The epithelial–mesenchymal transition (EMT) induced by chemotherapeutic agents promotes malignant tumor progression; however, the mechanism underlying the drug-induced EMT remains unclear. In this study, we reported that miR-448 is the most downregulated microRNA following chemotherapy. Suppression of miR-448 correlated with EMT induction in breast cancer in vitro and in vivo. With the use of chromatin immunoprecipitation-seq analysis, we demonstrated that miR-448 suppression induces EMT by directly targeting special AT-rich sequence-binding protein-1 (SATB1) mRNA, leading to elevated levels of amphiregulin and thereby, increasing epidermal growth factor receptor (EGFR)-mediated Twist1 expression, as well as nuclear factor κB (NF-κB) activation. On the other hand, we also found that the adriamycin-activated NF-κB directly binds the promoter of miR-448 suppressing its transcription, suggesting a positive feedback loop between NF-κB and miR-448. Furthermore, all patients who received cyclophosphamide (CP), epirubicin plus taxotere/CP, epirubicin plus 5-fluorouracil chemotherapy showed miR-448 suppression, an increased SATB1, Twist1 expression and acquisition of mesenchymal phenotypes. These findings reveal an underlying regulatory pathway, in which the autoregulation between NF-κB and miR-448 is important for restrain miR-448 suppression upon chemotherapy and may have a role in the regulation of chemotherapy-induced EMT. Disruption of the NF-κB-miR-448 feedback loop during clinical treatment may improve the chemotherapy response of human breast cancers in which EMT is a critical component.

We aimed to investigate the expression of RACK1 in breast cancer, evaluate its role in predicting prognosis and compare with commonly used biomarkers: Ki67, ER, PR and HER‐2 for patients with breast cancer. The RACK1 expression and its clinical significance were examined in 160 breast carcinoma patients using immunohistochemistry. Correlations of RACK1 expression with other commonly used biomarkers and survival analyses were assessed. Immunohistochemistry results showed that the number of RACK1 cases scoring 0, 1, and 2 were 66, 54, and 40, respectively. RACK1 staining was strongly related to clinical stage, histological grade, Ki67, ER, PR and HER‐2 (all p < 0.05). Consistently, all of the cases exhibiting RACK1 staining score 0 were survivors, whereas the majority (55.0%) of those exhibiting RACK1 staining score 2 were deaths. Kaplan‐Meier survival analysis of 160 cases revealed a correlation between higher RACK1 expression levels and shorter overall survival times (p < 0.001). Univariate and multivariate analyses revealed that RACK1, tumor size, lymph node metastasis, and HER‐2 were independent prognostic factors (all p < 0.05). Interestingly, receiver operator characteristic (ROC) curves showed that the ROC areas for RACK1, Ki67, ER, PR and HER‐2 were 0.833, 0.766, 0.446, 0.387, and 0.689, respectively, and the superiority of RACK1 in sensitivity and specificity as biomarker was demonstrated. To our knowledge, it is the first time to investigate the expression of RACK1, and identified that RACK1 is a superior independent biomarker for diagnosis and prognosis comparing with currently widely used diagnostic index in breast carcinoma.