Introduction
Breast cancer is a heterogenous disease whose progression from atypical ductal hyperplasia (ADH) to ductal carcinoma
in situ (DCIS) and invasive ductal carcinoma (IDC) is regulated by the aberrant expression of multiple mediators produced by the mammary tumor itself and the adjacent reactive stroma [
1]. These signals promote tumor cell proliferation, survival, establishment of tumor vasculature, invasion and ultimately metastasis to secondary organs. The ability of the tumor to create a state of local immune suppression allows tumor cells to evade clearance by the immune system [
2]. Signaling pathways that regulate cytokine/chemokine expression (ILs, IFNs and interferon regulatory factors (IRFs)) have recently been found within tumor microenvironments and in metastatic sites; some of these cytokines stimulate while others inhibit breast cancer proliferation and/or invasion [
2]. The role of these cytokines in disease progression, as markers of disease stage, and as novel treatment strategies requires further attention.
IRF5 is a transcription factor that regulates type I IFN signaling [
3] and cytokines/chemokines with lymphocyte-chemotactic activities, that is, RANTES, MIP1α/β, MCP1, I309, IL8 and IP10 [
4]. Subsequent studies demonstrated its critical role(s) in the cellular response to extracellular stressors including virus, DNA damage, Toll-like receptor (TLR) and death receptor signaling [
3‐
11]. Depending on the cell type, loss of
IRF5 yields cells incapable of a sufficient immune response to pathogens and/or undergoing apoptosis [
6,
8‐
11]. Northern blot analysis of
IRF5 tissue-specific expression revealed that it is primarily expressed in lymphoid tissues but can be induced in multiple cell types [
3,
12,
13]. IRF5 has been associated with the regulation of important cellular processes, such as cell growth, apoptosis, cell cycle arrest, and cytokine production [
6‐
9,
14].
Little is known of IRF5 tumor suppressor function.
IRF5 was mapped to chromosome 7q32 [
3] that contains a cluster of imprinted genes and/or known chromosomal aberrations and deletions in lymphoid, prostate, and breast cancer [
15‐
22].
IRF5 expression is absent or significantly decreased in immortalized tumor cell lines and primary samples from patients with hematological malignancies, suggesting for the first time its role as a tumor suppressor gene [
3,
7]. Recent data from
irf5
-/-
mice support its candidacy as a tumor suppressor gene [
9]. Mouse embryonic fibroblasts (MEFs) from
irf5
-/-
mice are resistant to DNA damage-induced apoptosis and can be transformed by
c-Ha-ras [
9]. Conversely, ectopic expression suppresses malignancy of cancer cell lines
in vitro and
in vivo [
7,
23]. While IRF5 has been shown to be a direct target of p53 [
23], data from our lab and others indicate that IRF5 acts on an apoptotic signaling pathway that is distinct from p53 [
7‐
9].
Loss of tumor suppressor genes represents a critical event in the development and progression of breast cancer. However, while an increasing number of oncogenes have been identified in breast cancer, few tumor suppressor genes have been directly implicated in the development/progression of this disease. Altered expression or function of tumor suppressor genes
BRCA1,
BRCA2 and
p53 do not fully account for the high prevalence of spontaneous breast cancers. Loss or mutation of
BRCA1 occurs in < 10% of all breast cancers, while
p53 is mutated in up to 30% of breast cancers [
24]. There are likely other tumor suppressor genes and oncogenes contributing to breast tumorigenesis. IRF1 was recently shown to have tumor suppressor function in breast cancer, while increased expression of IRF2 was associated with oncogenic activation [
25]. Overexpression of IRF1 induced apoptosis and inhibited tumor growth in mouse and human mammary cancer cells [
26‐
28]. The focus of the present study was to examine and compare IRF1 and IRF5 expression in human breast tissue and to determine whether IRF5 acts as a tumor suppressor. Data presented here support a unique role for IRF5 in regulating mammary epithelial cell growth and provide the first direct evidence that loss of IRF5 tumor suppressor function contributes to breast tumorigenesis.
Materials and methods
Cell lines and culture
Human immortalized breast cells MCF-12A, MCF-7, MDA-MB-231, -436, -468, and T47D were purchased from American Type Culture Collection (Manassas, VA, USA) in spring 2009, and aliquots were frozen in liquid nitrogen until time of use. Cells were cytogenetically tested and authenticated (by STR profiling from ATCC) before freezing. The amphotrophic helper-free Phoenix cells were provided by G. Nolan (Stanford, CA, USA). All breast cancer cells lines and 293T-derived Phoenix cells were propagated in Dulbecco's modified Eagle's medium (Sigma-Aldrich, St. Louis, MO, USA) containing 10% fetal bovine serum (Sigma) and 1 IU penicillin/1 μl/ml streptomycin (Mediatech, Hemdon, VA, USA) at 37°C in a humidified incubator with 5% CO2/95% air. MCF-12A were grown in DMEM F-12 supplemented with 5% horse serum (Sigma), 100 ng/ml cholera toxin (Sigma), 20 ng/ml EGF (Invitrogen, Carlsbad, CA, USA), 10 μg/ml insulin (Sigma), and 500 mg/ml hydrocortisone (Sigma). Each vial of frozen ATCC authenticated cells was thawed and maintained in culture for a maximum of six weeks. There were enough frozen vials for each cell line to ensure that all experiments were performed on cells that had been tested and in culture for six or more weeks.
Chemicals and treatments
Doxorubicin was from Sigma; Interferon (IFN)-γ from PBL InterferonSource (Piscataway, NJ, USA). Cells were treated with 0.1 or 1 μM Doxorubicin or 1,000 U/ml IFN-γ for the indicated time periods. Cells were exposed to 2, 5 or 10 Gray (Gy) of ionizing radiation (IR) using a self-shielded Cs-137 irradiator.
Retroviral construction and transduction
IRF5 was cloned into the pBabe-puromycin vector at BamHI/SalI sites transfected to Phoenix cells as described [
29]. Viral supernatants were collected 48 h post-transfection and used to infect MCF-7, MDA-MB-231 and -468 cells. After two days, media was exchanged for puromycin selection to obtain stable transfectants. Cultures were pooled from each cell line and positive infection determined by Western blot with mouse anti-IRF5 antibodies (M01, Novus Biologicals, Littleton, CO, USA).
Immunofluorescence (IF), immunohistochemistry (IHC) and semi-quantitative evaluation
H&E sections of formalin-fixed paraffin-embedded (FFPE) archival tissue specimens were reviewed by two pathologists (MH and NM) for histological evaluation of disease and grade. Slides from 19 patients with ADH, 24 with DCIS, 29 with IDC, and 11 with lymph node metastases were evaluated for IRF expression. Normal breast tissue from the same donors or adjacent to tumors were characterized in 51 patients. Sections were obtained from the Pathology Department at UMDNJ New Jersey Medical School (NJMS). The study was approved by the NJMS Institutional Review Board (IRB) and all participants provided written informed consent. Antigen retrieval was performed by heating slides at 95°C in citrate buffer (pH 6.0) for one hour before staining with mouse anti-IRF5 or rabbit anti-IRF1 (C-20, Santa Cruz Biotechnology, Santa Cruz, CA, USA) antibodies at 1:100 dilution in 4% BSA overnight. For IF, slides were incubated with anti-rabbit Cy3 and anti-mouse FITC (Molecular Probe, Eugene, OR, USA) antibodies at 1:1,000 in 4% BSA/PBST. Slides were mounted with DAPI mounting buffer (Vector Laboratories, Burlingame, CA, USA) and images captured on a Zeiss Axiovert 200 fluorescent microscope; quantification was performed using Axiovision software (Carl Zeiss Microimaging, Oberkochen, Germany). For IHC, slides were incubated with 1:200 diluted anti-IRF5 for two hours then 1:1,000 diluted Alkaline Phosphatase anti-mouse IgG (Vector Laboratories, AP-2000) and developed with the Vector® Blue Alkaline Phosphatase (BAP) Substrate Kit III (Cat. No. SK-5300). The second staining was with 1:200 diluted anti-IRF1, Peroxidase anti-Rabbit IgG (Vector Laboratories, PI-1000) and developed with DAB Substrate Kit (Vector Laboratories, SK-4100). The nucleus was stained with Nuclear Fast Red mounting buffer.
Evaluation of stained slides was assessed by one pathologist (MH) and two independent reviewers (XB and BJB or JA), who were unaware of the patient's characteristics. Two slides from different areas of the same tumor were examined and scored independently by each reviewer with a consensus being reached in difficult cases (< 5% for each antibody). Following initial review, an arbitrary grading system was defined for each antibody in which the density of positive cells within normal ducts and lobules or ADH, DCIS and IDC as defined by the tumor (and not the stroma) was assessed semi-quantitatively on the whole tissue section. This classification allowed the stratification of the tumors as positive or negative for IRF1 and IRF5.
Western blotting
Preparation of cellular lysates and immunoblotting were performed as described [
30,
31]. Proteins were transferred to nitrocellulose membrane and detected with horseradish peroxidase (HRP)-conjugated secondary antibody (1:2,000) followed by enhanced chemiluminescence (Amersham Biosciences, Piscataway, NJ, USA). Equal loading was confirmed with β-actin antibodies (Cell Signaling, Danvers, MA, USA) after stripping with Restore™ Western blot stripping buffer (Pierce, Rockford, IL, USA).
Colony survival assay
Colony survival was performed as described [
31]. Cells were plated and exposed to different sources of DNA damage. One hour post-treatment, cells were split into 2,000 cells per 10 cm plate and cell growth assessed after 14 days by staining with 0.5% crystal violet and 25% methanol. The colony number was calculated and plotted as the mean for triplicate samples and presented as percentages relative to the control.
Apoptosis assay
Apoptosis was assessed by flow cytometric analysis of cells stained with Annexin V-FITC and PI using a Becton Dickinson FACScan (St. Louis, MO, USA) [
8,
10]. Data analysis using CELLQuest™ software (Becton Dickinson) was performed; numbers of cells positive for Annexin V-FITC, PI, or combinations thereof, were calculated.
Suppression of IRF5 with siRNA
A modified protocol from Hu
et al. [
10] was used to transfect siRNAs into immortalized non-oncogenic mammary epithelial cells. MCF-12A cells were transfected using Qiagen (Valencia, CA, USA) RNAifect Transfection Reagent once with 5 nM of
IRF5 pooled siRNAs (Catolog #16708; Ambion, Austin, TX, USA) and harvested 24 h later, or twice (transfected a second time at the 24 h timepoint) and harvested an additional 24 h later. siGenome Lamin A/C Control siRNA (Catalog #D-001050-01-05; Dharmacon, Lafayette, CO, USA) was transfected in a similar manner. Knockdown of IRF5 was determined from Western blots by densitometry analysis of the mean pixel intensity of IRF5 normalized to β-actin.
3-Dimensional (3-D) culture in Matrigel and PCR array
3-D culture was performed as described [
32]. Cells were spread between two layers of Matrigel (Becton Dickinson) in eight-well chamber slides. Slides were incubated at 37°C in 5% CO
2/95% air for 10 days. Acini formation was visualized on a Zeiss microscope at 10 × magnification. 3-D colonies were harvested with Cultrex 3D culture Matrix™Cell Harvesting Kit (3448-020-K, Trevigen, Gaithersburg, MD, USA) following the manufacturer's instruction and total RNA isolated with Qiagen RNeasy Plus Mini kit (#74134). Total RNA was converted to cDNA with qScript™ cDNA SuperMix (Quanta BioSciences #84034; Gaithersburg, MD, USA) for PCR array and qPCR analysis. The effect of IRF5 overexpression on 84 known tumor metastases genes was analyzed using the Human Tumor Metastasis RT
2 Profiler™ PCR Array (SABiosciences PAHS-028A-2; Frederick, MD, USA) using RT
2 SYBR
® Green qPCR Master Mixes (SABiosciences, PA-012); qPCR was performed on the ABI 7300 instrument. Raw data were analyzed with SABiosciences online data analysis software. For standard q2PCR, iTaq™SYBR Green Supermix with Rox (Bio-Rad 172-5850; Hercules, CA, USA) was used. Primer sequences for standard qPCR are shown in Additional file
1, Table S1 obtained from the Quantitative PCR Primer Database [
33].
Chemotaxis assay
Chemotaxis assays were performed using 24-well transwell permeable supports (Corning Life Sciences, Lowell, MA, USA) in accordance with the manufacturer's instructions. Briefly, 100 ng/ml human recombinant CXCL12/SDF-1 (R&D Systems, Minneapolis, MN, USA) was added to 600 μl of phenol red-free DMEM medium supplemented with 10% FBS in the lower chamber. A total of 1 × 105 MDA-MB-231 cells in 100 μl of medium were added to the upper chamber, separated from the lower chamber by a membrane (6.5 mm diameter, 8 μM pore size, polycarbonate membrane). Total cell migration was obtained by calculating cell number in the lower chamber after 6 hr of incubation at 37°C in 5% CO2. Three samples were analyzed separately in duplicate, and the data were averaged for statistical analysis.
Cell surface expression of CXCR4
Cell surface expression of CXCR4 was measured by flow cytometry. MDA-MB-231 cells cultured with and without 100 ng/ml CXCL12 for six hours were stained with PE-conjugated anti-human CXCR4 antibodies or isotype control antibodies (BioLegend, San Diego, CA, USA) in accordance with the manufacturer's specifications. In brief, cells were harvested, washed in PBS, mixed with the appropriate antibody and incubated in the dark for 15 minutes before analysis by flow cytometry. A total of 10,000 events were accumulated for each analysis; samples were analyzed in triplicate.
A total of 1 × 10
6 MDA-MB-231/pBabe or MDA-MB-231/pBIRF5 cells were plated in 96-well format in triplicate four hours before transfection (SuperFect Transfection Reagent, Qiagen) with pGL3 empty vector control plasmid or the
CXCR4 luciferase promoter reporter pGL3-
CXCR4/3B/4-1(5'Δ3) (-191 to +88) [
34] from Dr. Nelson L. Michael (Walter Reed Army Institute of Research). In all wells, 40 ng of thymidine kinase driven Renilla luciferase reporter gene (Promega, Madison, WI, USA) was co-transfected to normalize for transfection efficiency. After 24 h of transfection, fresh media was added to cells with or without 100 ng/ml CXCL12 for 4 h. Post-stimulation, cell lysates were prepared, and reporter gene activity was measured using the Dual luciferase assay system (Promega) [
5]. Data are expressed as the mean relative stimulation ± S.D.
In vivo tumorigenicity assay
Four- to six-week ovariectomized, Ncr
nu/nu mice (
n = 18 per group (Charles Rivers Laboratory, Wilmington, MA, USA) were supplemented with 17 β-estradiol pellets (0.72 mg/pellet; Innovative Research of America, Sarasota, FL, USA) and used to determine the tumorigenicity of MCF-7 pooled stable transfectants [
26]. A total of 1 × 10
6 control (MCF-7/pBabe) or MCF-7/pBIRF5 cells were inoculated into opposite thoracic mammary fat pads. Ncr
nu/nu mice (
n = 15 per group) were also used for MDA-MB-231 pooled stable transfectants. A total of 2 × 10
6 control (MDA-MB-231/pBabe) or MDA-MB-231/pBIRF5 cells were inoculated into upright mammary fat pads. The primary endpoint was the incidence of proliferating tumors; secondary was tumor size. Tumor areas were estimated from the product of the two longest perpendicular measurements with a caliper. All
in vivo studies were conducted in accordance with UMDNJ New Jersey Medical School Animal Care and Use Committee approved protocols.
Statistical analyses
Data are presented as mean ± SD of data obtained from three or four independent experiments performed in duplicate. Representative experiments of multiple experiments are depicted in some figures. Comparisons between values were analyzed by the Student's t-test. Differences were considered significant at P-values ≤ 0.05. Statistical analyses were performed using Prism 4.0 (GraphPad Software, San Diego, CA, USA). Cumulative incidences of proliferating tumors in each experimental group were visualized by the Kaplan-Meier method and compared by the log rank test.
Discussion
Results presented here provide the first clear support of IRF5 tumor suppressor function and identify a new role for IRF5 in tumor cell invasion/metastasis. We demonstrated that loss of IRF5 expression correlated with advanced stages of breast cancer and invasion/metastasis. Loss of IRF5 preceded that of IRF1, but loss of IRF5 expression was not a prerequisite for IRF1 and IRF5 overexpression did not affect IRF1 levels (Figure
3A and data not shown). IRF1 was used as a comparative control given its known expression and function in breast cancer [
25]. The differential reactivity of the IRF1 and IRF5 antibodies by IF and IHC, as well as by Western blot showing they bind to discrete molecular weight bands (IRF1 approximately 48 kDa and IRF5 62 kDa), support their specificity; in addition, the same IRF1 antibody used in the manuscript by Doherty
et al. [
25] to examine IRF1 expression in FFPE samples was used in this study. Two distinct IRF5 antibodies, one from Novus Biologicals and the other from Cell Signaling, were tested and gave identical results by IF, IHC and Western blot analysis of IRF5 expression in immortalized transformed and untransformed cell lines (data not shown). Together, these data document both the specificity and non-cross-reactivity of anti-IRF1 and anti-IRF5 antibodies.
Although we found that IRF1 and IRF5 were similarly expressed in normal breast tissue and patients with ADH or IDC, significant differences were observed in DCIS suggesting the unique utilization of these two biomarkers for diagnosis and prognosis. Another important distinction between these two transcription factors was in cellular expression; IRF5 was predominantly expressed in MECs (Figure
1A, B). IRF5 was also detected in non-MECs and the surrounding stroma of early DCIS, late DCIS and IDC patients (Figure
1C). These data support distinct functions for IRF1 and IRF5 in breast tumorigenesis. MECs play a critical role in mammary gland development and loss of myoepithelial function is almost universally associated with breast cancer [
37]. MECs are localized between luminal epithelial cells and the stroma, which ideally position them to communicate with both compartments. They suppress tumor growth and invasion [
38] and degradation of the MEC layer and basement membrane is an absolute prerequisite for breast cancer invasion and metastasis [
39]. Mounting evidence also demonstrates the importance of surrounding stroma in tumor promotion [
40]. Recent data from Eguchi
et al. support a role for IRF5 in the fatty stroma [
41]. Additional experiments are necessary to determine the exact expression and function of IRF5 in tumor versus non-tumor MECs, stromal cells and non-MECs. Significant differences in gene expression have been observed between normal MECs and tumor MECs [
42,
43]. Given the known function(s) of IRF5 in regulating proinflammatory cytokine/chemokine expression [
3,
4,
6], combined with its cellular expression in breast tissue and high expression in infiltrating leukocytes in the tumor stroma of IDC patients (Additional file
5), suggest that IRF5 may play an important role in breast cancer invasion. Indeed, the van't Veer cohort placed IRF5 in a dominant gene cluster associated with lymphocytic infiltration and progressive disease [
44]. Furthermore, IRF5 is part of a 28-gene signature for predicting breast cancer recurrent and metastatic potential [
45]. Based on data presented here, we propose a two-fold function for IRF5 that is cell type-specific and lends support to the 'release' model of breast cancer invasion where phenotypic changes in MECs (loss of IRF5 expression), in coordination with the infiltration and influence of inflammatory cells (high levels of IRF5 expression), lead to the breakdown of ducts and release and invasion of tumor epithelial cells [
46].
Clinical data from tissue specimens combined with expression analyses and 3-D cultures provide the first clues that IRF5 may be involved in regulating tumor metastases, where loss of IRF5 enhances metastatic potential. A cursory review of the literature indicates that this function is unique to IRF5 and not IRF1. The molecular mechanism by which IRF5 inhibits invasion/metastasis is not yet clear but likely involves the dysregulation of genes, such as
CXCR4.
CXCR4, the receptor for chemokine CXCL12/SDF-1, was significantly down-regulated at both the transcript and protein level by IRF5 overexpression, and IRF5 inhibited promoter reporter activity (Figure
7A, C and Additional file
3). CXCR4 is an important factor in the migration, invasiveness and proliferation of breast cancer cells and silencing of
CXCR4 blocks breast metastasis [
47,
48]. Increased expression of CXCR4 in primary breast tumors has been associated with developing bone metastases [
49].
Further studies will be necessary to address the question of how or why IRF5 expression is altered in different stages of human breast cancer. Results from Q-PCR analysis of
IRF5 transcript expression (Additional file
2) support the presence of
IRF5 transcripts in cell lines that lack detectable IRF5 proteins, that is, MDA-MB-231 and T47D cells, yet the overall trend in IRF5 transcript and protein levels correlated. The
IRF5 promoter does contain a large CpG rich island [
13] suggesting that it may be susceptible to silencing by hypermethylation; yet, when MDA-MB-231, MDA-MB-436 and T47D cell lines were treated with 5-aza-2'-deoxycytidine and
IRF5 expression analyzed by RT-PCR, no change in transcript levels was detected (data not shown). It has recently been demonstrated that the
IRF5 promoter is frequently hypermethylated in hepatocellular carcinoma tissue samples [
50]. A similar study in immortalized cell lines from patients with Li-Fraumeni syndrome that had decreased
IRF5 expression showed no detectable methylation of CpG islands in the
IRF5 promoter [
51]. More recently, a single point mutation in the
IRF5 gene was identified in peripheral blood from patients with adult T-cell leukemia/lymphoma (ATL) and chronic lymphocytic leukemia (CLL) that altered the function of wild-type IRF5 [
52]. Together, these data suggest that multiple mechanisms may exist that regulate IRF5 expression and function in cancer.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
XB contributed to all aspects of the study including the design, statistical analysis and manuscript preparation, as well as carried out all in vitro and in vivo scientific assays. MH and NM analyzed H&E sections of FFPE tissue specimens, and graded and selected cases to be examined. EMP performed QPCR assays and JA participated in the staining of FFPE tissue specimens. BJB conceived and designed the study, performed statistical analyses, coordinated the requisition of FFPE tissue specimens and drafted the manuscript. All authors read and approved the final manuscript.