Global secretome analysis identifies novel mediators of bone metastasis

Abstract

Bone is the one of the most common sites of distant metastasis of solid tumors. Secreted proteins are known to influence pathological interactions between metastatic cancer cells and the bone stroma. To comprehensively profile secreted proteins associated with bone metastasis, we used quantitative and non-quantitative mass spectrometry to globally analyze the secretomes of nine cell lines of varying bone metastatic ability from multiple species and cancer types. By comparing the secretomes of parental cells and their bone metastatic derivatives, we identified the secreted proteins that were uniquely associated with bone metastasis in these cell lines. We then incorporated bioinformatic analyses of large clinical metastasis datasets to obtain a list of candidate novel bone metastasis proteins of several functional classes that were strongly associated with both clinical and experimental bone metastasis. Functional validation of selected proteins indicated that in vivo bone metastasis can be promoted by high expression of (1) the salivary cystatins CST1, CST2, and CST4; (2) the plasminogen activators PLAT and PLAU; or (3) the collagen functionality proteins PLOD2 and COL6A1. Overall, our study has uncovered several new secreted mediators of bone metastasis and therefore demonstrated that secretome analysis is a powerful method for identification of novel biomarkers and candidate therapeutic targets.

Figure 1

Schematic overview of (A) non-quantitative and (B) quantitative secretome analysis approaches.

Figure 2

Non-quantitative secretome analysis. (A) Overview of cell lines used for the three bone metastasis secretome signatures (BMSS). (B) Overlap of non-quantitative BMSSs. (C) Non-quantitative BMSSs were searched via the Gene Ontology (GO) database for significantly overrepresented Molecular Functions (MFs). The dashed line indicates significance cutoff (P = 0.05); green text indicates MFs significantly overrepresented in two of three BMSSs, and red text indicates overrepresentation in all three BMSSs.

Figure 3

Quantitative secretome analysis. (A) log₂ SILAC heavy (H) vs. light (L) ratios were compared for two distinct tryptic peptides, peptide 1 and peptide 2, from the same protein. (B) log₂ secreted protein and mRNA abundance ratios were compared for 1833 and MDA-231 cells. (C-E) MDA231, TSU, and 4T1 BMSSs were subjected to GO database searching (C) and Ingenuity Pathway Analysis (IPA) (D-E). The dashed lines indicate significance cutoffs (P = 0.05) for GO Molecular and Biological functions and IPA Canonical Pathways. The most significant IPA networks are shown for each BMSS, with the “collagen” connectivity node being shaded in each case (E). (F) Overlap of quantitative BMSSs.

Figure 4

Global and specific validation of BMSS proteins in experimental and clinical datasets. (A) Genes encoding proteins on the quantitative (“SILAC”) (left) and non-quantitative (right) MDA231 BMSSs were used as gene sets and tested for enrichment via Gene Set Enrichment Analysis (GSEA) in the list of genes ranked by degree of overexpression in 1833 as compared to MDA231 cell line microarray data. The panel to the right of each GSEA plot shows the corresponding heat map of the differential expression of BMSS genes sets in the 1833 and MDA231 microarray datasets. (B) The three quantitative (“SILAC”) BMSS gene sets were tested for enrichment in the list of genes ranked by degree of overexpression in primary breast cancer tumors from patients in the MSK82 dataset who developed bone metastasis as compared to patients who remained disease free. (C) Kaplan-Meier plot of relapse-free survival of breast cancer patients stratified by median primary tumor SRGN expression in the KM-Plotter database. Log-rank test P-value is displayed. (D) Kaplan-Meier plots of bone metastasis-free survival of breast cancer patients stratified by upper quartile primary tumor expression of the indicated genes in either the EMC286, NKI295, or MSK82 datasets, as indicated. Log-rank test P-values are displayed.

Figure 5

Endogenous and exogenous expression of BMSS genes selected for experimental analysis. (A-C) Endogenous mRNA expression of all genes selected for functional analysis in selected MDA231, TSU, and 4T1 family cell lines. (D) Endogenous protein expression of COL6A1 in conditioned media from MDA231 and TSU family cell lines. (E) SCP28 cells were engineered to stably overexpress the indicated genes by retroviral infection followed by drug selection. Fold overexpression of the indicated genes relative to vector (control) cells was quantified via qRT-PCR. (F) The indicated combinations of high titer retrovirus were mixed and used to infect SCP28 cells. After roughly one month of culture without drug selection, RNA was harvested and fold overexpression of the indicated genes relative to vector (control) cells was quantified via qRT-PCR.

Figure 6

Salivary cystatins, COL6A1 and PLOD2, and plasminogen activators promote bone metastasis in vivo. (A) Control and CST1-2-4 (top), COL6A1-PLOD2 (middle) and PLAT-PLAU (bottom) overexpressing SCP28 cells were xenografted into nude mice for experimental bone metastasis assays. Metastatic burden of hindlimb bone metastasis was quantified by weekly bioluminescent imaging (BLI). ^*P < 0.05. (B) BLI, (C) radiographic, and (D) TRAP staining (for mature osteoclasts) images are shown for representative mice from each experimental group. T = tumor, B = bone, OC = osteoclast. Scale bar, 50 μm in D.