Unveiling the role of osteosarcoma-derived secretome in premetastatic lung remodelling

Abstract

Background: Lung metastasis is the most adverse clinical factor and remains the leading cause of osteosarcoma-related death. Deciphering the mechanisms driving metastatic spread is crucial for finding open therapeutic windows for successful organ-specific interventions that may halt or prevent lung metastasis.

Methods: We employed a mouse premetastatic lung-based multi-omics integrative approach combined with clinical features to uncover the specific changes that precede lung metastasis formation and identify novel molecular targets and biomarker of clinical utility that enable the design of novel therapeutic strategies.

Results: We found that osteosarcoma-bearing mice or those preconditioned with the osteosarcoma cell secretome harbour profound lung structural alterations with airway damage, inflammation, neutrophil infiltration, and extracellular matrix remodelling with increased deposition of fibronectin and collagens by resident stromal activated fibroblasts, favouring the adhesion of disseminated tumour cells. Systemic-induced microenvironmental changes, supported by transcriptomic and histological data, promoted and accelerated lung metastasis formation. Comparative proteome profiling of the cell secretome and mouse plasma identified a large number of proteins involved in extracellular-matrix organization, cell-matrix adhesion, neutrophil degranulation, and cytokine-mediated signalling, consistent with the observed lung microenvironmental changes. Moreover, we identified EFEMP1, an extracellular matrix glycoprotein exclusively secreted by metastatic cells, in the plasma of mice bearing a primary tumour and in biopsy specimens from osteosarcoma patients with poorer overall survival. Depletion of EFEMP1 from the secretome prevents the formation of lung metastasis.

Conclusions: Integration of our data uncovers neutrophil infiltration and the functional contribution of stromal-activated fibroblasts in ECM remodelling for tumour cell attachment as early pro-metastatic events, which may hold therapeutic potential in preventing or slowing the metastatic spread. Moreover, we identified EFEMP1, a secreted glycoprotein, as a metastatic driver and a potential candidate prognostic biomarker for lung metastasis in osteosarcoma patients. Osteosarcoma-derived secreted factors systemically reprogrammed the lung microenvironment and fostered a growth-permissive niche for incoming disseminated cells to survive and outgrow into overt metastasis. Daily administration of osteosarcoma cell secretome mimics the systemic release of tumour-secreted factors of a growing tumour in mice during PMN formation; Transcriptomic and histological analysis of premetastatic lungs revealed inflammatory-induced stromal fibroblast activation, neutrophil infiltration, and ECM remodelling as early onset pro-metastatic events; Proteome profiling identified EFEMP1, an extracellular secreted glycoprotein, as a potential predictive biomarker for lung metastasis and poor prognosis in osteosarcoma patients. Osteosarcoma patients with EFEMP1 expressing biopsies have a poorer overall survival.

Keywords: EFEMP1; Extracellular matrix; Fibroblasts; Lung Metastasis; Neutrophils; Osteosarcoma; Premetastatic niche.

Fig. 1

Changes in the transcriptome and lung tissue architecture in mice treated with the secretome or with a primary tumour. A Schematic diagram illustrating the s.c. injection of the 143B Luc⁺ cells into the lower flank of mice. Lung tissue was harvested when the primary tumour (PT) reached a maximum volume of 50–60 mm³ (PT-bearing mice, n = 3–5). B Schematic diagram illustrating the preparation and administration schedule of the 143B cell-derived secretome (SCR) in mice. Animals received daily i.p. injections for 1 week (SCR-treated mice, n = 3–5). Lung tissue was harvested at the end of the treatment. C Representative images of H&E at x200 magnification (Scale bar: 20 μm), SEM (Scale bar: 40 μm), and TEM (Scale bars: 1000 and 2000 nm) of lung tissue sections from untreated mice (CTR), SCR-treated or PT-bearing mice. Black arrowheads: exudate of protein, vesicles, and fragments of the surfactant; White arrowheads: mucous granules (produced by peribronchial glands); LB, lamellar bodies; Nu, nucleus; R, red blood cells; Pn II: pneumocytes type II (alveolar cells); Fib: fibrosis. D Venn diagram of differentially expressed genes (DEGs) in lungs for each pairwise comparison: SCR vs. CTR and PT vs. CTR. E KEGG pathway enrichment analysis of DEGs. Circle sizes denote the number of genes included in a group and the colour indicates the p-value. F Bar plots depicting the manually curated common Gene Ontology (GO) terms found for the two comparison groups. Biological process (BP), cellular component (CC), and molecular function (MF) of altered genes reporting the intersections in lungs from SCR vs. CTR and PT vs. CTR. G Representative protein-protein interaction (PPI) network, constructed with the common DEGs, using the STRING database

Fig. 2

Lung microenvironmental changes in response to osteosarcoma-secreted factors. A, B Heatmaps of transcripts encoding genes involved in ECM remodelling, inflammation, and immune cell recruitment in lungs from SCR-treated or PT-bearing mice compared with controls. C qRT-PCR analysis of S100A8/A9, Prg4, Cxcr2, Cxcl2, IL-1β, IL6, TGF-β, TNF-α, and FN1 genes in lungs from SCR-treated mice or PT-bearing mice compared with controls. (n = 3–7, per group). D Representative images of fibronectin, reticulin, α-SMA, and vimentin immunostaining at x200 magnification (Scale bar: 30 μm) in lung sections from CTR mice, SCR-treated mice, or carrying a PT. E Western blot analysis of fibronectin and collagen type IV. Expression levels with graphic quantification. F Schematic diagram illustrating the analysis of lung-infiltrating neutrophils and the peripheral blood in mouse models by flow cytometry. G Flow cytometric quantification of infiltrating neutrophils in the lungs and in the peripheral blood of control, SCR-treated and PT-bearing mice (n = 5–6, per group). Data are presented as mean ± SEM from 3–7 independent biological samples. *p < 0.05, **p < 0.01 compared to control lungs (Mann-Whitney test)

Fig. 3

Increased deposition of fibronectin and collagen by activated lung fibroblasts favour the adhesion of OS cells to the lung ECM. A Schematic diagram of the establishment of the primary cultures of fibroblasts isolated from the lungs of untreated (NFs), SCR-treated (NAFs^SCR) and PT-bearing mice (NAFs^PT). B Representative image of immunofluorescence staining for α-SMA, FAP, vimentin, fibronectin and phalloidin at x20 magnification (Scale bar: 50 μm) in primary cultures of fibroblasts from CTR, SCR-treated or PT-bearing mice. Immunofluorescence quantification of α-SMA, FAP, vimentin and fibronectin (n = 3, per group). C Relative adhesion of 143B cells to increasing concentrations of fibronectin and collagen type IV ranging from 1 to 10 µg/mL (n = 3–4, performed in triplicate). Representative bioluminescence images of cell adhesion (143B cells) to different concentrations of fibronectin and collagen IV. D Representative SEM images (Scale bar: 50 μm) of decellularized lung sections and fibronectin and collagen immunostaining at x20 magnification (Scale bar: 50 μm) in decellularized lung sections from CTR mice, SCR-treated mice, or carrying a PT. E Relative adhesion of 143B cells to decellularized lung sections from CTR, SCR-treated, or PT-bearing mice (n = 5). Representative pictures of the decellularized fragments in the wells (upper row) and bioluminescent images of adhered cells (lower row). The bioluminescent signal is represented as radiance (p/s/cm²/sr). Data are presented as mean ± SEM. *p < 0.05 and ****p < 0.0001 are significantly different when compared to NFs (Kruskal-Wallis (B)); *p < 0.05 and **p < 0.01 when compared to healthy decellularized lungs (One-way ANOVA (E))

Fig. 4

Osteosarcoma-induced PMN formation promotes and accelerates the formation of lung metastasis. A Schematic diagram of the experimental model of lung metastasis. Mice were treated with 143B cells-derived secretome (SCR) for 1 week, followed by i.v. administration of 143B Luc⁺ cells into the tail vein (SCR + i.v. group). B Schematic diagram of the experimental model of lung metastasis. Mice received only the i.v. injection of 143B Luc + cells without pre-treatment with the SCR (i.v. group). C Representative bioluminescence images of lung metastasis formation in pre-treated (SCR + i.v. group) and untreated (i.v. group) mice with secretome before the i.v. injection of 143B cells. D Exponential fitting of the bioluminescence signals (photons/second) of metastatic lesions over time in the i.v. group (∇ n = 6) and the SCR + i.v. group (• n = 13 mice), and corresponding kinetic parameters. E Schematic diagram of the spontaneous metastatic mouse model. Animals were injected subcutaneously with the 143B cells. After reaching a volume of 60–100 mm³, the tumour was excised, and the animals were monitored by BLI for lung metastasis formation. F Images of the surgical resection of a primary tumour with a volume of 60 mm³ and bioluminescence images before and after the excision of the tumour. G Representative bioluminescence images at 4, 7 and 14 days after surgical resection of the primary tumour. H Histological H&E images at x100 magnification (Scale bar: 30 μm) and immunostaining for vimentin at x100 magnification (Scale bar: 30 μm) of the resected tumour. I Histological H&E images at x40 magnification (Scale bar: 20 μm) and IHC staining for reticulin (arrowed), fibronectin, α-SMA, and vimentin at x100 magnification (Scale bar: 30 μm) of lung metastatic lesions in both experimental and spontaneous mouse models. Data are presented as mean ± SEM

Fig. 5

Proteomic analysis identified EFEMP1 as a potential metastatic-related biomarker in osteosarcoma. A, B Gene ontology analysis (GO) output. Biological process (BP) and molecular function (MF) of differently expressed proteins (DEPs) in the two pairwise-comparison groups (SCR 143B vs. MG-63; Plasma PT vs. CTR). Circle sizes in BP denote the number of genes involved in the process. C Reactome pathway enrichment analysis of identified proteins in the two pairwise- comparison groups. Circle sizes denote the number of genes included in a group and colour indicates the p-value. The common pathways are highlighted. D Venn diagram showing specific and common proteins among the two pairwise groups: SCR 143B vs. MG-63 and Plasma PT vs. CTR. E EFEMP1 levels in the SCR of the metastatic 143B and non-metastatic MG-63 OS cells. F Representative western blot of EFEMP1 in the metastatic 143B and non-metastatic MG-63 OS cells. G Plasma levels of EFEMP1 in control mice (CTR), mice treated with the 143B SCR, bearing a primary tumour (PT) or with lung metastasis (Lung Met). H Representative western blot of EFEMP1 expression in 143B cells (CTR), non-targeting (NT) and siRNA knockdown of EFEMP1 cells. I EFEMP1 levels in the SCR of 143B cells, NT and siRNA EFEMP1-knockdown cells. J Representative bioluminescence images of mice treated with the SCR of control 143B (SCR + i.v. group) or EFEMP1-knockdown cells (SCR siRNA EFEMP1 + i.v. group) followed i.v. injection of 143B-Luc⁺ cells. Data are presented as mean ± SEM, from 7 to 8 independent experiments. **** p < 0.0001 were significantly different when compared with the SCR from MG-63 (unpaired t-test (E)); *p < 0.05, **p < 0.01 and ***p < 0.001 were significantly different when compared with the values present in the plasma from healthy mice (Kruskal-Wallis test (G); *p < 0.05, **p < 0.01 were significantly different when compared with 143B SCR (Kruskal-Wallis test (I))

Fig. 6

EFEMP1 expression in primary tumours correlates with poor prognosis in OS patients. A Representative images of EFEMP1 staining and H&E in biopsy samples of high-grade OS patients at x100 magnification (Scale bar: 50 μm). B, C Kaplan-Meier analysis of overall survival in chondroblastic, fibroblastic and osteoblastic OS patient samples (n = 73 patients) and with metastatic disease (n = 37 patients). Scan cut-off was used to group samples into high (blue) and low (red) EFEMP1 expressions. p-values were determined by a log-rank test