Nakazawa, K.T. tumor cells and promoted invasion in residual tumors. Knockdown of periostin inhibited outgrowth and invasion of mesenchymal tumor cells upon chemotherapy. Furthermore, chemotherapy upregulated cancer-specific variants of periostin and application of a blocking antibody specifically targeting those variants overcame chemoresistance and halted disease progression without toxicity. Together, these data indicate that periostin plays a key role in EMT-dependent chemoresistance and is a promising target to overcome chemoresistance in triple-negative breast cancer. Introduction TNBC is an aggressive subtype of breast cancer that is closely related to basal-like breast cancer with a strong EMT gene signature with poor overall prognosis. Due to a lack of estrogen receptor (ER), progesterone receptor (PR) or human epidermal growth factor receptor 2 (HER-2) expressions for targeted therapy in TNBC, its treatment largely consists of cytotoxic chemotherapies with anthracyclines and taxanes. Although a subset of TNBCs are responsive to initial chemotherapy, the probability of early relapse is significantly higher compared to other subtypes of breast cancer1. Several studies report that a number of EMT markers are associated with chemoresistance or poor prognosis in TNBCs2C4. However, the cause of chemoresistance in patients with TNBC is unclear and there is an urgent need to overcome chemoresistance and prolong therapeutic response. The EMT program is a critical process during embryo development by which epithelial cells lose epithelial polarity, weaken cell-cell adhesions and obtain a migratory phenotype5. Many studies reveal a functional involvement of EMT in various steps of cancer progression: invasion, metastasis, chemoresistance and relapse6,7. Therefore, the EMT regulatory pathway has attracted great therapeutic interest in cancer treatment. Specifically, a number of studies showed that tumor cells Propiolamide undergo EMT to escape from chemotherapy and other targeted therapies. However, therapeutic strategy targeting EMT has not been developed to overcome therapy resistance. Periostin (POSTN, osteoblast-specific factor 2, OSF-2) is a secreted extracellular matrix (ECM) protein, which was originally identified in osteoblasts8. Periostin overexpression and function have been implicated in several types of human cancer9,10. In this report, we set to determine the functional role of periostin in EMT-mediated chemoresistance in TNBC and to explore its clinical implications in overcoming chemoresistance in TNBC. Results Chemotherapy upregulates periostin and enriches the mesenchymal population in TNBC xenografts and patient samples To Propiolamide explore the role of EMT in chemotherapy, we generated tumor xenografts using a human basal-like TNBC model, MCF10DCIS. MCF10DCIS cell line is derived from a premalignant epithelial breast tumor xenograft and lacks the expression of ER, PR, or HER-2. Mice carrying MCF10DCIS human mammary tumors were treated with paclitaxel (tubulin inhibitor, PTX), doxorubicin (DNA intercalater, DXR) and cyclophosphamide (DNA alkylator, CPA), which are standard chemotherapies for TNBC. All three treatments inhibited tumor growth but did not induce complete regression (Fig.?1A). We then collected vehicle-treated and PTX-treated tumors and performed high-throughput mRNA quantitation of 770 genes associated with cancer progression using the nCounter analysis system. Among the 236 genes upregulated upon chemotherapy, 57 are mesenchymal-annotated genes (Supplementary CCNB2 Table?S1), suggesting the induction of EMT. Since nCounter analysis system cannot quantitatively differentiate between human mRNAs from tumor cells and mouse mRNAs from stromal cells, we used qPCR analysis with human gene-specific primers to confirm that chemotherapies upregulated EMT-inducing transcription factors specifically in residual tumor cells (Fig.?1B). Open Propiolamide in a separate window Figure 1 Chemotherapy upregulated periostin expression in chemoresistant triple-negative breast cancer xenografts and human TNBC patient samples. (A) Tumor growth curve in response to paclitaxel (PTX), doxorubicin (DXR) and cyclophosphamide (CPA) in MCF10DCIS xenografts. Arrows indicate times of drug administration. Data are presented as mean??standard error (n?=?6). (B) Quantitative PCR analysis of indicated genes in tumors harvested at the end of the experiment in A. Relative Propiolamide values are presented as mean??standard error (n?=?3). *test. (C) Correlation analysis of expression in 59 human breast cancer cell lines. Heatmap of gene expression and Pearson product-moment correlation coefficient (r value) are shown. (D) Patient survival analysis in basal-like breast cancer patients (n?=?580) according to expression (low: 66.6%, high: 33.3%). Hazard ratio (HR), 95% confidence intervals. value, log rank test. (E) Immunohistochemistry staining of periostin with human TNBC samples. Scale bar, 100?m. (F) Periostin score before and after chemotherapy in 26 matched pair TNBC samples which showed residual disease after chemotherapy. ***mRNA expression is significantly correlated with an EMT gene signature in Cancer Cell Line Encyclopedia (CCLE) database (Supplementary Fig.?S1C). Among various.