RUNX1 regulates corepressor interactions of PU.1

Abstract

The transcription factor (TF) RUNX1 cooperates with lineage-specifying TFs (eg, PU.1/SPI1) to activate myeloid differentiation genes, such as macrophage and granulocyte macrophage colony-stimulating factor receptors (MCSFR and GMCSFR). Disruption of cooperative gene activation could contribute to aberrant repression of differentiation genes and leukemogenesis initiated by mutations and translocations of RUNX1. To investigate the mechanisms underlying cooperative gene activation, the effects of Runx1 deficiency were examined in an in vitro model of Pu.1-driven macrophage differentiation and in primary cells. Runx1 deficiency decreased Pu.1-mediated activation of Mcsfr and Gmcsfr, accompanied by decreased histone acetylation at the Mcsfr and Gmcsfr promoters, and increased endogenous corepressor (Eto2, Sin3A, and Hdac2) coimmunoprecipitation with Pu.1. In cotransfection experiments, corepressors were excluded from a multiprotein complex containing full-length RUNX1 and PU.1. However, corepressors interacted with PU.1 if wild-type RUNX1 was replaced with truncated variants associated with leukemia. Histone deacetylase (HDAC) enzyme activity is a major component of corepressor function. HDAC inhibition using suberoylanilide hydroxamic acid or MS-275 significantly increased MCSFR and GMCSFR expression in leukemia cell lines that express PU.1 and mutated or translocated RUNX1. RUNX1 deficiency is associated with persistent corepressor interaction with PU.1. Thus, inhibiting HDAC can partly compensate for the functional consequences of RUNX1 deficiency.

Figure 1

In Runx1-deficient (shRunx1) cells, Pu.1 (OHT) repressed stem cell genes but myeloid differentiation gene activation was impaired. (A) The pattern of mRNA expression in PUER shRunx1 compared with PUER empty vector (control) cells treated with OHT. Gene expression after OHT addition was measured by quantitative RT-PCR. Stem cell genes are Bmi-1, Hoxb4, and c-Kit. Differentiation genes are F4/80, Mcsfr, and Gmcsfr. Data are mean ± SD. shRunx1 versus control for each time point: *P < .05, **P < .01 (Student t test). Experiments were performed in triplicate. (B) The pattern of c-Kit and F4/80 protein expression. Measured by flow cytometry 48 hours after OHT. MFI indicates mean fluorescence intensity of all cells. (C) OHT effect on cell proliferation of shRunx1 and control cells. Cell counts were measured daily with an automatic cell counter. Experiments performed in triplicate. Error bars represent SE. (D) OHT effect on cell morphology of shRunx1 and control cells. Cell morphology was evaluated on Giemsa-stained cytospin slides 48 hours after OHT. Microscope: Leica DMR, 63×/1.32 oil PH3 type N immersion oil. Image capture: CRI Nuance NzMSI-FX with Nuance 2.8 software. (E) Decreased histone acetylation at Gmcsfr and Mcsfr proximal promoters in shRunx1 compared with control cells. The Gcsfr promoter, which is not known to be regulated by Runx1/PU.1, was analyzed as control. ChIP performed with anti-H3Ac. Coimmunoprecipitation of promoter regions was analyzed by quantitative RT-PCR. Data are mean ± SD. shRunx1 vs control for each time point after OHT: *P < .05, **P < .01 (Student t test). Experiments performed in triplicate.

Figure 2

Increased coimmunoprecipitation (co-IP) of endogenous corepressor (Eto2, Sin3A, and Hdac2) with Pu.1 in Runx1-deficient cells. (A) Increased co-IP of endogenous corepressor with Pu.1 from PUER shRunx1 compared with PUER empty vector (control) cells. Input indicates nonimmunoprecipitated cell lysate; IgG, control IP with isotype antibody; C, PUER control; and shR, PUER shRunx1. Cells were lysed 4 hours after addition of OHT. (B) Runx1 protein expression is decreased in primary Runx1 haploinsufficient (Runx1^+/−) compared with wild-type littermate control (WT) bone marrow cells. Mice were genotyped as previously described. (C) Co-IP of endogenous corepressor with endogenous Pu.1 from Runx1^+/− compared with WT bone marrow. Lysates were generated from freshly harvested bone marrow.

Figure 3

The corepressors SIN3A and ETO2 are excluded from the RUNX1/PU.1 complex but not the RUNX1-ETO/PU.1 complex. (A) Increasing concentrations of PU.1 decrease the amount of ETO2 coimmunoprecipitated with RUNX1. 293T cells were transfected with expression vectors for flag-RUNX1, HA-ETO2, and PU.1 in increasing amounts. In the absence of PU.1, ETO2 co-IPs with RUNX1 (lane 3). As PU.1 amounts are increased, PU.1 preferentially co-IPs with the RUNX1 and ETO2 is progressively excluded (lanes 6, 9, and 12). (B) Increasing concentrations of PU.1 decrease the amount of endogenous SIN3A coimmunoprecipitated with RUNX1. Sin3A co-IPs with RUNX1 in the absence of PU.1 (lane 3). As PU.1 concentrations are increased, PU.1 preferentially co-IPs with the RUNX1 and Sin3A is progressively excluded (lanes 6, 9, and 12). (C) Addition of PU.1 to RUNX1-ETO increases ETO2 recruitment. 293-T cells were transiently transfected with expression vectors for flag-RUNX1-ETO, ETO2, and increasing amounts of PU.1. Increasing amounts of PU.1 increased the amount of ETO2 in the RUNX1-ETO/PU.1 complex (lanes 6, 9, and 12). (D) Addition of PU.1 to RUNX1-ETO increases SIN3A recruitment. 293-T cells were transiently transfected with expression vectors for flag-RUNX1-ETO and increasing amounts of PU.1. Increasing amounts of PU.1 increased the amount of SIN3A in the RUNX1-ETO/PU.1 complex.

Figure 4

The RUNX1 C-terminus is required to exclude corepressors from the RUNX1/PU.1 complex. (A) Increasing concentrations of RUNT did not prevent the co-IP of ETO2 or SIN3A with PU.1. 293T cells were transfected with HA-ETO2, Flag-PU.1, and increasing amounts of HA-RUNT. Increasing amounts of RUNT did not decrease co-IP of ETO2 or endogenous SIN3A with PU.1 (lanes 5-8). The vertical line separates repositioned gel lanes (lanes 9-11) that showed anti-Flag did not IP HA-RUNT, HA-ETO2, or SIN3A in the absence of Flag-PU.1. Dashed outline shows expected position of protein in these lanes. (B) In the reverse co-IP experiment, increasing amounts of RUNT did not prevent the co-IP of PU.1 with ETO2. 293T cells were transfected with HA-ETO2, PU.1, and increasing amounts of Flag-RUNT. Increasing amounts of RUNT did not decrease the co-IP of PU.1 with ETO2 (lanes 5-8). The vertical line separates repositioned gel lanes (lanes 9 and 10), which showed that anti-HA did not immunoprecipitate PU.1 or Flag-RUNT in the absence of HA-ETO2. Dashed outline shows expected position of protein in these lanes. (C) In contrast, increasing amounts of RUNX1 decreased co-IP of PU.1 with ETO2. 293T cells were transfected with expression vectors for HA-ETO2, PU.1, and flag-RUNX1. PU.1 co-IPs with ETO2 (lane 5). Increasing amounts of RUNX1 decreased co-IP of PU.1 (lanes 6-8). The vertical line separates repositioned gel lanes (lanes 9 and 10) that showed anti-HA did not IP PU.1 or Flag-RUNX1 in the absence of HA-ETO2. Dashed outline indicates expected position of protein in these lanes.

Figure 5

Inhibition of HDAC rescues MCSFR and GMCSFR expression in AML cells that express lineage-specifying TFs and contain mutated or translocated RUNX1. Kasumi-1 cells contain RUNX1-ETO. CG-SH cells contain mutated RUNX1 and normal cytogenetics. (A) AML cells containing abnormal RUNX1 express high levels of the lineage-specifying TFs PU.1 and CEBPA. Western blot 72 hours after addition of drug. (B-C) Treatment with HDAC inhibitors significantly increased MCSFR and GMCSFR mRNA and protein expression. Kasumi-1 and CG-SH were treated once with SAHA 1μM or MS-275 0.5μM. mRNA expression was measured by quantitative RT-PCR 72 hours after HDAC inhibitor treatment. Data are mean ± SD. HDAC inhibitor treatment versus vehicle treatment: *P < .05, **P < .01 (Student t test). Protein expression was measured by flow cytometry. Experiments were performed in triplicate.

Figure 6

Model for RUNX1 regulation of PU.1 interaction with corepressors. Proposed model for RUNX1-mediated regulation of the transcriptional activity of PU.1 in normal hematopoiesis, and effects of RUNX1 deletions and translocations on this cooperation. RUNX1 and PU.1 independently interact with SIN3A or ETO2 (rows 1 and 2). When both RUNX1 and PU.1 are present, ETO2 and SIN3A are excluded from the RUNX1:PU.1 complex (row 3). Corepressor exclusion requires the RUNX1 C-terminus. Therefore, RUNT does not inhibit PU.1 interaction with corepressor (row 4). Similarly, corepressor is recruited to the RUNX1-ETO/PU.1 complex (RUNX1-ETO lacks the C-terminus). In this complex, corepressor is recruited to both the ETO moiety of the leukemia fusion protein and to PU.1 (row 5).