNA levels were measured in MCF-7 and MDA-MB-231 cells following transfection with either a control vector, FOXA1 or the NT construct. The p27 mRNA levels increased in MCF-7 cells but not in MDA-MB-231 cells, following transfection with FOXA1 and the NT construct (p = 0.05 for either FOXA1 or NT vs. control. Fig. 1g).

Differential transcriptional regulation of the ERE by FOXA1

The role of FOXA1 in the regulation of ER activity was studied using the ERE-luciferase reporter construct, which consists of 2 repeats of the upstream region of the vitellogenin ERE promoter, followed by the luciferase gene. This construct has been used extensively for the study of the ER pathway.21 MCF-7 cells were cotransfected with the ERE-luciferase reporter together with the various FOXA1 constructs or FOXA1 directed siRNA, and treated with either E2 or 4HT. Overexpression of FOXA1 decreased ERE activity after E2 stimulation by 65% (p = 0.03, Fig. 2a) while silencing of FOXA1 resulted in the opposite effect: a 5.5-fold enhancement of the ERE activity (p < 0.0007, Fig. 2b). Both the FH and the N-terminal part, but not the C-terminal part, were important for ERE inhibition (Fig. 2c). Similar results were observed for the BT-474 cells (ERα-positive cells, data not shown).

Figure 2. Transcriptional regulation of the ERE by FOXA1. (a) MCF-7 cells were transiently transfected with either FOXA1 or control vector, together with the ERE-luciferase and the renilla-luciferase reporters. Cells were serum-starved for 48 hr followed by treatment with E2 (10 nM) or 4HT (1 nM) as indicated, for 24 hr. The luciferase activities were analyzed and normalized to renilla activities, and are shown relative to the control vector. (b) MCF-7 cells were transiently transfected with either FOXA1-directed siRNA or scrambled siRNA, together with ERE-luciferase and renilla-luciferase reporters and analyzed for luciferase activities. (c) MCF-7 cells were transiently transfected with either a FOXA1 construct or the control vector, together with the ERE-luciferase and renilla-luciferase reporters. Luciferase activities were analyzed 48 hr after transfection. (d) MDA-MB-231 cells were transiently transfected with either a FOXA1 construct or the control vector, together with the ERE-luciferase and renilla-luciferase reporters. Luciferase activities were analyzed 48 hr after transfection. (e) MDA-MB 231 cells were transiently transfected with either FOXA1-directed siRNA or scrambled siRNA, together with the ERE-luciferase and renilla-luciferase reporters. Luciferase activities were analyzed 48 hr after transfection. (f) MDA-MB-231 cells were transiently transfected with either the FOXA1 construct, ER expressing vector or both as indicated, together with the ERE-luciferase and renilla-luciferase reporters. Luciferase activities were analyzed 48 hr after transfection. Data are shown as mean ± S.D., from at least 3 independent experiments, each performed in triplicate. Asterisk indicates p < 0.05 for comparison between cells transfected with either the FOXA1 construct or vector control. Double asterisk indicates p < 0.05 for the comparison between cells transfected with FOXA1 and FOXA1+ER.

Transfection of the ERα-negative MDA-MB-231 cells22 with FOXA1 resulted in a 4.1-fold activation of the ERE (p = 0.01, Fig. 2d), while silencing of it inhibited ERE activity by 48% (Fig. 2e). FOXA1 activity in MDA-MB-231 cells was also dependent on the FH and the N-terminal parts of the protein (Fig. 2d).

MDA-MB-231 cells were transiently transfected with the ERE reporter together with either FOXA1 alone or in combination with ERα and the luciferase activity was measured (Fig. 2f). Transfection with ERα resulted in an 8.6-fold elevation in ERE activity and the addition of FOXA1 resulted in a 15.3-fold increase in ERE activity. Thus, coexpression of ERα together with FOXA1 resulted in synergistic activation of the ERE. These results suggest that ERα itself is insufficient to induce the inhibitory effect of FOXA1 on ERE activity.

Regulation of FOXA1 levels by growth stimuli

MCF-7 cells, which had been cultured for 72 hr in estrogen-free media, were treated with either E2 (10 nM) alone or in combination with 4HT (1 nM) for 24 hr. Exposure to estrogen resulted in a 45% decrease in FOXA1 levels, which was only partly reversed by the addition of 4-HT (Fig. 3a).

Figure 3. Regulation of FOXA1 levels by growth stimuli. (a) MCF-7 cells were stimulated with either E2 (10 nM) alone or in combination with 4HT (1 nM) for 24 hr, lysed and immunoblotted with antibodies to FOXA1 or GAPDH. (b) MDA-MB-231 cells were transiently transfected with the ERα expressing vector or control vector as indicated, lysed and immunoblotted with antibodies to FOXA1 or to GAPDH. The quantification of FOXA1 expression relative to GAPDH expression represents a mean ± S.D. of 3 independent experiments.

Transfection of ERα into MDA-MB-231 cells activates the estrogen pathway but inhibits their growth.23 We found that transfection of the ERα into MDA-MB-231 cells resulted in upregulation of FOXA1 protein levels (Fig. 3b). Higher expression of FOXA1 was also noted when MDA-MB-231 cells were grown in high confluence or starvation (data not shown). Thus, FOXA1 expression inversely correlated with growth stimulation but did not correlate with the activation of ER pathway.

Overexpression of FOXA1 inhibits proliferation of breast cancer cell lines

The effect of FOXA1 on the growth rate of breast cancer cell lines was assessed using colony formation assays. The cells were transfected with either a FOXA1 expression vector (pcDNA3.1-FOXA1) or an empty vector (pcDNA3.1) as a control. Transfected cells were cultured in selection media for 2 weeks and then stained to determine the number of surviving colonies. FOXA1 expression significantly reduced the number and size of surviving colonies of all cell lines compared to empty vector-transfected controls (Fig. 4).

Figure 4. FOXA1 inhibits the clonal proliferation of breast cancer cell lines. Breast cancer cell lines were transfected with either pCDNA3.1 or pCDNA3.1-FOXA1 expression vector. Transfected cells were treated for 2 weeks with G418, fixed, stained and photographed. The quantification represents a mean ± S.D. of 3 independent experiments. Asterisk indicates p < 0.05.

FOXA1 expression in breast tissue arrays

Immunohistochemistry analysis of FOXA1 expression in breast tissues was conducted on tissue arrays of 100 infiltrating ductal carcinoma (IDC) samples and 88 pure ductal carcinoma in situ (DCIS) samples. The analysis revealed a direct correlation between FOXA1 and tumor differentiation: higher levels in pure DCIS (94%) than in DCIS adjacent to IDC (85%) or ER-positive (83%) and ER-negative IDC (58%, Table I, Fig. 5). Using linear trend analysis, we found this correlation to be highly significant (p = 0.0004). The expression of FOXA1 was high in normal tissues adjacent to breast cancer samples, as well as in various benign changes, such as atypical ductal and lobular hyperplasia, but low in 17 samples of normal breast. In all the samples, positive staining of FOXA1 was noted only in the nuclei.

Figure 5. FOXA1 expression in tissue array samples: immunohistochemical staining of FOXA1 protein in breast tissue arrays shown at ×40 (Left panels) and ×100 magnifications (Right panels). (a), (b) DCIS showing high expression of FOXA1. (c), (d) ER positive IDC showing high expression of FOXA1. (e), (f) ER-negative IDC showing low expression of FOXA1.

Table I. FOXA1 Expression in Ductal Carcinoma in situ and Invasive Ductal Carcinoma

	Pure DCIS (N = 88)	DCIS adjacent to IDC (N = 78)	IDC, ER-positive (N = 69)	IDC, ER-negative (N = 19)
DCIS, ductal carcinoma in situ; IDC, invasive ductal carcinoma; ER, estrogen receptor. p < 0.05 for comparisons of DCIS vs. IDC ER-positive and IDC ER-negative, and for IDC ER-positive vs. IDC ER-negative. p = 0.0004 for trend of reduction in FOXA1 expression from pure DCIS, DCIS adjacent to IDC, IDC ER-positive and IDC ER-negative.
Low FOXA1 (N, %)	5 (6)	11 (14)	12 (17)	8 (42)
High FOXA1 (N, %)	83 (94)	67 (86)	57 (83)	11 (58)

Comparison of 77 IDC samples that expressed high levels of FOXA1 to 23 IDC samples that expressed low FOXA1 levels, revealed that high FOXA1 expression was associated with favorable prognostic factors such as lower tumor grade, lower lympho-vascular invasion rate, expression of the ER and higher levels of Bcl-2 (Table II). These results were also verified by using univariate analysis (data not shown). Stepwise logistic regression identified lympho-vascular invasion as a significant predictor of low FOXA1 expression (estimate 0.128, confidence interval (CI) 0.03–0.48), while the association between ER and FOXA1 expression was of only borderline statistical significance (estimate 3.4, CI 0.96–12.0).

Table II. Distribution of Clinicopathological Variables in IDC Patients by FOXA1 Levels

	Low FOXA1, N (%)1	High FOXA1, N (%)2	Overall p value
IDC, invasive ductal carcinoma; ER, estrogen receptor; PR, progesterone receptor. 1 N = 23. 2 N = 77.
Age (mean ±SD)	53 ± 12	56 ± 15	0.48
Tumor size (T)
1	12 (52)	51 (66)	0.31
2	8 (35)	22 (29)
3	2 (9)	2 (3)
Unknown	1 (4)	2 (3)
Lymph nodes involvement (N)
Negative	11 (48)	48 (62)	0.17
Positive	9 (39)	18 (23)
Unknown	3 (13)	11 (14)
Histology IDC
IDC	21 (91)	75 (97)	0.54
Mixed	1 (4)	2 (3)
Grade
1–2	9 (39)	50 (65)	0.04
3	13 (57)	25 (33)
Unknown	1 (4)	2 (3)
Lympho-vascular invasion
Negative	6 (26)	47 (61)	0.004
Positive	17 (74)	30 (39)
ER
Negative	8 (35)	11 (14)	0.03
Positive	12 (52)	57 (74)
Unknown	3 (13)	9 (12)
PR
Negative	7 (30)	15 (20)	0.25
Positive	13 (57)	53 (69)
Unknown	3 (13)	9 (12)
KI67
1	8 (35)	42 (55)	0.09 (1 vs. 2.3)
2	6 (26)	4 (5)
3	3 (13)	14 (18)
Unknown	6 (26)	17 (22)
BCL2
0	2 (9)	0 (0)	0.02 (0.1 vs. 2.3)
1	11 (48)	19 (25)
2	7 (30)	32 (42)
3	3 (13)	17 (22)
Unknown	0 (0)	9 (12)

Discussion

In the present study, we identify, for the first time, growth inhibitory activities of FOXA1 in breast cancer and discover an association between FOXA1 expression and favorable prognostic factors. We identified 2 mechanisms that may be involved in the growth inhibitory role of FOXA1: activation of p27 transcription and differential regulation of the ER pathway.

p27 is a cyclin-dependent kinase inhibitor, which, like FOXA1, is expressed in most breast cancers and correlates with favorable outcome and ER expression.24, 25, 26 Although p27 levels are primarily controlled at the posttranscriptional level, transcriptional regulation of its expression has been observed in breast cancer cells.24, 25, 27 We have recently shown a direct activation of p27 transcription by FOXA1 in the HCT-116 colon cancer cell line,17 and show here similar role for FOXA1 in breast cancer cells. These findings may explain the observation of high levels of p27 in cell lines that express high levels of ER and FOXA1, such as MCF-7.25, 26

We noticed cell-type dependent regulation of the ER pathway, as measured by ERE-luciferase activity, by FOXA1. FOXA1 inhibited the ER pathway in ERα-positive cells (MCF-7 and BT-474), and slowed cells' growth. On the other hand, in ERα-negative MDA-MB-231 cells with forced ERα expression, FOXA1 activated the ER pathway, leading as well to slowed growth.23 These findings thus further support a growth inhibitory role for FOXA1 in breast cancer. Cell-type dependent transcriptional activity of FOXA proteins is not unique for breast cancer and has also been demonstrated in hepatoma and neuroblastoma cells.28, 29 Interestingly, another member of the FOX family, FOXO1 (also known as FKHR) also shows cell-type dependent activity on the ER pathway: repression in HepG2 cells and activation in MCF-7 cells.30, 31

Two global screens for ERα-responsive genes identified a role for FOXA1 in the activation of genes that contain both ERE and FOXA1 consensus sites in their promoter region.18, 19 However, these genes account for only 12% of all the ERα-responsive genes.18 We used for our studies a well-validated ERE-luciferase construct, which does not contain FOXA1 consensus site.21 Our results may thus represent the role of FOXA1 in the regulation of the subset of ERα-responsive genes that do not contain FOXA1 consensus sites in their promoter region. Coexpression of FOXA1 and ERα in MDA-MB-231 cells resulted in a synergistic activation of the ERE. This observation suggests that genes other than ERα are involved in the FOXA1 transcriptional inhibition of the ER pathway. Indeed, extensive coimmunoprecipitation assays failed to demonstrate direct interaction between FOXA1 and the ERα in various conditions and cell lines (data not shown).

Although recent studies emphasized the role of FOXA1 as an activator of the ER,18, 19, 32 it is important to note that most of the known FOXA1-responsive genes are not ERα-responsive genes6 and some of them are cell cycle inhibitors, such as p21 and IGFBP-1.6, 33 The role of FOXA1 in the induction of these genes in breast cancer remains to be elucidated.

In support of a growth inhibitory role of FOXA1, we observed an inverse correlation between FOXA1 levels and various growth stimuli. Similarly, 2 recent microarray analyses of genes regulated by E2 in MCF-7 cells also identified downregulation of FOXA1 levels, as soon as 4 hr following E2 stimulation.15, 16 Moreover, retinoic acid and BRCA1, which have growth inhibitory effects on breast cancer cells, can upregulate FOXA1 levels.17, 34, 35 Of note, Laganiere et al. observed upregulation of FOXA1 4–8 hr following E2 stimulation in MCF-7 cells.18 The discrepancy between the studies may be explained by the use of different antibodies and different study conditions.

Two transcriptional regulatory domains have been identified in FOXA proteins, one in the C-terminal region and the other in the N-terminus. Truncation of these domains in FOXA2 reduced its transcriptional activity in hepatoma cells, and deletion of the C-terminal domain of FOXA1 inhibited its transcriptional activity in prostate cancer cell lines.4, 5, 10In vitro experiments suggest that the DNA binding domain of FOXA proteins, which share similarity with the linker histones H1 and H5, can regulate transcription by opening compacted chromatin.3, 36 However, in our system, deletion of the C-terminal domain enhanced rather than inhibited FOXA1 transcriptional activity. This finding may result from different regulation of FOXA1 activity in breast cancer cells compared to hepatoma and prostate cells.

Immunohistochemical analysis of FOXA1 expression in a large cohort of IDC and DCIS samples revealed high expression in the majority of tumors and, in accordance with several global gene analyses, was associated with higher differentiation and ERα expression.12, 13, 32, 37 However, 42% of the ERα-negative tumors showed high, and 17% of the ERα-positive tumors showed low FOXA1 expression; and FOXA1 levels were inversely correlated with lympho-vascular invasion, even after adjustment for ERα status. These data suggest that at least some of FOXA1 activities in breast cancer are ERα-independent. Indeed, overexpression of FOXA1 inhibited growth not only of the ER-positive MCF-7 cells but also of the ER-negative MDA-MB-231 and SK-BR-3 cells. These findings are supported by a recent study in which high expression of FOXA1 was noted in a subset of ER-negative tumors.14

In summary, we have found a growth inhibitory role of FOXA1 in breast cancer cells and identified an association between FOXA1 protein levels and favorable prognostic factors. FOXA1 may thus be a potential novel prognostic marker, and specific activation of it may have a potential therapeutic role.

Acknowledgements

We thank Dr. K. Zaret (Brown University, Providence, R.I.) for the FOXA1 expression vector; Dr. P. Chambon (University of Strasbourg) for the ER alpha expression vector, Dr. W. Jiang (Salk Institute) for the GFP-ras expression vector and Dr. D. Harris (UCLA) for the ERE-luc expression vector. H.P.K. is a member of the Molecular Biology Institute and Jonsson Comprehensive Cancer Center at UCLA, and holds the endowed Mark Goodson Chair of Oncology Research at Cedars-Sinai Medical Center/UCLA School of Medicine. I.W. is the Mary Barry Medical Bridges Foundation Fellow.