Background Ovarian cancer has the highest mortality among gynecological cancers due to late diagnosis and lack of effective targeted therapy. Although the study of interplay between cancer cells with their microenvironment is emerging, how ovarian cancer triggers signaling that coordinates with immune cells to promote metastasis is still elusive. Methods Microarray and bioinformatics analysis of low and highly invasive ovarian cancer cell lines were used to reveal periostin (POSTN), a matrix protein with multifunctions in cancer, with elevated expression in the highly invasive cells. Anchorage independent assay, Western blot, RNA interference, confocal analysis and neutralizing antibody treatment were performed to analyze the effects of POSTN on tumor promotion and to explore the underlying mechanism. Chemotaxis, flow cytometry and cytokine array analyses were undertaken to analyze the involvement of POSTN in cancer-associated fibroblast (CAF) and macrophage modulation. Correlations between POSTN expression levels and clinical characteristics were analyzed using the Oncomine, commercial ovarian cancer cDNA and China Medical University Hospital patient cohort. In vivo effect of POSTN on metastasis was studied using a mouse xenograft model. Results Expression of POSTN was found to be elevated in highly invasive ovarian cancer cells. We observed that POSTN was co-localized with integrin β3 and integrin β5, which was important for POSTN-mediated activation of ERK and NF-κB. Ectopic expression of POSTN enhanced whereas knockdown of POSTN decreased cancer cell migration and invasion in vitro, as well as tumor growth and metastasis in vivo. POSTN enhanced integrin/ERK/NF-κB signaling through an autocrine effect on cancer cells to produce macrophage attracting and mobilizing cytokines including MIP-1β, MCP-1, TNFα and RANTES resulting in increased chemotaxis of THP-1 monocytes and their polarization to M2 macrophages in vitro. In agreement, tumors derived from POSTN-overexpressing SKOV3 harbored more tumor-associated macrophages than the control tumors. POSTN induced TGF-β2 expression from ovarian cancer cells to promote activation of adipose-derived stromal cells to become CAF-like cells expressing alpha smooth muscle actin and fibroblast activation protein alpha. Consistently, increased CAFs were observed in POSTN overexpressing SKOV3 cells-derived metastatic tumors. In clinical relevance, we found that expression of POSTN was positively correlated with advanced-stage diseases and poor overall survival of patients. Conclusions Our study revealed a POSTN-integrin-NF-κB-mediated signaling and its involvement in enhancing M2 macrophages and CAFs, which could potentially participate in promoting tumor growth. Our results suggest that POSTN could be a useful prognosis marker and potential therapeutic target.

Optimal therapy for each patient depends on their subtype, anatomic cancer stage, gene status, and preferences. Neoadjuvant chemotherapy-treated tumors have shown attenuated tumor growth, but the therapy cannot completely reduce tumor cell dissemination to blood stream and distant metastasis. Though it has been indicated that the protein of the collagen type IV alpha 1 (COL4A1) gene is induced by p53 to inhibit angiogenesis and tumorigenic activity in cancer cells, its prognostic significance in breast cancer (BC) patients has not yet been fully elucidated. We analysed 206 BC and fresh paired-match adjacent normal breast tissue from tissue microarrays (TMAs) and COL4A1-stained TMAs using immunohistochemistry. These were used to evaluate COL4A1 expression in BC and to analyse the relationship between this expression and clinicopathological factors and prognosis. The expression of the COL4A1 protein was significantly higher in normal adjacent tissue than in the tumor tissues of BC (P<0.0001). The low COL4A1 expression of the BC patients had decreased metastasis incidence ratio than those exhibiting high COL4A1 expression (P=0.034). Low COL4A1 expression in the tumor cells of BC patients was found to significantly reduce the overall survival (OS) and relapse-free survival (RFS) rates of neoadjuvant chemotherapy patients (P=0.047 and P=0.025, respectively). We also validated the results to ensure their consistency with a web server program for survival analysis from the Cancer Genome Atlas (TCGA) database (P=0.057). Additionally, COL4A1 expression was positively correlated with p53 expression (P=0.00076). Thus, we present clinical evidence that COL4A1 expression can be used as a biomarker of better prognosis of BC patients receiving neoadjuvant chemotherapy.

No abstract