ATRA resolves the differentiation block in t(15;17) acute myeloid leukemia by restoring PU.1 expression

Abstract

Tightly regulated expression of the transcription factor PU.1 is crucial for normal hematopoiesis. PU.1 knockdown mice develop acute myeloid leukemia (AML), and PU.1 mutations have been observed in some populations of patients with AML. Here we found that conditional expression of promyelocytic leukemia-retinoic acid receptor alpha (PML-RARA), the protein encoded by the t(15;17) translocation found in acute promyelocytic leukemia (APL), suppressed PU.1 expression, while treatment of APL cell lines and primary cells with all-trans retinoic acid (ATRA) restored PU.1 expression and induced neutrophil differentiation. ATRA-induced activation was mediated by a region in the PU.1 promoter to which CEBPB and OCT-1 binding were induced. Finally, conditional expression of PU.1 in human APL cells was sufficient to trigger neutrophil differentiation, whereas reduction of PU.1 by small interfering RNA (siRNA) blocked ATRA-induced neutrophil differentiation. This is the first report to show that PU.1 is suppressed in acute promyelocytic leukemia, and that ATRA restores PU.1 expression in cells harboring t(15;17).

Figure 1.

Conditional expression of PML-RARA in myeloid cells suppresses PU.1. (A) Top: U937 cells stably transfected with the PML-RARA cDNA (U937-PR9) under the control of the zinc-inducible metallothionein promoter were analyzed by Northern blot for PU.1 mRNA expression before and after treatment with 100 μM zinc. Bottom: The same blot was hybridized with a GAPDH probe. Zn indicates days of zinc treatment. (B) Western blots using a RARA antibody (top), a PU.1 antibody (middle), and a β-tubulin antibody (bottom). The PML-RARA protein was induced by treating U937-PR9 cells with 100 μM zinc. RARA indicates the endogenous RARA protein as present in U937 cells. Right panel: U937 cells lacking the zinc-inducible construct were treated with zinc to exclude nonspecific effects of zinc. (C) EMSA of binding activity to the PU.1 site in the M-CSF receptor promoter (left panel). U937-PR9 cells were induced with zinc. X represents nonspecific binding commonly observed with this oligonucleotide; ivt; in vitro-translated PU.1 protein is run as a control; P, free probe alone; Cos, nuclear extracts from Cos cells serve as a negative control; SS, supershift with specific antibody; and PU.1, shift of PU.1 protein binding to this particular site. The right panel represents binding of the same extracts to an NFY site as a control for loading and integrity of the extracts. C indicates competition with cold oligo.

Figure 2.

Leukemic cells from a patient with APL (HT93 cells) have suppressed PU.1 expression. (A) Northern blot analysis using a PU.1 probe showing myeloid U937 cells lacking the PML-RARA translocation (first lane), t(15;17) HT93A cells (second lane), and uninduced U937-PR9 cells (third lane). The somewhat lower PU.1 expression in U937-PR9 versus U937 cells may be due to leakiness of the system and thus low level expression of PML-RARA in the absence of zinc. The blot was reprobed with a GAPDH probe. (B) Prolonged exposure of a Western analysis using a PU.1 antiserum (top). U937 cells served as a positive control. The blot was reprobed with a β-tubulin antibody (bottom). (C) EMSA studying the binding activity to the PU.1 site in the M-CSF receptor promoter. U937 cells (U9; lanes 4 and 6) were compared with HT93A cells (HT; lanes 3 and 5). ivt indicates in vitro-translated PU.1 protein is run as a control; P, free probe alone; SS, supershift with specific antibody; PU.1, shift of PU.1 protein binding to this particular site; and X, nonspecific binding.

Figure 3.

ATRA restores PU.1 expression in APL cells. (A) Northern blot analysis of HT93A cells treated with 1 μM ATRA and harvested at the time points indicated. The blot was probed for PU.1 (top), G-CSF receptor (middle), and GAPDH expression (bottom). (B) HT93A cells were treated with 1 μM ATRA. PU.1 mRNA expression was determined by quantitative real-time PCR. Results are given as fold induction of the PU.1/GAPDH ratio compared with the value of HT93A cells before treatment (0 hours). Mean values and SD (error bars) are depicted. (C) Western analysis using a RARA antiserum (top), a PU.1 antibody (second), an M-CSF receptor antibody (third), and a β-tubulin antiserum (bottom). HT93A cells were harvested before and after ATRA treatment for 24 hours. (D) EMSA of DNA-binding activity to the PU.1 site in the M-CSF receptor promoter. HT93A cells were stimulated with 1 μM ATRA and harvested at the time points indicated. ivt indicates in vitro-translated PU.1 protein; P, free probe alone; SS, supershift with specific antibody; PU.1, shift of PU.1 protein binding to this particular site; and X, nonspecific binding. The panel on the right represents binding of the same extracts to an NFY site as a control for loading and integrity of the extracts. N indicates treatment of the extracts with normal rabbit serum; and C, competition with cold oligo. (E) ATRA treatment overcomes the suppressive effect of PML-RARA on PU.1 expression. Northern blot showing PU.1 mRNA expression of U937-PML/RARA (U937-PR9) cells (top). Lane 1 depicts U937-PML-RARA cells in the absence of zinc or ATRA. Lane 2 shows U937-PR9 cells after 2 days of treatment with zinc (100 μM); lane 3 demonstrates U937-PR9 cells after 2 days of treatment with zinc (100 μM) and also treated with both ATRA (1 μM) and zinc for the last 24 hours. The same blot was probed for GAPDH (bottom). (F) Western blot analysis of whole-cell lysates of U937-PR9 cells. Lane 1 represents unstimulated cells. Cells were induced with zinc for 1 and 2 days (lanes 2 and 3). In lane 4 cells were treated for 48 hours with zinc, and then for 18 hours with both zinc and ATRA. Top: Staining with a RARA antibody detecting the fusion protein PML-RARA and the endogenous RARA. Middle: Probing the same blot with an antibody-detecting PU.1 protein. Bottom: The blot was restained for β-tubulin. (G) ATRA sensitivity is a prerequisite for PU.1 restoration in APL cells. Whole-cell lysates were analyzed by Western blot using a PU.1 antibody. Unstimulated U937-PR9 cells (lane 1) served as a positive control and Jurkat cells as a negative control (lane 3). Lane 2 shows U937-PR9 cells after 24 hours of PML-RARA induction. NB4 cells (lanes 4 to 6) and the ATRA-resistant APL cell line NB4-R1 (lanes 7 to 9) were induced for 12 or 24 hours with 1 μM ATRA. The blot was restained for β-tubulin (bottom).

Figure 4.

CEBP factors and OCT-1 induce activity of the proximal PU.1 promoter in HT93A APL cells following ATRA stimulation. (A) A series of PU.1 promoter constructs were ligated to the luciferase gene in the pxp2 vector. HT93A cells were transfected with equal amounts of each of the deletion constructs and induced for 18 hours with 1 μM ATRA. The results are given as fold activation compared with HT93A cells transfected with the empty pxp2 vector only. (B) The sequence between -48 bp and -86 bp of the human PU.1 promoter contains a CEBP and an octamer consensus binding site (with the sites underlined and italicized). The brackets demonstrate the sequences used in the EMSA assays. (C) EMSA studying binding activity to the CEBP consensus site present in the PU.1 promoter with the CEBP probe (-80 to -60 bp). HT93A cells were stimulated with 1 μM ATRA and harvested at the time points indicated. SS indicates supershift using specific antibodies against CEBPA, CEBPB, and CEBPE. Specificity of the antibodies was verified by supershifting extracts from K562 cells transfected with human CEBPA, CEBPB, or CEBPE cDNAs; no cross-reactivity was observed (data not shown). C indicates competition with cold oligo. (D) EMSA of binding to the OCT site in the PU.1 promoter before and after 18 hours of stimulation with 1 μM ATRA (left panel). The probe is depicted in panel B as “OCT-1” extending from -66 to -42. Cos OCT-1 indicates Cos cells transfected with an OCT-1 expression plasmid as a positive control (lanes 2 and 3). Cos OCT-2 indicates Cos cells transfected with an OCT-2 expression plasmid as a positive control (lanes 4 and 5). Supershifts are depicted for OCT-1 (lanes 3, 7, 9, 11) and OCT-2 (lane 5). U937 indicates binding activity in lane 6, and supershift with OCT-1 in lane 7; HT93A, increase of binding activity before and after induction with ATRA (lanes 8 and 10), with OCT-1 supershifts in lanes 9 and 11; and P, free probe alone. Right panel shows EMSA with a probe with a mutation in the OCT site; no binding activity was detectable with this probe. (E) Western blot analysis of HT93A cells treated with 1 μM ATRA using a CEBPA antibody (top), a CEBPB antibody (second), a CEBPE antibody (third), an OCT-1 antibody (fourth), and a β-actin antibody (bottom). (F) H1299 cells were transiently transfected with the -86-bp PU.1 promoter (top 8 bars) and pcDNA3 vector (V) or equal amounts of expression plasmids for CEBPA, CEBPB, CEBPE, and/or OCT-1. OCT-1 was always cotransfected with equimolar amounts of BOB-1. Results are given as fold induction of the -86-bp promoter alone (to P = 1-fold). CEBP mut bars represent transfections with the -86-bp construct with the CEBP site mutated, whereas OCT mut bars are experiments with the -86-bp construct with the OCT site mutated. The bottom 4 bars are transfections with the -86-bp construct with both sites mutated (CEBP+OCT). (G) ChIP assay of the PU.1 promoter in HT93A cells treated with 1 μM ATRA. Depicted are the results of the quantitative RT-PCR for CEBPA, CEBPB, CEBPE, and OCT-1 binding to the PU.1 promoter at days 0, 1, 3, and 5 after ATRA treatment. (A, F-G) Mean values and SD (error bars) are depicted.

Figure 5.

PU.1 expression in primary APL cells: inverse correlation with PML-RARA at diagnosis, and induction of expression following treatment with ATRA in vivo. (A) Correlation of PU.1 and OCT-1 mRNA expression and inverse correlation between PML-RARA vs. PU.1 and OCT-1 transcripts in bone marrow cells at diagnosis of 9 patients with newly diagnosed t(15;17) AML-M3 as assessed by quantitative real-time PCR. The median value was determined as 100% for PML-RARA, OCT-1, and PU.1, respectively. The percentage of blasts was more than 90% in all patients. The correlation coefficient PML-RARA versus OCT-1 was r = -0.887, for PML-RARA versus PU.1 it was r = -0.912, and for OCT-1 versus PU.1 it was r = 0.907. (Different symbols indicate different patients.) (B) RNA expression was determined by quantitative RT-PCR of CEBPA, PU.1, CEBPB, OCT-1, and CEBPE in primary cells from a patient with newly diagnosed APL treated with orally given ATRA (45 mg/m²). No additional chemotherapy was given. The cells analyzed on a daily basis were obtained from peripheral blood (Table S1). Error bars depict the standard deviation as the result of triplicate mRNA determinations.

Figure 6.

Restoring PU.1 expression in PML-RARA leukemic cells induces granulocytic differentiation. (A) Western blot for PU.1 (top) detecting the PU.1-ER fusion protein in HT93A cells transduced with a PU.1-ER-expressing pBabePuro retrovirus (HT-PER) or with the vector alone (HT-V). Cells were treated with 1 μM β-estradiol as indicated. C indicates that Cos cells serve as a negative control and U937 cells as a positive control for endogenous PU.1 expression. The blot was also stained with an antibody against the M-CSF receptor (middle) and β-tubulin (bottom). (B) HT-PER cells and the control line HT-V were induced for 6 days with 1 μM β-estradiol. Fluorescence-activated cell sorting (FACS) analysis was performed at day 0 (left panels) and day 6 (right panels) using a PE-conjugated CD11b antibody (top panels) or a G-CSF receptor antibody (bottom panels). (C) HT-PER and HT-V cells were treated with 1 μM β-estradiol (Est) or 1 μM ATRA. Cell-cycle analysis was performed using propidium iodide staining at day 0 and 2. ▪ indicates the percentage of cells in G₀-G₁ stage, whereas □ indicates cells in G₂-M stage and indicates cells in S phase. Median values are shown and error bars depict standard deviation from 3 independent experiments. (D) Wright-Giemsa staining of HT93A cells expressing the PU.1-ER fusion (HT-PER cells). The top 3 panels show unstimulated HT-PER cells (left panel), after 6 days of ATRA treatment indicating that the cells still differentiate toward neutrophils after having established the stable estrogen inducible system (middle panel), and after 6 days of treatment with 1 μM β-estradiol indicating differentiation toward neutrophils after 6 days (right panel). The bottom 3 panels depict the HT-V control line, which fails to show morphologic changes after β-estradiol treatment (right panel), but differentiates toward neutrophils after 6 days of ATRA treatment (middle panel). Images were visualized using a Zeiss Axioplan 2 microscope (Carl Zeiss, Oberkochen, Germany) and a 100 ×/2 numeric aperture objective. Images were acquired using a Zeiss AxioCam camera.

Figure 7.

Reduction of PU.1 expression by siRNA blocks ATRA-induced differentiation of HT93A cells. (A) HT93A cells were stably transfected with an MSCV retrovirus expressing PU.1 siRNA to reduce PU.1 expression (siPU.1-1 and siPU.1-2). Two pools of cells were expanded and analyzed after 1 month. Western blot for PU.1 demonstrates that treatment with 1 μM ATRA induced only small amounts of PU.1 protein in siPU.1 cells after 2 days compared with vector-transfected siV cells. (B) siPU.1 cells (pools siPU.1-1 and siPU.1-2) were induced with 1 μMATRAfor 5 days. FACS analysis was performed at day 0 (d0) and day 5 (d5) using a G-CSF receptor antibody. Parental HT93A cells transfected with the siRNA vector (siV) served as a positive control (right panel). (C) Wright-Giemsa staining of siPU.1 cells expressing PU.1 siRNA is shown before and after 5 days of ATRA treatment, indicating that siPU.1 cells morphologically fail to respond to ATRA treatment. Right panel: ATRA treatment induces neutrophil differentiation in the siV vector-transfected HT93A cells. (D) HT93A cells were transiently transfected with siRNAs against OCT-1, CEBPA, CEBPB, or CEBPE. PU.1 mRNA expression was determined by RT-PCR after 48 hours of ATRA treatment (and transfection). V indicates the silencer negative control siRNA no. 2 (Ambion) was used to control for nonspecific effects in siRNAexperiments.