Distinct modes of gene regulation by a cell-specific transcriptional activator

Abstract

The architectural layout of a eukaryotic RNA polymerase II core promoter plays a role in general transcriptional activation. However, its role in tissue-specific expression is not known. For example, differing modes of its recognition by general transcription machinery can provide an additional layer of control within which a single tissue-restricted transcription factor may operate. Erythroid Kruppel-like factor (EKLF) is a hematopoietic-specific transcription factor that is critical for the activation of subset of erythroid genes. We find that EKLF interacts with TATA binding protein-associated factor 9 (TAF9), which leads to important consequences for expression of adult beta-globin. First, TAF9 functionally supports EKLF activity by enhancing its ability to activate the beta-globin gene. Second, TAF9 interacts with a conserved beta-globin downstream promoter element, and ablation of this interaction by beta-thalassemia-causing mutations decreases its promoter activity and disables superactivation. Third, depletion of EKLF prevents recruitment of TAF9 to the beta-globin promoter, whereas depletion of TAF9 drastically impairs beta-promoter activity. However, a TAF9-independent mode of EKLF transcriptional activation is exhibited by the alpha-hemoglobin-stabilizing protein (AHSP) gene, which does not contain a discernable downstream promoter element. In this case, TAF9 does not enhance EKLF activity and depletion of TAF9 has no effect on AHSP promoter activation. These studies demonstrate that EKLF directs different modes of tissue-specific transcriptional activation depending on the architecture of its target core promoter.

Fig. 1.

EKLF interacts with TAF9 in vivo. (A) Full-length or various deletion constructs (as indicated) of HA-TAF9 were cotransfected with full-length EKLF into 293T cells. Whole cell lysates were prepared and immunoprecipitation was performed with anti-EKLF mAb (7B2) or control isotype IgG as indicated on the left. The immunoprecipitates were subjected to immunoblot analysis with anti-HA antibody to detect TAF9 and reprobed with anti-EKLF antibody to confirm successful immunoprecipitation (indicated on the right). Input represents 5% of the whole cell lysate used for immunoprecipitation. The asterisk denotes the specific EKLF band. (B) MEL cells were grown in the presence of 1.5% DMSO for varying lengths of time (days) to induce hemoglobin expression as confirmed by real-time RT-PCR of adult β-globin RNA. Extracts were immunoprecipitated (IP) with 7B2 EKLF antibody or control IgG and analyzed by Western blotting using anti-TAF9 or anti-EKLF antibody. Input represents 10% of the lysates used for immunoprecipitation. The asterisk denotes the specific EKLF band. (C) Extracts from E13.5 murine fetal livers were immunoprecipitated with control IgG or anti-EKLF mAb (7B2) as indicated and probed by Western blot analysis using anti-TAF9. Input represents 5% of extract used for immunoprecipitation; molecular mass markers (in kDa) are on the left.

Fig. 2.

TAF9 superactivates EKLF-mediated transcriptional activity of the adult β-globin promoter. (A) K562 cells were transfected with a luciferase reporter containing the adult β-globin promoter together with expression plasmids for EKLF and TAF9; TAF9 levels were titrated up as indicated. (Inset) An anti-EKLF blot from the indicated samples to demonstrate its constant expression level. (B) K562 cells were transfected with the β-globin/luciferase reporter along with expression plasmids for TAF9 and where indicated with full-length and various deletion constructs of EKLF. In all cases, luciferase activity was normalized against Renilla activity from a cotransfected control vector. The relative luciferase activity reflects the values obtained in triplicates.

Fig. 3.

Interaction of TAF9 with β-globin promoter. (A) Alignment of β-globin gene sequences from different species. Conserved INI and DPE elements are boxed. Asterisks indicate the position of known β-thalassemic mutations (at +1, +22, and +33). (B) (Left) EMSA was carried out by using radiolabeled DNA oligonucleotides encompassing the initiator (INI), the DPE box (+20), or a downstream E-box (+60) after incubation with bacterial extracts containing overexpressed TAF9 protein. (Center) EMSA with TAF9 was carried out with a radiolabeled oligonucleotide encompassing the WT DPE box (+20) or the same oligonucleotide with a mutation at +22 or +33 as indicated. (Right) EMSA with TAF9 was carried out by using a radiolabeled fragment encompassing the WT DPE box (+20) after preincubation of the TAF9 extract with nonradioactive (cold) WT (+20) or β-thalassemic oligonucleotide (+22 or + 33) as indicated below. Asterisks indicate the specific TAF9/DNA complexes. (C) EMSA was performed with WT or mutant radiolabeled DPE oligonucleotides (as described in B) after incubation with an extract from MEL cells. Competition with cold oligonucleotides (shown below) was performed as for B. (D and E) K562 cells were transfected with the β-globin luciferase reporter containing WT or point mutated (+22 in D; +33 in E) promoters along with expression plasmids for EKLF and TAF9. Luciferase activity was normalized against Renilla activity from a cotransfected control vector. The relative luciferase activity reflects the values obtained in triplicates.

Fig. 4.

TAF9 does not superactivate EKLF activity on the AHSP promoter. K562 cells were transfected with either a β-globin or an AHSP luciferase reporter along with EKLF and TAF9 as indicated. (Inset) An anti-EKLF blot from the indicated samples demonstrates its constant expression level. Luciferase activity was normalized against Renilla activity from a cotransfected control vector. The relative luciferase activity reflects the values obtained in triplicates.

Fig. 5.

Binding of TAF9 to the endogenous β-globin promoter is mediated by EKLF. (A) Quantitative ChIP analysis of TAF9 occupancy on β-major globin promoter or a region 3 kb upstream was performed with anti-TAF9 antibody or control IgG in MEL cells without differentiation (day 0) or after DMSO-induced differentiation (day 4) as indicated. Standard error reflects values from triplicate experiments. (B) Western blot analysis of Dox-inducible knockdown of EKLF in a derivative of MEL cells. shEKLF represents an independent clone of MEL cells that expresses a hairpin against EKLF (38). Control (parental) MEL cells express only the TetR. Extracts were prepared from the indicated MEL cell lines before or after treatment with Dox for 2 days, and extracts were probed with anti-EKLF or anti–Erk1 (control) antibodies. (C) Total RNA was prepared from control (parental) or shEKLF MEL cell lines (as indicated) grown in the absence of Dox (−) or differentiation (−); after Dox treatment for 2 days (+2) as in B; after Dox treatment for 2 days followed by differentiation for 4 days with 1.5% DMSO (+6); or after differentiation alone (+4). Real-time RT-PCR was performed by using primers for EKLF and β-major globin. The average of duplicates from a single experiment that is representative of 2 is shown. (D) TAF9 occupancy at the β-major promoter was measured by quantitative ChIP using TAF9 or IgG antibodies on untreated (−) or Dox-treated (+) control (parental) or shEKLF MEL cells that were also differentiated for 4 days with DMSO. An average of 2 independent experiments each performed in triplicate are shown, with error bars representing standard deviation.

Fig. 6.

TAF9 is required for β-major globin, but not AHSP, promoter activity in MEL cells. MEL cells were transfected with TAF9 specific siRNAs (lanes 1–4) or scrambled siRNA (c, control). Cells were differentiated with 1.5% DMSO starting from 18 h posttransfection. Total RNA or cell extracts were prepared from cells collected after 3 days of differentiation. TAF9 and Erk1 protein levels were monitored by Western blot (A), and RNA was analyzed by quantitative RT-PCR for TAF9, EDEM, β-major globin, EKLF, GATA1, or AHSP expression using the indicated detection primers (B). The average of duplicates from a single experiment that is representative of 2 is shown.