Differential gene expression between blastsnaive and blastsresist. (A) Confirmation of microarray data by Q-PCR. Genes associated with cell cycle progression (CCNB2 and CCNE1) and anti-apoptosis (TNFAIP3), and genes the functions of which in leukemogenesis were unknown (ALDH1A1, JAK3, MMP15) were up-regulated; whereas ABCA5, HPGD, and those involved in interferon pathway including IFI44L, OAS1, and RSAD2, were down-regulated in blastsresist. (B) Aldefluor activities were up-regulated by 2- to 3-fold in blastsresist in 3 patients from whom paired blastsnaive and blastsresist samples were available. (C) The CD34+ALDH+ population (of myeloblasts from patients not in this study) preferentially expressed ALDH1A1, as shown by Q-PCR. (D) CD34+ALDH+ population showed superior engraftment in NOD/SCID mice compared with CD34+ALDH− population at equal cell doses. (E) Western blot showing increase in total JAK3 and/or phosphorylated-JAK3 (p-JAK3) in 3 patients when the response to sorafenib was lost. Protein was extracted from CD34+CD33+ myeloblasts (F) Western blot of paired samples from 5 patients. Both total and phosphorylated (p) levels of FLT3 and STAT5 were down-regulated in blastsresist (R) compared with blastsnaive (N). ERK1/2 phosphorylation was specifically down-regulated in blastsresist. The changes in total and phosphorylated AKT proteins were variable. Total and phosphorylated AKT in blastsresist were decreased in AML1, AML3, and AML4 and increased in AML7. Total AKT protein was unchanged in AML10 but p-AKT was decreased. Arrows indicate specific and asterisks nonspecific bands. The numbers on top of the panel indicate the percentage of blasts in the clinical samples. Protein was extracted from LD cells. None of the samples have been cultured or manipulated ex vivo. At progression, or “resistance,” patients were given the therapeutic dosage of sorafenib, although the blood levels had not been measured.