Tumor-associated Endo180 requires stromal-derived LOX to promote metastatic prostate cancer cell migration on human ECM surfaces

Abstract

The diverse composition and structure of extracellular matrix (ECM) interfaces encountered by tumor cells at secondary tissue sites can influence metastatic progression. Extensive in vitro and in vivo data has confirmed that metastasizing tumor cells can adopt different migratory modes in response to their microenvironment. Here we present a model that uses human stromal cell-derived matrices to demonstrate that plasticity in tumor cell movement is controlled by the tumor-associated collagen receptor Endo180 (CD280, CLEC13E, KIAA0709, MRC2, TEM9, uPARAP) and the crosslinking of collagen fibers by stromal-derived lysyl oxidase (LOX). Human osteoblast-derived and fibroblast-derived ECM supported a rounded 'amoeboid-like' mode of cell migration and enhanced Endo180 expression in three prostate cancer cell lines (PC3, VCaP, DU145). Genetic silencing of Endo180 reverted PC3 cells from their rounded mode of migration towards a bipolar 'mesenchymal-like' mode of migration and blocked their translocation on human fibroblast-derived and osteoblast-derived matrices. The concomitant decrease in PC3 cell migration and increase in Endo180 expression induced by stromal LOX inhibition indicates that the Endo180-dependent rounded mode of prostate cancer cell migration requires ECM crosslinking. In conclusion, this study introduces a realistic in vitro model for the study of metastatic prostate cancer cell plasticity and pinpoints the cooperation between tumor-associated Endo180 and the stiff microenvironment imposed by stromal-derived LOX as a potential target for limiting metastatic progression in prostate cancer.

Keywords: Bone; Cell migration; Collagen; Fibroblast; Osteoblast; Prostate cancer.

Fig. 1

LOX inhibition increases type I collagen fiber curvature in matrices produced by human stromal cells. a, b Immunofluorescent images of type I collagen fibers produced by primary human trabecular bone osteoblasts (a) and human HCA2 dermal fibroblasts (b) in the absence (-BAPN) or presence (+BAPN) of LOX inhibitor (1 mM), scale bar = 50 μm. c, d Relative levels of total type I collagen deposited by osteoblasts (c) and fibroblasts (d) in the absence (–) or presence of BAPN (0.01–1 mM) (average ± s.d. of integrated fluorescent intensity of immunostaining per µm²). e, f Collagen fiber curvature ratio in the matrices deposited by osteoblasts (e) and fibroblasts (f) in the absence (–) or presence of BAPN (0.01–1 mM); 20 collagen fibers measured for each condition in duplicate. e, f Average ± s.d. values for three individual experiments, within which each experimental condition in duplicate, are shown; significant differences compared to control (no BAPN treatment) are indicated (*p < 0.05, **p < 0.001 and ***p < 0.0003). g Von Kossa staining of osteoblast derived ECM generated in the absence (−BAPN) or presence (+BAPN) of LOX inhibitor (1 mM), scale bar = 50 μm

Fig. 2

Human stromal cell-derived surfaces promote plasticity of movement in metastatic prostate cancer cells. a Stills extracted from Videos 1–12 in Online Resources 1–12 of PC3, VCaP and DU145 migrating on tissue culture plastic, commercial rat-tail type I collagen and native ECM generated by human HCA2 dermal fibroblasts (fibroblast ECM) and primary human trabecular bone osteoblasts (osteoblast ECM); scale bar = 50 μm. b, c The percentages of VCaP cells (b) and DU145 cells (c) participating in grouped and singular modes of cell migration on the four substrata. d The rounded:bipolar ratios for singular PC3, VCaP and DU145 cells migrating on the four substrata; significant increases compared to type I collagen (*p < 0.01) and fibroblast ECM (⁺p < 0.01) are indicated. b–d Average ± s.d. values for three individual experiments, within which each experimental condition was tested in quadruplicate, are shown

Fig. 3

The migration, adhesion and proliferation rates of metastatic prostate cancer cells are differentially modulated on human stromal cell-derived ECM surfaces. a–c The migratory velocities (μm/h) over a 24 h time-frame (a), adhesion levels at 1 h (b) proliferation rates at 48 h (c) of PC3, VCaP and DU145 cells cultured on tissue culture plastic (plastic), rat-tail type I collagen (type I collagen), human HCA2 fibroblast-derived ECM (fibroblast ECM) and human trabecular bone osteoblast-derived ECM (osteoblast ECM). a–c Average ± s.d. values for three individual experiments, within which each experimental condition was tested in quadruplicate, are shown. Significant increases (*p < 0.05) and decreases (⁺p < 0.05) compared to plastic are indicated

Fig. 4

Endo180 is upregulated in metastatic prostate cancer cells cultured on human stromal cell-derived ECM surfaces. a Flow cytometry analysis of PC3, VCaP and DU145 cells using three anti-human Endo180 monoclonal antibodies (A5/158, E1/183, 39.10). b Immunofluorescent staining of Endo180 (A5/158 mAb) in PC3-Endo180 cells cultured on plastic and ECM generated by primary human trabecular bone osteoblasts (osteoblast ECM); scale bar = 50 μm. c Relative levels of Endo180 expression (integrated fluorescent intensity; A5/158 mAb immunostaining) in PC3, PC3-Endo180 and DU145 cells cultured on tissue culture plastic (plastic) and ECM generated by primary human trabecular bone osteoblasts (osteoblast ECM) and human HCA2 dermal fibroblasts (fibroblast ECM); average ± s.d. values for three individual experiments, within which each experimental condition was tested in quadruplicate, are shown; significant differences compared to plastic are indicated (*p < 0.05). d Immunoblot shows Endo180 expression (A5/158 mAb) in VCaP cells cultured on plastic, rat-tail type I collagen (collagen), osteoblast ECM and fibroblast ECM for 6, 24 and 48 h (GAPDH = loading control). e Relative levels of Endo180 expression in (d) quantified using densitometry (n = 1)

Fig. 5

Silencing Endo180 inhibits metastatic prostate cancer cell migration on human stromal-derived ECM surfaces. a Brightfield and immunofluorescent images of PC3 cells transfected with control shRNA vector containing a targeting Endo180 oligonucleotide sequence (PC3-shEndo180); scale bar = 100 μm. b, c Immunoblot analysis (A5/158 mAb) (b) and corresponding denistometric analysis (c) confirming decreased Endo180 expression in PC3-shEndo180 cells compared to control PC3 cells transfected with a vector containing a non-targeting (scrambled) Endo180 oligonucleotide sequence (PC3-shSCN) (α-tubulin = loading control). d Stills extracted from Videos 13–16 in Online Resources 13–16 of PC3-shSCN and PC3-shEndo180 cells migrating on native ECM generated by human HCA2 dermal fibroblasts (fibroblast ECM) and primary human trabecular bone osteoblasts (osteoblast ECM); scale bar = 50 μm. e The rounded:bipolar ratios of PC3-shSCN and PC3-shEndo180 cells migrating on fibroblast ECM and osteoblast ECM. f Migration velocities (μm/h) of PC3-shSCN and PC3-shEndo180 cells migrating for 24 h on human fibroblast-derived ECM and human osteoblast-derived ECM; significant decreases in the velocity of PC3-shEndo180 cells compared to pc3-shSCN cells are indicated (⁺p < 0.01). g, f The adhesion levels at 1 h (g) and proliferation rates at 48 h (h) of PC3-shSCN and PC3-shEndo180 cells are shown; significant increases (*p < 0.05) and decreases (⁺p < 0.05) compared to plastic are indicated. e–h Average ± s.d. values for three individual experiments, within which each experimental condition was tested in quadruplicate, are shown

Fig. 6

Endo180 requires LOX-dependent ECM crosslinking of human fibroblast-derived ECM surfaces to promote metastatic prostate cancer cell migration. a Migration velocities (μm/h) of DU145, PC3 and PC3-Endo180 cells for a duration of 24 h on ECM derived from human HCA2 dermal fibroblasts in the absence (−) or presence of the LOX inhibitor BAPN (0.01–1 mM); significant increases (*p < 0.05) and decreases (⁺p < 0.05) compared to cells migrating on control ECM generated by fibroblasts in the absence of BAPN (−) are indicated. b Relative levels of Endo180 expression (integrated fluorescent intensity; A5/158 mAb immunostaining) in DU145, PC3 and PC3-Endo180 cells cultured for 24 h on ECM derived from human HCA2 dermal fibroblasts in the absence (−) or presence of BAPN (0.01–1 mM); significant increases compared to corresponding each corresponding cell line cultured on control fibroblast ECM (−) are indicated (*p < 0.05). c–e Migration velocities (μm/h) of PC3 (c), VCaP (d) and DU145 (e) cells migrating for 24 h on ECM derived from primary human trabecular bone osteoblasts in the absence (−) or presence of the LOX inhibitor BAPN (1 mM); no significant differences were observed. a–e Average ± s.d. values for three individual experiments, within which each experimental condition was tested in quadruplicate, are shown. f Schematic diagram that summarizes the findings of this study and their potential therapeutic implications in the context of bone and visceral metastasis. Metastatic prostate cancer cells adopt a rounded mode of cell migration on fibroblast-derived and osteoblast-derived ECM that is associated with increased Endo180 expression. Silencing Endo180 (-Endo180) prevents the translocation of prostate cancer cells on fibroblast-derived ECM and osteoblast-derived ECM due to a detachment defect and the potential loss of Endo180-dependent mechanotransduction. Inhibition of fibroblast-derived LOX activity (-LOX) uncouples Endo180 from its pro-migratory capacity, possibly via loss of its capacity for mechanotransduction in the more compliant non-crosslinked ECM. In contrast the inhibition of osteoblast-derived LOX does not affect prostate cancer cell migration, potentially due to the non-crosslinked mineralized ECM being stiff enough to promote Endo180-dependent mechanotransduction and migration. This model supports the use of anti-LOX therapy for metastasis in visceral tissue and anti-Endo180 therapy for metastasis in bone