Background
The fate of patients with breast carcinoma is determined by distant organ metastasis rather than local disease. Life-threatening metastatic disease is the end-result of a cascade of events that begins with local invasion at the primary site and includes intravasation (into blood or lymphatic vessels), survival of tumor cells in the blood stream, extravasation at the distant site and outgrowth of metastatic lesions. Recent experimental evidence suggests that the initial steps of the metastatic cascade including intravasation occur relatively early and that the later events like extravasation and outgrowth may be the rate-limiting steps [
1]. Although progress has been made in uncovering the biology of some aspects of metastatic spread, little is known about the mechanisms that govern adaptation of disseminated tumor cells to the environment at the distant site and determine whether tumor cells remain dormant or actively proliferate. It is becoming increasingly clear, however, that complex reciprocal interactions between disseminated tumor cells and cells in the local microenvironment (i.e. the metastatic niche) play a crucial role [
2].
Syndecan-1 (Sdc1; CD138) belongs to a four-member family of transmembrane heparan sulfate proteoglycans (HSPGs) with roles in cell signaling and adhesion [
3]. Sdc1 is primarily expressed by plasma cells and epithelia, including their malignant counterparts. During development, Sdc1 expression is transiently induced in the mesenchyme and the molecule participates in paracrine epithelial-stromal interactions [
4]. This mesenchymal induction is recapitulated during malignant progression, when Sdc1 expression is observed in stromal fibroblasts in a variety of carcinoma types, including carcinoma of the breast [
5,
6]. Little is known about the mechanisms of Sdc1 induction in fibroblasts. Induction in mesenchymal cells has been linked to transcriptional regulation by members of the fibroblast growth factor family and extracellular matrix (ECM) constituents [
7‐
9].
Via its heparan sulfate (HS) chains, Sdc1 engages HS-binding ligands including growth factors and many ECM molecules - a property it shares with other HSPGs [
10]. Sdc1 core protein-specific binding interactions have been observed between its ectodomain and both integrin cell adhesion receptor subunits and receptor tyrosine kinases [
11]. Thus, Sdc1 can act as a cell surface docking station that complexes integrins and receptor tyrosine kinases (RTKs), thereby regulating cell growth and migration. Mice globally deficient in Sdc1 have a surprisingly subtle phenotype with a slight reduction in size and weight noted as the sole abnormality [
12]. When the animals are challenged, however, some defects emerge. Sdc1-knockout animals display impaired inflammatory responses, have altered vascular and endothelial cell biology and reduced tumorigenesis [
12‐
15]. Sdc1 deficiency also affects lipid metabolism and reduces tolerance to cold temperatures [
16,
17]. The molecular pathways involved in these impaired responses to external and intrinsic challenges are largely unknown.
In breast cancer, Sdc1 generally acts as a promoter of tumor growth and progression via multiple mechanisms of action. Sdc1 overexpression in human breast carcinoma correlates with a proliferative state and poor prognosis [
18‐
20]. Mice lacking Sdc1 are relatively resistant to Wnt-induced tumorigenesis, demonstrating that Sdc1 is required for efficient tumorigenesis in this model [
12]. Sdc1 modulates tumor progression not only by cell autonomous but also by cell non-autonomous mechanisms. The induction of Sdc1 expression in stromal fibroblasts triggers a reciprocal paracrine signaling loop that stimulates mammary tumor growth in vitro and in vivo [
6,
21,
22]. Sdc1-expressing stromal fibroblasts also produce an altered ECM that is characterized by parallel, aligned fibronectin and collagen fibers, which is permissive to carcinoma cell migration and invasion and thus has the potential to promote carcinoma spread and metastasis [
23]. Collectively, these findings indicate that Sdc1 can stimulate breast tumor progression at many levels.
The goal of the present study was to determine whether host Sdc1 plays a role in mammary carcinoma metastasis. We showed in two mouse strains that the ability of highly aggressive mouse mammary tumor cells to metastasize to the lungs is diminished in mice genetically deficient in Sdc1. The requirement of host Sdc1 for efficient metastasis is observed both after orthotopic (fat pad) and tail vein injection, which suggests that Sdc1 exerts its effect during the later steps of the metastatic cascade; likely during metastatic outgrowth. Elevating the ambient housing temperature to thermo-neutral conditions reduces metastatic efficiency in the wild-type animals to the level seen in the knockout mice suggesting that the Sdc1-dependent mechanism affecting metastasis is regulated by the thermogenic response.
Methods
Cells, tumor cell inoculations and scoring of metastases
The 4T1 mouse mammary tumor cells were obtained from American Type Culture Collection (ATCC) (CRL-2539) and were cultured in Roswell Park Memorial Institute (RPMI-1640) medium supplemented with 10% fetal bovine serum (FBS), 2 mM L-glutamine and penicillin/streptomycin at 37 °C in a humidified atmosphere containing 5% CO2. E0771 cells were purchased from CH3 Biosystems (Amhurst, NY, USA) and cultured in RPMI-1640 medium supplemented with 10 mmol/L HEPES and 10% FBS.
For fat pad injections, mice were anesthetized with isoflurane, and 1 × 10
7 cells in 10 μL serum-free DMEM were injected into the exposed, intact left 4th mammary fat pad as described by Miller [
24]. Tumors were allowed to grow for 30 days and then mice were humanely killed. For tail vein injections, mice were anesthetized with isoflurane and 1 × 10
5 tumor cells in 100 μL serum-free DMEM were injected through the tail-vein. After 15 days the mice were humanely killed.
Upon completion of in vivo studies, tissues were fixed for 12–18 h in 10% buffered formalin (Fisher Scientific, Waltham, MA, USA) and then processed and paraffin embedded. Hematoxylin and eosin (H&E)-stained slides were either scanned with an Aperio whole slide scanner (Leica Biosystems) or imaged by stitching individually acquired images in Adobe Photoshop. The image files showing sections of whole lungs were carefully examined and metastatic lesions were circled with an Intuos input device (Wacom) and analyzed using a combination of Photoshop (Adobe) and ImageJ (
https://imagej.nih.gov/ij/). The number of lesions per mouse, the area of each lesion and the area of total lung tissue were recorded. Tumor burden per mouse was defined as area of lung tissue occupied by metastases divided by total area of lung.
Antibodies, reagents and histological analyses
The antibodies used for immunolabeling are listed in Table
1. Mouse tissue sections were deparaffinized and for antigen retrieval, sections were boiled with citrate buffer (pH 6.0, with 0.05% Tween 20) for 30 min. In preparation for fibroblast antibody (ER-TR7) labeling, sections of mouse tissues were incubated with proteinase K working solution (20 μg/mL in TE buffer, pH 8.0) for 15 min at 37 °C. After incubation with the primary antibody (overnight at 4 °C for CD4 and CD8; 1 h at room temperature for all others) and extensive washes, horseradish peroxidase chromogenic (Ventana) or TSA Plus fluorescence detection kits (Perkin Elmer) were applied following the manufacturers’ instructions. Nuclei were counterstained with hematoxylin, 4′,6-diamidino-2-phenylindole (DAPI) or Hoechst 33,342 as appropriate. CD4 and CD8 positive T cells were visualized in mouse lung sections by manual immunolabeling using the ImmPRESS polymer detection system and diaminobenzidine (DAB) substrate.
Table 1
Antibodies used for immunolabeling
CD31 | 1:400 | Goat anti-mouse | AF3628 | R&D Systems, Minneapolis, MN, USA |
Ki67 | 1:1000 | Mouse anti-human, clone MIB-1 | M7240 | Dako, Santa Clara, CA, USA |
Ki67 | 1:200 | Rabbit anti-mouse, clone D3B5 | 12,202 | Cell Signaling Technology, Danvers, MA, USA |
αSMA | 1:500 | Rabbit anti-mouse | ab5694 | Abcam, Cambridge, MA, USA |
F4/80 | 1:400 | Rabbit anti-mouse | ab100790 | Abcam |
ER-TR7 | 1:200 | Rat anti-mouse | ab51824 | Abcam |
CD45 | 1:200 | Rabbit anti-mouse | ab10558 | Abcam |
CD68 | 1:200 | Rat anti-mouse | MCA1957T | BioRad |
Cleaved Caspase-3 | 1:200 | Rabbit anti-mouse, D175 | 9661 | Cell Signaling Technology |
Sdc-1 | 1:200 | Rat anti-mouse | NA | gift from Dr. Rapraeger |
Sdc-1 | 1:100 | Mouse anti-human, clone B-B4 | MCA681H | Serotec |
Vimentin | predilute | Mouse anti-human | 790–2917 | Ventana Medical Systems, Tucson, AZ, USA |
CD4 | 1:200 | Rabbit anti-mouse | ab221775 | Abcam |
CD8 | 1:750 | Rabbit anti-mouse | ab209775 | Abcam |
For the analysis of immunohistochemically labeled slides, at least five images per sample were captured using a brightfield or epifluorescence microscope depending on the label. The signal was quantified using a combination of Photoshop and the National Institutes of Health (NIH) ImageJ software [
25]. Lymphocyte densities were determined in the two smallest and the two largest metastases per lung. CD4+ and CD8+ lymphocyte densities were expressed as cells per 1 × 10
6 pixel (megapixel) on images taken at consistent magnification and resolution.
Tissue sections of breast carcinoma metastases to the lung from seven patients were analyzed by dual-labeling for Sdc1 and the mesenchymal/stromal marker vimentin. After deparaffinization and heat-induced antigen retrieval (EDTA, pH 8.5, 95–100 °C for 44 min), sections were incubated with anti-Sdc1 antibody (8 min, 37 °C). After rinsing, the UltraMap mouse HRP polymer kit (Ventana, Roche, catalog number 760–4313) was applied following the manufacturer’s instructions. After denaturing and rinsing, a second round of epitope retrieval under the same conditions was applied and the sections were incubated with prediluted anti-vimentin antibody (16 min, 37 °C). After rinsing, HQ mouse polymer (Roche, 760–4814) was applied for 8 min at 37 °C followed by anti-HQ HRP solution (Roche, 760–4820; 8 min, 37 °C) and the Discovery purple detection kit (Roche 760–229).
Second harmonic generation microscopy and collagen fiber analysis
H&E-stained histology slides were placed onto an optical workstation built around a Nikon Eclipse TE300 (Nikon, Tokyo, Japan), with a Ti:Sapphire laser (Spectra-Physics-Millennium/Tsunami, Mountain View CA, USA) excitation source tuned to 890 nm, focused with a × 20 Nikon Plan Apo lens (Nikon, Tokyo, Japan), filtered with a 445 nm narrow band pass filter (TFI Technologies, Greenfield MA, USA) and back-scattered second harmonic generation (SHG) signal collected with an H7422P-40 detector (Hamamatsu, Japan) and WiscScan acquisition software developed at the Laboratory for Optical and Computational Instrumentation (LOCI), University of Wisconsin, Madison, WI, USA).
Collagen fiber angles relative to the tumor boundary were analyzed using CTFIRE and the MATLAB-based CurveAlign software developed by the LOCI (
http://loci.wisc.edu/software/curvealign) [
26]. Intensity of the SHG images was analyzed using the NIH ImageJ Software (as described above).
Animals
The generation of Sdc1−/− mice has been described previously [
12]. Mice were housed at room temperature (20–23 °C unless otherwise specified) and maintained on a 12-h light and dark cycle with free access to water and food. For all tumor experiments, 6–8 week old female mice were used. For experiments using thermo-neutral conditions, mice were individually caged and housed at 31 °C (± 1 °C) for 2 weeks in a controlled environment prior to cell inoculation and were monitored daily.
Statistics
Data are expressed as mean +/− standard error of the mean. Statistical tests were performed using MStat, which is JAVA-based software written at the University of Wisconsin-Madison (
http://mcardle.wisc.edu/mstat/) or Prism 7 software (Graphpad). The non-parametric Wilcoxon rank sum, Mann-Whitney or Kruskal-Wallis tests were used unless stated otherwise and
p values less than 0.05 were deemed to be significant.
Discussion
Here we show that host Sdc1 is required for efficient metastasis of mammary carcinoma cells and that the HSPG acts by enhancing the outgrowth of metastatic lesions. The host Sdc1 effect is lost when the animals are placed in thermo-neutral housing conditions.
These observations describe a new pathway by which stromal cell-derived Sdc1 can drive cancer progression by stimulating proliferation of disseminated carcinoma cells at distant organ sites. This is relevant because it ascribes a role to Sdc1 at a critical transition step during the natural history of breast cancer. Local disease is typically controlled with surgical and radiation therapy. However, at the time of diagnosis, carcinoma cells may already have disseminated to distant organ sites, where they may lie dormant for many years. The mechanisms that govern escape from dormancy and outgrowth into clinically apparent metastatic lesions are unknown but are thought to rely on microenvironmental cues from the metastatic niche.
Co-localization studies identified Sdc1 expression in intratumoral fibroblasts within the metastatic niche and failed to detect Sdc1 in endothelial cells or leucocytes, pointing to metastasis-associated fibroblasts as the key cell type, regulating outgrowth. This is consistent with our understanding of stromal Sdc1 activity in primary breast carcinomas. However, airway epithelial cells also express Sdc1 and it is possible that epithelial Sdc1 is responsible for or contributes to metastatic outgrowth in the lung microenvironment. We also cannot rule out the possibility that low levels of Sdc1 expression in cell types other than stromal fibroblasts or airway epithelial cells trigger pathways that stimulate disseminated carcinoma cell proliferation.
The exact mechanism of Sdc1-stimulated metastatic outgrowth is uncertain. Judging from the decreased Ki67 proliferation index seen in Sdc1−/− mice, host Sdc1 stimulates proliferation of disseminated carcinoma cells. In 2D and in 3D co-culture models, fibroblast-derived Sdc1 promotes breast carcinoma cell proliferation via paracrine pathways [
6,
21]. In these models, paracrine growth stimulation requires proteolytic cleavage of the Sdc1 core protein resulting in shedding of the Sdc1 ectodomain into the pericellular space. Prior work from several groups including ours revealed matrix metalloproteinase (MMP)14 (aka MT1-MMP) as the obligatory “sheddase” [
37,
40]. Other investigators identified heparanase combined with MMP9 as a critical enzyme involved in Sdc1 shedding [
41]. In our previously published in vitro model, fibroblast growth factor 2 (FGF2) and stroma-derived factor 1 (SDF-1) complete the paracrine signaling loop that begins with the induction of Sdc1 expression in stromal fibroblasts [
21]. Whether or not Sdc1 shedding plays a role in the lung microenvironment in vivo is currently unknown. Sdc1 and its intracellular adapter syntenin are also key molecules in the generation of extracellular vesicles of the exosome class [
42]. Furthermore, Sdc1 regulates exosome cargo composition [
43]. Since exosomes participate in tumor cell-stroma interactions and exosomes have been shown to prepare the pre-metastatic niche [
44], it is conceivable that host Sdc1 stimulates metastasis by modulating exosome production or loading.
Although our results point to an Sdc1 effect on metastatic outgrowth, we cannot rule out the possibility that host Sdc1 levels affect tumor cell extravasation. Götte et al. have shown that Sdc1 deficiency increases the adhesion of leucocytes to retinal endothelium [
13]. Any role of endothelial Sdc1 in tumor cell extravasation is speculative at this point.
Sdc1 expression in fibroblasts also leads to the production of ECM with an aligned fiber architecture that is permissive to the directionally persistent migration and invasion of carcinoma cells [
23]. ECM fiber alignment in vitro requires activation of the αvβ3 integrin [
45]. Beauvais and coworkers have shown that clustering of Sdc1 on the cell surface results in the assembly of a trimeric complex that also contains an αv containing integrin and insulin-like growth factor 1 receptor (IGF1R) [
11]. Ligand-independent activation of the IGF1R triggers inside-out activation of αvβ3, which could execute ECM fiber alignment. A migration and invasion-permissive microenvironment may enable carcinoma cells to escape microenvironmental niches that suppress growth of disseminated carcinoma cells. Ghajar and colleagues have described a perivascular niche in the lung that traps disseminated mammary carcinoma cells in a dormant state and ascribed a dormancy-inducing activity to endothelial-derived thrombospondin-1 [
46].
The dependence of the Sdc1-induced metastasis-promoting effect on sub-thermo-neutral ambient temperatures is intriguing. Relief of cold stress does not readily explain the observation since Sdc1−/− animals in sub-thermo-neutral temperature (i.e. typical, mandated housing temperature) are the only ones in this experiment that experience significant cold stress and their metastasis rate does not change when moved to the higher ambient temperature. Kokolus and co-workers report that raising the temperature to thermo-neutral conditions increases CD8+ T-cells number and activity in the primary tumors, while decreasing T-helper cells [
17]. In our study, no significant difference was identified in CD4+ or CD8+ lymphocytes between Sdc1+/+ and Sdc1−/− animals. Therefore, it is unlikely that T cells are responsible for mediating the effect of host Sdc1 on metastasis but we cannot entirely rule out differences in lymphocyte activity.
Targeting disseminated breast carcinoma cells during the long period of dormancy or preventing escape from dormancy is a promising therapeutic goal. Sdc1-mediated outgrowth of metastatic lesions may be targetable by interfering with Sdc1 core protein interactions using peptide competitors [
47] or by blocking other molecules associated with the Sdc1 pathway such as integrin cell adhesion receptors or receptor tyrosine kinases like IGF1R. However, any therapeutic intervention will require a detailed understanding of the mechanism of Sdc1 action in the metastatic microenvironment.
Conclusions
In summary, we show that Sdc1 expression is induced in stromal fibroblasts in the lung metastatic microenvironment and that host Sdc1 is required for efficient outgrowth of mammary carcinoma metastases. In thermo-neutral (higher temperature of 31 °C) ambient housing conditions, Sdc1 deficiency in the host had no impact on metastasis, suggesting that the Sdc1 effect is temperature-sensitive and likely dependent on mild cold stress. These observations assign an important role to Sdc1 during the late stages of the metastatic cascade, the molecular mechanism of which requires further study.
Acknowledgements
This article is dedicated to the memory of Patricia Keely, Ph.D., collaborator, mentor and friend. The authors thank the University of Wisconsin Translational Research Initiatives in Pathology Laboratory, in part supported by the UW Department of Pathology and Laboratory Medicine and UWCCC grant P30 CA014520, for use of its facilities and services. We are also grateful to Dr. Alan Rapraeger for providing anti-Sdc1 antibody. The contents do not represent the views of the Dept. of Veterans Affairs or the United States Government.