Evidence for a direct link between BMP4, BMPR1b, and TWIST-1 expression and BCR-ABL cell resistance to IM. (A) Correlation matrix representing bilateral Pearson correlation tests performed on transcriptional expression of TWIST-1, BMP2, BMP4, and BMPR1b in CD34⁺ BM or PB cells from 140 patients (under TKI treatment or CP-Diag CML patients). Positive correlations are presented in red, and negative ones in blue. (B) KCL22S overexpressing BMP4 or KCL22S transfected with an empty vector were analyzed for BMPR1b and TWIST-1 transcript. Results from each analyzed sample are expressed in arbitrary units. (C) Flow cytometry detection of BMPR1b at the cell surface of KCL22S overexpressing BMP4 or transfected with an empty vector. Data represent the percentage of positive cells of paired analyzed samples of control or BMP4-transfected TKI-sensitive KCL22. (D) CD34⁺ cells from newly diagnosed CP patients treated or not in serum-free media with IM at 1 or 2 µM, and with or without sBMPR1b. After 5 days of culture, TWIST-1 gene expression was analyzed. BMPR1b expression was modulated in the KCL22 cell line by transfection with an empty vector as a control or a BMPR1b-expressing sequence to induce its expression in KCL22-sensitive cells (E) or using a vector containing short hairpine (sh) sequence against BMPR1b to decrease its level in KCL22-resistant cells (F). After puromycin selection, cells were incubated with 1% fetal calf serum in the presence of IM at the indicated dose. After 3 days of culture, cell proliferation and viability were determined by trypan blue staining. The percentage of proliferation was determined by reference to the untreated number of viable cells ± SEM from four experiments. (G) Model for BMPR1b-mediated TKI resistance of CML LSC. CT, control; NS, not significant (P > .05). *P ≤ .05; **P ≤ .01; ***P ≤ .001; ****P ≤ .0001.