Background
Breast cancer is one of the most prevalent malignancies and constitutes a tremendous medical and socio-economic burden [
1]. Increased awareness combined with advancements in screening methods and therapies has led to an improved prognosis and survival rate of breast cancer patients [
2]. Nevertheless, after initial treatment, distant metastases frequently occur after years or even decades of a disease-free interval. Bone is the most frequent site for metastases in breast cancer, and more than 70% of patients with an advanced disease suffer from bone metastases [
3,
4]. These patients have a high morbidity caused by skeletal-related events due to the predominantly osteolytic nature of the disease. Furthermore, the survival rate remains poor with only 25% of patients living more than 3 years upon diagnosis of breast cancer bone metastases [
5].
Breast cancer-induced bone destruction is a consequence of a disturbed bone remodeling, a well-characterized process known as the “vicious cycle” of bone metastases [
6]. Briefly, tumor-derived factors including parathyroid hormone-related protein (PTHrP) stimulate bone-forming osteoblasts to secrete receptor activator of nuclear factor kappa-B ligand (RANKL) as well as other cytokines. The increase in RANKL activates bone-resorbing osteoclasts and subsequent bone destruction. This results in the release of matrix-derived growth factors, such as transforming growth factor-β1 (TGF-β1), which further stimulate tumor growth [
6,
7]. Although novel therapeutic options and targets are emerging (e.g., tyrosine kinase inhibitors, microRNAs [
8‐
11]), the disease remains incurable once osteolytic lesions have developed.
Upon entering the bone, breast cancer cells are exposed to a heterogeneous bone microenvironment, which comprises different cell types and non-cellular cues including growth factors, cytokines, and the extracellular matrix [
12]. Each component of the bone microenvironment has important roles in supporting tumor cell quiescence, metastases initiation, and progression as well as resistance and/or response to anti-cancer therapy [
13‐
20]. Within the bone microenvironment, the endosteal (osteoblasts, osteoclasts, adipocytes) and the vascular (vascular endothelial cells, pericytes) niches regulate hematopoietic stem cell (HSC) renewal and differentiation via cytokines, intracellular signals, and cell-cell contacts (e.g., integrins, cadherins) [
21‐
25]. It has been suggested that tumor cell homing to bone, quiescence, and metastatic growth are mediated via these niche-controlling signals [
15,
17‐
20,
26‐
28]. Hence, modification of the bone microenvironment including the niches might offer novel therapeutic approaches to target metastatic bone disease. However, although the understanding of how tumor cells and the bone microenvironment influence each other increases continuously, the knowledge about early events of bone metastases, especially regarding the stimuli that initiate metastatic growth, is rather limited.
High abundance of the homeodomain protein TG-interacting factor 1 (Tgif1) has been associated with poor patient survival in various cancers, including upper urinary tract urothelial carcinoma, colorectal cancer, and breast cancer [
29‐
31]. Besides its role in carcinogenesis, we recently identified Tgif1 as a novel target gene of the canonical Wnt and of the Parathyroid hormone receptor type 1 (PTH1R)-dependent signaling pathways in bone [
32]. Furthermore, we determined its important role as a novel regulator of bone homeostasis by demonstrating that the deletion of Tgif1 results in a low bone turnover phenotype in vivo [
32]. Given the crucial regulatory function of Tgif1 in both tumorigenesis and bone remodeling, we hypothesized that Tgif1 could also impact the initiation and progression of breast cancer bone metastases.
Here, we report that Tgif1 expression is strongly increased in osteoblasts upon stimulation by metastatic breast cancer cells, suggesting a potential role of Tgif1 in the osteoblast-breast cancer cell interaction. Furthermore, our results reveal that medium conditioned by osteoblasts stimulates breast cancer cell migration in a Tgif1-dependent manner. This indicates that Tgif1 in osteoblasts may also play a role during the early stages of bone metastasis. In vivo deletion of Tgif1 in mice attenuates the progression of bone metastases and protects from breast cancer-induced bone destruction. Together, our findings establish osteoblasts and Tgif1 as important regulators of breast cancer cells in the bone microenvironment and of the formation of breast cancer bone metastases.
Methods
Cell culture
Two sub-clones of the mouse mammary epithelial cell line 4T1, stably expressing green fluorescence protein (GFP) or GFP together with luciferase (luc), hereafter referred to 4T1-GFP and 4T1-GFP-luc were used. For the establishment of a stable luciferase expression in 4T1 cells (purchased from ATCC), the luc2 gene was cloned into the LeGO-iG2 plasmid (PMID: 18362927) using the EcoRI restriction site. To produce infectious particles, the second-generation lentiviral packaging system was used by transient transfection of 293T cells using Lipofectamine 2000. For transduction, the viral supernatant was added in a 1:10 dilution to 50% confluent recipient cultures. Positive selection of GFP-positive cells was performed 72 h after transduction by fluorescence-activated cell sorting (FACS). The 4T1-GFP cells were kindly provided by Dr. Sonja Loges. The 4T1 cells were cultured in RPMI 1640 medium (Gibco, 61870-010) supplemented with 10% fetal bovine serum (FBS, Gibco, 10270-106) and 1% penicillin-streptomycin (P/S, Gibco, 15070-063). The mouse osteoblast cell line MC3T3-E1 and the human breast cancer cell line MDA-MB-231 (both obtained from ATCC) were cultured in α-MEM (Gibco, 22571-020) supplemented with 10% FBS and 1% P/S.
For calvarial osteoblast cultures, the calvariae were dissected from 1- to 3-day-old mice and digested sequentially (4 × 25 min) in α-MEM containing 0.1% collagenase and 0.2% dispase (both from Roche). The first cell fraction was discarded; fractions 2 to 4 were collected, combined, and expanded in α-MEM containing 10% FBS and 1% P/S.
To obtain conditioned medium, cells were washed twice with phosphate-buffered saline (PBS) and serum starved for 24 h in α-MEM supplemented with 1% FBS and 1% P/S. On the next day, the medium was collected, centrifuged for 5 min at 900g, and stored at − 80 °C. For all experiments, 50% medium conditioned by cancer cells (CCM) or medium conditioned by osteoblasts (ObCM) was used (diluted in α-MEM + 1% FBS + 1% P/S). MC3T3-E1 cells were stimulated with CCM (from 4T1 or MDA-MB-231 cells) for the indicated periods of time. Treatment of osteoblasts with a selective ERK1/2 inhibitor (SCH772984, Santa Cruz Biotechnology) was performed prior to the stimulation with CCM.
Transwell migration assay
For transwell migration assays (BD Biosciences, BIOCOAT® Cell culture inserts, 354578), 2 × 104 breast cancer cells per well were allowed to migrate through 8-μm pores towards control medium (α-MEM + 1% FBS + 1% P/S, referred to as Ctrl), ObCM, or towards α-MEM supplemented with recombinant Semaphorin 3E (Sema3E, R&D Systems, 100-500 ng/ml) for 6 h. Migrated cells were stained using Giemsa’s azur eosin methylene blue solution (Merck, HX8389304). Cells within 4 fields of view of interest were counted using the OsteoMeasure system (Osteometrics) using a × 10 objective (Olympus UPlan Fl 10x/0.30 ∞/).
RNA sequencing
For RNA sequencing, osteoblasts were isolated from the calvariae of Tgif1−/− mice and Tgif1+/+ control littermates as described above. Libraries were prepared from 1 μg total RNA using the NEBNext Ultra RNA Library Preparation Kit for Illumina (NEB). The size of the library was measured using a Bioanalyzer 2100 (Agilent Technologies), and a 51-bp single-end sequencing was used for RNA sequencing. After aligning the reads using Bowtie2 with mm9 cDNA transcriptome, reads were counted with a custom ruby script and DESeq was applied to identify differentially expressed genes.
To determine the role of Tgif1 during the establishment and progression of breast cancer bone metastases, 8–10-week-old female mice with a germ-line deletion of Tgif1 (
Tgif1−/−) or control littermates (
Tgif1+/+) were used [
33]. Mice were backcrossed from C57Bl/6 background to BALB/c background for at least ten generations. Mice were injected intracardially with 1 × 10
5 4T1 breast cancer cells (4T1-GFP or 4T1-GFP-luc) and sacrificed 5, 7, and 9 days after tumor cell injection. Metastasis formation was monitored on day 7 using bioluminescence imaging (BLI) and quantified using the Living Image Software. To quantify the number of metastases formed, BLI signals were counted per leg for each mouse. For dynamic histomorphometry, 8-week-old tumor-free mice were injected intraperitoneally 7 and 2 days prior to the sacrifice with calcein (Sigma, C0875, 20 mg/kg) and demeclocycline (Sigma, D6140, 20 mg/kg), respectively.
Sample preparation
For paraffin embedding, bones were fixed in 4% paraformaldehyde (PFA, pH 7.4, in PBS) for 48 h at 4 °C, followed by decalcification in 0.5 M EDTA/0.5% PFA for 14 days. Decalcified bones were cut into 5-μm-thick sections. For embedding in methylmethacrylate (MMA), bones were fixed in 4% PFA for 48 h. Fixed, non-decalcified bones were embedded in MMA and cut into 4-μm-thick sections.
Immunofluorescence staining and imaging
To visualize the bone marrow vasculature and single tumor cells by immunofluorescence, long bones were fixed in 4% PFA for 4 h at 4 °C and decalcified in 0.5 M EDTA (pH 8, in PBS) for at least 24 h. Bones were then embedded in gelatin [
34‐
36]. Samples were stored at − 80 °C prior to cutting into 30-μm-thick sections. Immunofluorescence staining of the vascular endothelial cell marker Endomucin (1:100, Endomucin V.7C7, rat monoclonal, Santa Cruz, sc-65495) and of the osteoblast marker Osterix (1:200, Osterix antibody (A-13), rabbit polyclonal, Santa Cruz, sc-22536) was performed on bone sections [
34‐
36] (see Table
1 for antibody details). The GFP signal of the 4T1-GFP cells was retained during gelatin embedding, allowing the visualization of the immunofluorescence without prior staining. Images were acquired using the Leica SP5 confocal microscope, × 20 objective (20x HC PL APO CS IMM/CORR, NA: 0.70, WD (mm): 0.59 (W), =.17 (oil)). The presence and localization of 4T1-GFP breast cancer cells in long bones were determined using confocal microscopy 5 days after injection. For each mouse, 3–4 non-serial sections were analyzed.
Table 1
Antibodies used in this study
Endomucin V.7C7, rat monoclonal | Santa Cruz, sc-65495 | 1:100 |
Osterix A-13 rabbit polyclonal | Santa Cruz, sc-22536 | 1:200 |
Alexa Fluor 546 goat-anti rat | Life Technologies, A11081 | 1:400 |
Alexa Fluor 546 donkey-anti rabbit | Life Technologies, A10040 | 1:400 |
Tgif1, rabbit monoclonal | Abcam, ab52955 | 1:500 |
Erk1/2 | Cell Signaling, #9107 | 1:1000 |
Phosphorylated Erk1/2 (Thr202/Tyr204) | Cell Signaling, #4379 | 1:1000 |
AKT | Cell Signaling, #4691 | 1:1000 |
Phosphorylated AKT (Ser473) | Cell Signaling, #4060 | 1:1000 |
Actin, mouse monoclonal | Abcam, MAB1501 | 1:1000 |
Peroxidase-labeled anti-mouse | Promega, W402B | 1:10,000 |
Peroxidase-labeled anti-rabbit | Promega, W401B | 1:10,000 |
Analysis of the bone marrow vasculature
To determine whether the bone marrow vasculature is altered in a Tgif1-deficient bone microenvironment, the number (/mm
2 tissue area), length (mm), and size (mm
2) of the Endomucin-positive bone marrow vasculature were analyzed on 3–4 non-serial, gelatin-embedded sections of the tibiae from mice that were sacrificed 5 days after tumor cell injection. Only mice without tumor cells were used for quantification. The analysis was performed as outlined in Additional file
1: Figure S1A using the Osteomeasure software and an Olympus BX50 microscope (Olympus UPlan Fl 10x/0.30 ∞/-). Briefly, an area of 1125 mm
2 in the metaphysis was quantified starting 180 μm away from the growth plate and 225 μm away from the medial cortex.
Micro-computed tomography
For micro-computed tomography (μCT), the tibiae were scanned during tissue fixation either in 4% PFA or in 70% ethanol at later time points. Trabecular bone volume (bone volume per total volume, BV/TV), trabecular number (Tb.N), trabecular thickness (Tb.Th), and trabecular separation (Tb.Sp) were analyzed using a vivaCT 80 scanner (Scanco Medical). Bones were scanned at 70 kVp with a pixel size of 15.6 μm. All trabecular bone surfaces 78 μm away from the growth plate were analyzed with a length of 1 mm of the region of interest. Following the acquisition of the grayscale images, images were converted into binary images with thresholds being consistent within one study, and bone parameters were calculated using the proprietary scanner software.
Immunoblot analysis
Whole-cell protein lysates were obtained using lysis buffer (150 nM NaCl, 0.5% NP-40, 0.25% sodium deoxycholate, 50 mM Tris pH 7.5) supplemented with a protease inhibitor (Roche Diagnostics, 11873580001) and a phosphatase inhibitor (Roche Diagnostics, 4906837001). Protein concentration was quantified using the Pierce™ BCA Protein Assay Kit (Thermo Fisher, 23225) according to the manufacturer’s protocol. Lysates were separated on a 12% acrylamide gel and immobilized onto nitrocellulose blotting membranes (Amersham Protran®, 0.2 μm NC, Roth) using a Tans-Blot Turbo Transfer System (30 min, 25 V, BioRad). The membranes were blocked using 5% skimmed milk (Roth, Art Nr.T145.2) for 1 h at room temperature before overnight incubation at 4 °C with primary antibodies against Tgif1 (1:500, rabbit monoclonal antibody, Abcam, ab52955), Erk1/2 (1:1000, mouse monoclonal antibody, Cell Signaling, #9107), phosphorylated Erk1/2 (Thr202/Tyr204) (1:1000, rabbit monoclonal antibody, Cell Signaling, #4379), AKT (1:1000, rabbit polyclonal antibody, Cell Signaling, #4691), and phosphorylated AKT (Ser473) (1:1000, rabbit monoclonal antibody, Cell Signaling, #4060). Immunoblot for actin was used as a loading control (1:1000, mouse monoclonal antibody, Abcam, MAB1501). Secondary antibodies were incubated for 1 h at room temperature, followed by signal detection using the Clarity Western ECL Substrate (BioRad) and the ChemiDoc™ MP Imaging System (Bio-Rad).
RNA extraction and gene expression analysis
Total RNA was isolated from cultured cells using the RNeasy Plus Mini Kit (Qiagen, 74136) according to the manufacturer’s protocol. After sacrifice, the lungs were snap-frozen in liquid nitrogen and stored at − 80 °C. Total RNA was isolated from the lungs using TRIzol reagent (Sigma, T9424) according to the manufacturer’s instructions. cDNA was synthesized using 1 μg RNA and the ProtoScript First Strand cDNA Synthesis Kit (E6300S). Quantitative real-time PCR (qRT-PCR) was performed with the CFX Connect Real-Time PCR Detection System (Bio-Rad) using the SYBR™ Select Master Mix for CFX (Applied Biosystems, 4472942). Data were normalized to the expression of the housekeeping genes glycerinaldehyde-3-phosphate-dehydrogenase (GAPDH) or beta-2-microglobulin (B2M, Table
2). GFP expression was used to determine the presence of tumor cells in the lungs 5 days after tumor cell injection.
Table 2
Oligonucleotide sequences used in this study
mGFP | CAGGAGCGCACCATCTTCTT | CTCGATGTTGTGGCGGATCT |
mB2M | CTGCTACGTAACACAGTTCCACCC | CATGATGCTTGATCACATGTCTCG |
mTgif1 | GAGGATGAAGACAGCATGGA | TTCTCAGCATGTCAGGAAGG |
mSema3E | GGGGCAGATGTCCTTTTGA | AGTCCAGCAAACAGCTCATTC |
mGAPDH | TCACCACCATGGAGAAGGC | GCTAAGCAGTTGGTGGTGCA |
Histological analysis
For the analysis of bone volume, MMA-embedded sections were stained with von Kossa/van Gieson staining. To identify osteoblasts and osteoclasts, bone sections were stained using toluidine blue and tartrate-resistant acid phosphatase (TRAP) staining, respectively. Bone cell number/mm trabecular bone surface as well as dynamic histomorphometric parameters were analyzed using the OsteoMeasure software attached to an Olympus BX50 microscope, × 20 objective (Olympus UPlan Fl 20x/0.50 ∞/0.17) according to the standards of the American Society for Bone and Mineral Research (ASBMR) [
37]. The tumor volume and the trabecular bone volume in the femora of mice were quantified using hematoxylin and eosin- or van Giemsa-stained sections (2 non-serial sections per mouse, 20 μm apart) 9 days after tumor cell injection. The analysis was performed using the OsteoMeasure software and an Olympus BX50 microscope (× 10 objective, Olympus UPlan Fl 10x/0.30 ∞/-). The total tumor area and the trabecular bone area (Additional file
1: Figure S1B) of a tissue area with a length of 2700 μm were quantified.
Enzyme-linked immunosorbent assay
Blood was collected by cardiac puncture and allowed to coagulate at room temperature, followed by centrifugation at 6200g for 10 min. The serum was collected and stored at − 80 °C until quantification of the bone formation marker pro-collagen type I N propeptide (P1NP, Immunodiagnostic Systems, AC-33F1) and of the bone resorption marker tartrate-resistant acid phosphatase (TRAP, Immunodiagnostic Systems, SB-TR103). ELISA analyses were performed according to the manufacturer’s instructions.
Statistical analyses
Statistical analyses were performed using the Prism GraphPad software (Version 8.0.1). Data were analyzed using Student’s t test when comparing two groups or by one-way analysis of variance (ANOVA), followed by Tukey’s post hoc test when comparing more than two groups. The applied test is indicated in each figure legend with a p value < 0.05 being considered as statistically significant.
Discussion
In this study, we identified Tgif1 as a novel regulator of the osteoblast-breast cancer cell interaction. We propose that the lack of Tgif1 in osteoblasts attenuates breast cancer cell migration and metastasis formation, presumably through suppression of Sema3E expression (Fig.
5e). Thus, our findings establish osteoblasts and Tgif1 as important regulators of bone metastasis.
The metastatic cascade is a multistep process, and in particular, migration and invasion are considered as hallmarks of cancer malignancy [
45]. For successful colonization of distant organs, an interaction between tumor cells and the local environment is important. Here, we demonstrate that osteoblasts mediate the chemotactic migration of breast cancer cells in vitro. In support of our results, medium conditioned by pre-osteoblasts was recently demonstrated to stimulate collective cell migration of metastatic breast cancer cells in a wound healing assay [
46]. Together, these findings suggest that osteoblasts attract breast cancer cells to the metastatic site, thereby playing a key role during the formation of bone metastases.
Tgif1 has been identified as a stimulator of cancer cell migration as shown by an enhanced migration of colon cancer cells upon overexpression of Tgif1 [
30]. Consistently, the absence of Tgif1 impaired the migration of non-small lung cell cancer cells and MDA-MB-231 breast cancer cells in vitro [
47,
48]. In breast cancer, long-term exposure to the carcinogen cadmium was shown to promote breast cancer cell migration and invasion by increasing the expression of Tgif1 [
49]. Thus, these studies indicate a cell-autonomous effect of Tgif1 in stimulating breast cancer cell migration. Here, we aimed to investigate whether Tgif1 in osteoblasts could participate in regulating the osteoblast-induced breast cancer cell migration. Indeed, medium conditioned by control osteoblasts significantly stimulated the migration of breast cancer cells, while medium conditioned by osteoblasts lacking Tgif1 failed to activate breast cancer cell migration. Hence, our data show for the first time that Tgif1 in osteoblasts supports breast cancer cell migration in a non-cell-autonomous manner.
To obtain further insights into the molecular mechanisms underlying the role of Tgif1 in the osteoblast-mediated breast cancer cell migration, we performed an unbiased RNA sequencing analysis in osteoblasts isolated from
Tgif1+/+ and
Tgif1−/− mice. In this screen, Sema3E, a class 3 Semaphorin, was identified as abundantly expressed by
Tgif1−/− osteoblasts. Class 3 Semaphorins comprise a group of secreted molecules that were originally described as chemorepulsive molecules regulating axon guidance [
50,
51]. Besides their function in the nervous system, class 3 Semaphorins have been shown to restrict cell migration in a variety of biological systems in a context-dependent manner [
52]. For instance, Sema3E plays an important role in leukocyte trafficking by inhibiting inflammation-induced neutrophil migration and recruitment to the lungs [
53]. Furthermore, migration of vascular smooth muscle cells, osteoblasts, and thymocytes is inhibited by Sema3E [
54‐
56]. In addition, Sema3E produced by immature dendritic cells inhibits the migration of natural killer cells, demonstrating the role of Sema3E in regulating cell-cell interaction [
57]. Consistently, our findings suggest that osteoblast-derived Sema3E suppresses breast cancer cell migration, providing a novel paracrine function for Sema3E.
Besides stimulating breast cancer cell migration, an increased abundance of Tgif1 has been associated with mammary tumorigenesis [
31]. In support of this finding, knockdown of Tgif1 in MDA-MB-231 breast cancer cells reduced the presence of lung metastases in mice [
48], suggesting that Tgif1 promotes cell-autonomous breast cancer growth and metastasis. While these studies established the role of Tgif1 in breast cancer cells, we determined whether Tgif1 in the tumor microenvironment affects the progression of bone metastases in a non-cell-autonomous manner. Interestingly, the deletion of Tgif1 in the bone microenvironment reduced the presence of single breast cancer cells and breast cancer micro-metastases in the bone. Furthermore, metastatic growth was attenuated in a Tgif1-deficient bone microenvironment, resulting in a reduced breast cancer-mediated bone destruction. Recently, it has been proposed that static environments (i.e., endosteal surfaces covered by lining cells, stable vasculature) maintain disseminated tumor cells quiescent, while active environments (i.e., sprouting vasculature) trigger tumor cell growth in bones [
20]. Thus, modifying the bone microenvironment might offer promising therapeutic approaches to restrict tumor growth in bone [
58].
As components of the heterogeneous bone microenvironment, the vascular and endosteal niches are of particular importance for metastatic breast cancer growth in bones [
16,
38,
59]. Our dedicated analysis of the vascular network revealed no alterations of the number, length, or size of the Endomucin-positive vasculature in the long bones of Tgif1-deficient mice. Previously, it has been shown that silencing Tgif1 expression decreased the proliferation while it increased the tube formation of endothelial cells in vitro [
60]. In addition, an increased angiogenic potential as determined by vascular network formation assays was observed upon silencing of Tgif1 expression [
60]. However, our results indicate that these in vitro findings do not translate into in vivo conditions and cause changes in the mineralized surface, bone marrow vasculature. In contrast, the number and activity of osteoblasts including the bone formation rate, osteoid volume, and surface were significantly reduced in
Tgif1−/− mice compared to control littermates. This suggests that the attenuated metastatic burden in Tgif1-deficient mice is, at least in part, mediated by osteoblasts rather than by the bone marrow vasculature. In vitro, Tgif1-deficient osteoblasts reduced breast cancer cell migration in a Sema3E-dependent manner. It is therefore likely that Tgif1 in osteoblasts also regulates breast cancer cell migration to the metastatic site in vivo. However, the strong reduction of the number of active osteoblasts suggests that additional mechanisms independent of cell migration may exist that control osteoblast function and consequently breast cancer cell proliferation and disease progression in vivo. While the experimental approaches used in this study do not allow distinguishing between these possibilities, the contribution of the migration-stimulating effect of Tgif1 and the effect on osteoblast activity would need to be elucidated in the future. Furthermore, to better understand which step of the metastatic cascade Tgif1 controls precisely, additional in vivo models could be employed. In the present study, a syngeneic mouse model was chosen to preserve an intact immune system, which has an important role in bone metastasis progression. However, the 4T1 breast cancer cells grow very aggressively in the bone, which makes it challenging to distinguish between the different steps of disease progression such as cancer cell homing, dormancy, micrometastasis formation, and relapse. Therefore, future experiments may include less aggressive xenograft models that recapitulate the corresponding clinical situation more closely.
In summary, this work demonstrates that Tgif1 in the bone microenvironment is implicated in the establishment and progression of breast cancer bone metastases and might therefore provide novel therapeutic opportunities to treat the initiation and progression of breast cancer metastasis to bones.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.