Background
Human breast cancer is a heterogeneous disease. Using expression profiling Perou
et al.[
1] described several molecular subtypes among which are three main categories: Luminal cancers that express Estrogen (ER) and/or Progesterone (PR) Receptors but lack Human Epidermal Growth Factor Receptor-2 (HER2); HER2-overexpressing cancers; and basal-like cancers that lack ER, PR and HER2. Luminal cancers are by far the most common (>70% of cases) and have the best prognosis among the three major subtypes [
2]. Clinically Luminal cancers are treated with endocrine therapies like fulvestrant or tamoxifen that target ER or aromatase inhibitors that suppress estrogen production. Despite the initial effectiveness of these therapies, approximately one-third of Luminal cancers acquire hormone resistance, enhancing their aggressiveness and leading to local recurrence or distant metastases [
3]. Several mechanisms have been proposed for growth resumption associated with loss of regulation by estrogens [
4]. Here we address the role of the tumor microenvironment.
The microenvironment, including peritumoral stroma that surrounds malignant epithelial cells, plays a major role in cancer progression and metastasis [
5]. Carcinoma Associated Fibroblasts (CAFs) are the most abundant cells in such stroma and secrete a variety of factors that encourage tumor progression [
6]. In xenografted human Luminal MCF-7 cells, CAFs enhance tumor growth and metastasis and promote angiogenesis by recruiting endothelial progenitor cells. Other studies suggest that CAFs play a critical role in hormone resistance. For example, co-culture of tamoxifen-sensitive cells with CAFs significantly decreases their sensitivity to tamoxifen while activating MAPK and AKT pathways [
7]. In MCF-7 cells, CAFs can also induce tamoxifen and fulvestrant resistance by altering mitochondrial functions [
8]. Thus the effects of peritumoral stromal cells in breast cancers are complex and effective therapeutic targets for hormone resistance generated by the proximity of such cells may delay disease recurrence.
Growth factors are important cancer regulators. Platelet-Derived Growth Factors (PDGF A, B, C and D) belong to a four-member family of factors that activate two tyrosine kinase receptors PDGFR-α and PDGFR-β [
9‐
11]. PDGF was identified 40 years ago [
12] as a constituent of whole blood serum and it is synthesized by many different cell types, including fibroblasts [
13]. PDGF is a mitogen, but it also has significant angiogenic effects on endothelial cells. In fact, PDGF and Vascular Endothelial Growth Factor (VEGF) are structurally and functionally related and are conserved throughout the animal kingdom [
10].
PDGF signaling is altered or overactive in many malignancies [
9,
14‐
16]. Therapies targeting the PDGFR pathway reduce tumor growth in prostate [
17], endometrial [
18], pancreatic [
19] and lung [
20] cancers, as well as in osteosarcomas [
21]. In breast cancers, TGF-β initiates an autocrine PDGF/PDGFR signaling loop critical for epithelial-to-mesenchymal transition and metastasis [
22].
In vitro, PDGF-BB increases proliferation of breast cancer cells that can be inhibited by drugs targeting the PDGF pathway [
23]. PDGF also plays a major role in initiating the “desmoplastic response” of breast cancers [
24,
25]. Lastly, patients with recurrent disease have elevated circulating PDGF levels, suggesting that it may serve as a recurrence marker [
26].
We have developed a model to study the effects of peritumoral stroma
in vitro. We previously isolated and cultured malignant mouse mammary gland stromal cells we called BJ3Z cells [
27]. They are derived from normal mouse fibroblasts that were transformed by proximity to human breast cancer cells grown
in vivo as xenografts in immuno-compromised mice. BJ3Z cells are tumorigenic when injected into mice and enhance angiogenesis and proliferation of co-injected human MCF-7 cells [
28]. Here we address
in vitro mechanisms by which BJ3Z cells control growth and aggressiveness of human breast cancer cells using normal mammary gland fibroblasts (NMFs) as controls. We find that unlike NMFs, BJ3Z cells enhance proliferation of co-cultured Luminal but not basal-like breast cancer cells. Gene expression profiling shows that malignant BJ3Z cells overexpress PDGF ligands. We demonstrate that PDGF increases proliferation of Luminal breast cancer cells in the absence of estrogens. PDGF also stimulates angiogenesis in an
in vitro model. Both effects can be prevented by Imatinib Mesylate; a potent PDGF receptor kinase inhibitor. Our studies suggest that stroma-directed therapies including anti-PDGFR agents may be useful in combination therapies for Luminal cancers.
Methods
Ethics statement
This study did not involve human subjects or clinical materials. The human breast cancer cell lines are commercially available. The research was approved by University of Colorado institutional review committees and granting agencies.
Cell lines
MCF-7 human breast cancer cells were obtained from the Michigan Cancer Foundation; BT-474, MDA-MB-231, BT-20 and Human Umbilical Cord Vascular Endothelial Cells (HUVEC) were from the ATCC (Manassas VA). Transformed mouse mammary stromal cells (BJ3Z) were developed in our laboratory [
27,
29]; normal mouse mammary fibroblasts (NMF) were a kind gift of L. Wakefield (NCI) [
27,
29]. All cell lines were authenticated by Single Tandem Repeat analysis at the CU Cancer Center Sequencing Core and were mycoplasma-free. Cells were routinely passaged in minimum essential medium (MEM; Invitrogen, Carlsbad CA) containing 5% fetal calf serum (FCS; HyClone, Logan UT). For estrogen-free conditions the medium was phenol red-free and the serum was stripped of endogenous hormones by two incubations with dextran-coated charcoal (DCC). HUVEC cells were grown in F-12 K medium (ATCC) supplemented with 0.1 mg/ml heparin, 0.05 mg/ml endothelial cell growth supplement (ECGS; Cat N. 356006 BD Biosciences, Bedford, MA) and 10% FCS.
BrdU and phosphohistone H3 assays
5-bromo-2'-deoxyuridine (BrdU or BrdUrd) incorporation in MCF-7 and BT-474 cells was calculated by dual staining with human CK18 (rabbit polyclonal AP1021; Calbiochem, La Jolla CA) and BrdU (mouse monoclonal #347580; Becton-Dickinson, San Jose CA), followed by red Alexa-555 goat anti-rabbit and green Alexa-488 goat anti-mouse antibodies (Invitrogen). Basal MDA-MB-231 and BT-20 were stained for human CD44 (rabbit monoclonal 1998–1; Epitomics) or CK5 (rabbit monoclonal 2290–1; Epitomics) instead of CK18. For cells grown in conditioned media, BrdU quantitation was performed by immunocytochemistry (ICC) using Image J software. For 3D cultures immunohistochemistry (IHC) was used. Total cells were quantified by counterstaining with blue fluorescent 4’-6-diamidino-2-phenylindole (DAPI). Antibody against phosphorylated Histone H3 (Rabbit pAb Millipore # 06–570) was used for IHC as described [
30].
Proliferation rates were calculated by the ratio of BrdU + nuclei (green) to DAPI + nuclei (blue) in CK18+, CD44+ or CK5+ cells (red) using Image Pro 4.5 software (Media Cybernetics). Quantification of BrdU incorporation and phosphorylated Histone H3 assays were performed in a minimum of five different fields from three independent experiments.
For conditioned media, stock 5% FCS-containing MEM was removed from BJ3Z cells or NMFs growing in T-75 flasks at 70-80% confluence, and replaced with phenol red-free medium containing 5% DCC-stripped FCS for 24 h. Media from these cells were collected, filtered and added to breast cancer cells.
3D colonies
3D culture was performed as described [
30]. Briefly, cells were trypsinized and resuspended in phenol red-free MEM containing 5% DCC-stripped serum. Breast cancer cells (10,000-50,000) and/or BJ3Z or NMF cells (50,000) were layered on Growth Factor Reduced (GFR)-phenol red-free Matrigel (hereafter called Matrigel) and cultures were maintained for ~7 days, adding fresh medium every 2 or 3 days. To calculate proliferation indices, 7 day-old colonies were incubated 1 hr. with 0.25 mg/ml BrdU in 5% phenol red-free, DCC-stripped serum.
Expression profiling
Briefly, triplicate independent sets of mouse NMF and BJ3Z cells were grown in Matrigel as 3D colonies [
30]. On day 4, colonies were incubated 5–10 min in dispase (Cat # 354235; BD Biosciences), cells were pelleted, resuspended in RLT buffer (QIAgen, Germany), homogenized (QIAshredder™; Cat # 79654, QIAgen) and RNA was extracted (RNEasy mini-kit™; Cat # 74104; QIAgen). Microarray analyses were performed at the University of Colorado Microarray Core using Affymetrix mouse gene ST 1.0 chips. Data were analyzed with Partek Genomics Suite Software 6.6 (Partek Ltd.). Raw data were normalized and analyzed to obtain significant differences by comparing NMF versus BJ3Z using
t-test of unequal variances. The results were organized by p values (false discovery rates, FDR) of 0.05 and 0.01. This generated lists of 5,049 and 234 probesets that were significantly and differentially up- or down-regulated in NMF
vs. BJ3Z cells respectively (Additional file
1: Table S1).
Semi-quantitative RT-PCR
Briefly, total RNA samples (1–2 μg) were reverse transcribed using SuperScript II RT (Invitrogen), and cDNA samples were amplified by PCR using specific primers against FGF-5, IGF2BP3, IGF2BP1, PDGF-A, PDGF-B, DAPK-1, Caveolin-1, TGF-β1, TGF-β2, PDGFR-α and PDGFR-β (primer sequences are summarized in Additional file
1: Table S1); β-actin was used as a loading control. In all cases a PCR cycle versus intensity curve was obtained to confirm that amplified PCR products were in the linear area of the curve.
PDGFR inhibition and recombinant PDGF
The PDGFR inhibitor Imatinib as its mesylate salt (IM; also Gleevec™, Glivec or STI571), was kindly provided by Novartis Pharma AG (Switzerland). Cells were incubated with 10 μM IM in sterile water for 6 h (tubule formation assay) or 48 h (3D proliferation assay). 3D colonies of MCF-7 cells were treated 48 h with 30 ng/ml recombinant human PDGF-BB (Cat N. 14-8501-80; eBioscience, San Diego, CA) [
31].
IHC for PDGFs & PDGFRs
IHCs were performed on paraffin sections (4–5 μm) of NMF and BJ3Z cells cultured in 3D colonies as described [
30]. They used antibodies directed against PDGF-A (sc-7958; rabbit polyclonal, Santa Cruz), PDGF-B (sc-7878; rabbit polyclonal, Santa Cruz), PDGFR-α (AF1062; goat polyclonal, R&D systems) and PDGFR-β (AF1042; goat polyclonal, R&D systems). Briefly, sections were deparaffinized and antigen retrieval was performed in a pressure cooker (Bio-care Medical) at 20 psi for 5 min in citrate buffer (10 mM sodium citrate, 0.05% Tween-20, pH 6.0). Sections were blocked 30 min with 10% normal goat serum and primary antibodies were applied for 1 hr. Fluorescent secondary antibodies were: Alexa fluor 555 (red) donkey anti-goat IgG (1:300) and Alexa fluor 488 (green) goat anti- rabbit IgG (1:400; both Invitrogen). Cell nuclei were counterstained with 4-6-Diamidino-2-phenylindole (DAPI). Fluorescent images were obtained using a Nikon Eclipse E600 fluorescent microscope coupled to a RGB-MSC micro color camera, and Image Pro Plus software version 4.5 (Media Cybernetics, Silver Spring, MD).
HUVEC cells were counted, pelleted and resuspended in phenol red-free medium containing 5% DCC-stripped serum; in F-12 K (a positive control containing Endothelial Cell Growth Supplement, BD Biosciences #356006); or in 48 hr-conditioned media from NMFs or BJ3Z cells grown in phenol red-free medium and 5% DCC-stripped serum. HUVEC were seeded in duplicate into 8-well chambers (40,000 cells per well) pre-coated with Matrigel. Cells were incubated for 24 h, with images captured every 2 hours. Cells were photographed in a Nikon Eclipse Ti microscope coupled to a Nikon digital camera DS-Qi1Mc. Images were analyzed and quantified in NIS-Elements AR software version 3.2 (Nikon Corp.). For quantification, the lengths of ten tubular structures/field were measured in duplicate per condition at 6 h of incubation. The experiment was repeated three independent times with similar results.
Acknowledgements
Supported by NIH CA026869-35, the Breast Cancer Research Foundation and the National Foundation for Cancer Research (KBH), Current study was also funded by a Susan G Komen post-doctoral fellowship PDF #0706748 (MPP) and The Avon Foundation (BMJ). The authors would like to acknowledge: Dr. J. Chuck Harrell for reading and reviewing our manuscript. The microarray core and the flow cytometry core at the University of Colorado (Anschutz Medical Campus) for gene profiling and flow sorting, respectively. We thank the University of Colorado Cancer Center Histology Shared Resource facility.
Competing interests
The authors declare that they have no competing interests.
Authors’ contributions
MP, designed and carried out the experiments in Matrigel and immunohistochemistry assays, quantified results and wrote the manuscript. WD, helped designing microarray experiments and performed microarray analysis, and statistical analyses. BJ, designed experiments and helped writing the manuscript, discussion. KH, designed experiments and wrote the manuscript. All authors have read and approved the final version of the manuscript.