The distinctive roles of erythroid specific activator GATA-1 and NF-E2 in transcription of the human fetal γ-globin genes

Abstract

GATA-1 and NF-E2 are erythroid specific activators that bind to the β-globin locus. To explore the roles of these activators in transcription of the human fetal stage specific γ-globin genes, we reduced GATA-1 and p45/NF-E2 using shRNA in erythroid K562 cells. GATA-1 or p45/NF-E2 knockdown inhibited the transcription of the γ-globin genes, hypersensitive site (HS) formation in the LCR and chromatin loop formation of the β-globin locus, but histone acetylation across the locus was decreased only in the case of GATA-1 knockdown. In p45/NF-E2 knockdown cells, GATA-1 binding was maintained at the LCR HSs and γ-globin promoter, but NF-E2 binding at the LCR HSs was reduced by GATA-1 knockdown regardless of the amount of p45/NF-E2 in K562 cells. These results indicate that histone acetylation is dependent on GATA-1 binding, but the binding of GATA-1 is not sufficient for the γ-globin transcription, HS formation and chromatin loop formation and NF-E2 is required. This idea is supported by the distinctive binding pattern of CBP and Brg1 in the β-globin locus. Furthermore GATA-1-dependent loop formation between HS5 and 3'HS1 suggests correlation between histone modifications and chromatin looping.

Figure 1.

Knockdown of GATA-1 and p45/NF-E2 in erythroid K562 cells. (A) The human β-globin locus is presented. Vertical black arrows indicate DNase I HSs in LCR. The exons of the globin genes are represented by squares. Vertical bars named below the diagram denote the locations of TaqMan amplicons used in real-time PCR. (B) Western blotting was performed using antibodies specific to GATA-1 or p45/NF-E2 in protein extract from K562 cells expressing a control, GATA-1 or p45/NF-E2 shRNA. Blotting with β-tubulin antibody was used an experimental control. ChIP was performed with GATA-1 (C) or p45/NF-E2 (D) antibodies in K562 cells expressing each shRNA. Relative intensity was determined by quantitatively comparing input with immunoprecipitated DNA for the indicated amplicons. Actin served as negative control. Normal rabbit and goat IgG (Con IgG) served as experimental control. The results of two to four independent experiments ± SEM are graphed.

Figure 2.

Transcription of the globin genes in GATA-1 or p45/NF-E2 knockdown K562 cells. (A) cDNA was prepared from RNA isolated from K562 expressing each shRNA and then amplified by real-time PCR using primers and probes for exons of the ε-, γ-, δ- and β-globin genes. Transcript levels of the globin genes were compared to transcript levels of the Actin control gene. The level of the ε-globin gene was multiplied by five to show changes in the knockdown cells. The results of two to four independent experiments ± SEM are graphed. (B) ChIP was performed with antibody specific to RNA polymerase II in K562 cells expressing each shRNA. Relative intensity was determined by quantitatively comparing immunoprecipitated DNA with input for the indicated γ-globin gene amplicons and then normalizing to the intensity at the Actin. Normal rabbit IgG (Con IgG) served as experimental control. The results of two to four independent experiments ± SEM are graphed.

Figure 3.

Histone modification of the β-globin locus in the GATA-1 or p45/NF-E2 knockdown cells. ChIP was performed with antibodies specific to H3K9/K14ac (A), H3K27ac (B), H3K4me2 (C), H3K4me3 (D), H3K27me2 (E) and H3K27me3 (F) in K562 cells expressing each shRNA. Relative intensity was determined by normalizing to the intensity at the Actin as described in Figure 2. (G) The relative intensity of CBP ChIP was determined as described in Figure 1. Normal rabbit IgG (Con IgG) served as experimental control. The results of two to four independent experiments ± SEM are graphed.

Figure 4.

DNase I sensitivity at the β-globin LCR HSs, γ-globin promoter and 3′HS1 in the GATA-1 or p45/NF-E2 knockdown cells. (A) Nuclei of K562 cells expressing each shRNA were digested with 150–600 U DNase I. DNA extracted from the digest was run on 1% agarose gel. (B) DNase I sensitivity was determined by quantitatively comparing digested DNA with undigested DNA and normalizing to the sensitivity at the Actin gene at each concentration. The results are averages of two to four nuclei preparations ± SEM. (C) ChIP was performed using Brg1 antibody described in Figure 1.

Figure 5.

Relative proximity between HSs and the γ-globin gene in the β-globin locus in the GATA-1 or p45/NF-E2 knockdown cells. The 3C assay was performed with HindIII restriction enzyme. (A) HindIII sites and PCR primers in the β-globin locus were represented by vertical bars and horizontal arrows, respectively. The black shading represents the anchor fragment for HS5 (B). 3′HS1 (C). HS2 (D). HS1 (E) and ^Gγ-globin gene (F) in PCR. The gray shadings are fragments generated by HindIII digestion. Relative cross-linking frequency was determined by quantitatively comparing ligated DNA in cross-linked chromatin with control DNA and then normalizing to the cross-linking frequency at the ERCC gene. The results are averages of four to six independent experiments ± SEM. (G) ChIP was performed using CTCF antibody described in Figure 1.

Figure 6.

Chromatin structure in the GATA-1 knockdown cells with restoration of p45/NF-E2 protein. (A) Western blotting was performed using antibodies specific to GATA-1 or p45/NF-E2 in protein extract from K562 cells where co-transduction of control shRNA and control expression vector (Con⁺) or co-transduction of GATA-1 shRNA and p45/NF-E2 cDNA expression vector (NF⁺⁾ was performed. ChIP was performed with GATA-1 or p45/NF-E2 antibodies in Con⁺ and NF⁺ cells and analyzed as described in Figure 1. (B) RT–PCR and ChIP using antibody specific RNA polymerase II were performed in Con⁺ and NF⁺ cells and analyzed as described in Figure 2. (C) ChIP was performed with antibodies specific to H3K27ac and H3K27me2 in Con⁺ and NF⁺ cells and analyzed as described in Figure 2. (D) DNase I sensitivity was measured in Con⁺ and NF⁺ cells as described in Figure 4 and ChIP was performed with Brg1 antibody and analyzed as described in Figure 1. (E) The 3C assay was performed with HindIII restriction enzyme in Con⁺ and NF⁺ cells and fragment containing the ^Gγ-globin gene was used as anchor. Relative cross-linking frequency was determined as described in Figure 5. ChIP was performed with CTCF antibody and analyzed as described in Figure 1.

Figure 7.

Summary of the changes by GATA-1 or p45/NF-E2 knockdown in the human fetal β-globin locus in K562 cells. The results presented in Figures 1–6 were combined and represented in model of the human β-globin locus.