Expression of RARα2 is diminished in AML cell lines and patient samples, whereas RARα1 expression is significantly reduced only in patient samples. (A) RARα1 and RARα2 expression levels (♦ and ▩, respectively) were calculated by real-time RT-PCR. The samples analyzed were: CD133⁺ (n = 4) and CD33⁺ (n = 4) CB cells from healthy full-term infants; primary non-APL AML cells from consenting patients (n = 19; consisting of AML FAB subtypes M0, M1, M2, M4, and M5); primary APL cells (AML M3) from consenting patients (N = 8); and AML cell lines (n = 4; NB4 (APL), Kasumi-1, THP1 and U937). n indicates the number of individuals or cells lines. Indicated values refer to the mean; error bars, plus or minus SEM. Asterisks indicate significant differences from control and + symbols between the indicated cell populations (^+/*P < .05; ^++/**P < .01; ^+++/***P < .001). RNA isolation, cDNA preparation, and real-time PCR analysis were performed as previously described.⁷  RARα expression was normalized against the beta-2-microglobulin (β₂M) housekeeping gene. For absolute quantification of RARα, PCR products were cloned into the pDrive vector (Qiagen, Valencia, CA), and the corresponding plasmid DNAs were used to generate standard curves. Negatives (nontemplate controls) were always included. Statistical analyses were performed as previously described.⁷  (B) RARα1, RARα2, C/EBPϵ, and CYP26A1 expression levels were calculated by real-time RT-PCR, except results for AML patient samples (n = 5) are shown relative to the values obtained for CD133⁺ CB cells. Pearson correlation coefficient reveals that RARα1 expression positively correlates with that of ATRA-responsive RARα2, C/EBPϵ, and CYP26A1 (r = .916, .877, and .748, respectively). The following PCR primers were used: C/EBPϵ-Fwd, 5′-GCTGTGGCGGTGAAGGAGGAG-3′ and C/EBPϵ-Rev, 5′-CAGGGGGGTGCGGCAGTGGC-3′; CYP26A1-Fwd, 5′-CGCATCGAGCAGAACATTCGC-3′ and CYP26A1-Rev, 5′-AAAGAGGAGTTCGGTTGAAGATT-3′. Statistical analyses were performed as previously described.⁷  Error bars represent SD.