Erythroid Krüppel-like factor directly activates the basic Krüppel-like factor gene in erythroid cells

Abstract

The Sp/Krüppel-like factor (Sp/Klf) family is comprised of around 25 zinc finger transcription factors that recognize CACCC boxes and GC-rich elements. We have investigated basic Krüppel-like factor (Bklf/Klf3) and show that in erythroid tissues its expression is highly dependent on another family member, erythroid Krüppel-like factor (Eklf/Klf1). We observe that Bklf mRNA is significantly reduced in erythroid tissues from Eklf-null murine embryos. We find that Bklf is driven primarily by two promoters, a ubiquitously active GC-rich upstream promoter, 1a, and an erythroid downstream promoter, 1b. Transcripts from the two promoters encode identical proteins. Interestingly, both the ubiquitous and the erythroid promoter are dependent on Eklf in erythroid cells. Eklf also activates both promoters in transient assays. Experiments utilizing an inducible form of Eklf demonstrate activation of the endogenous Bklf gene in the presence of an inhibitor of protein synthesis. The kinetics of activation are also consistent with Bklf being a direct Eklf target. Chromatin immunoprecipitation assays confirm that Eklf associates with both Bklf promoters. Eklf is typically an activator of transcription, whereas Bklf is noted as a repressor. Our results support the hypothesis that feedback cross-regulation occurs within the Sp/Klf family in vivo.

FIG. 1.

Bklf protein is diminished in Eklf-null murine fetal liver. (A) EMSA showing the identity of proteins found in erythroid cells which recognize the β-major globin promoter CACCC box. MEL cell nuclear extracts were allowed to bind to a radiolabeled probe containing the β-major globin promoter CACCC box in the presence of either preimmune serum or an antibody specific for Eklf, Bklf, Sp1, and/or Sp3. (B) EMSA analysis of murine fetal liver nuclear extracts prepared from E14.5 Eklf-null and heterozygous (denoted below each lane) littermates. Lane 1 contains the β-major globin CACCC probe alone. Bklf and Eklf bands were confirmed by antibody supershifting. All lanes that lack either Eklf- or Bklf-specific antibody (αEklf or αBklf) contain preimmune serum as a control.

FIG. 2.

Characterization of the transcriptional start sites of the alternative exons of Bklf. (A) 5′ RACE from MEL cDNA generates two distinct bands which represent transcripts containing different lengths of the novel Bklf exon 1b. (B) Primer extension of Bklf mRNA using a primer specific for Bklf exon 1a. Lanes 1 and 2 contain primer-extended total RNA from MEL cells and NIH 3T3 cells, respectively. Lane 3 contains yeast tRNA as a negative control. Arrows indicate the two major start points of transcription for Bklf exon 1a. (C) A schematic of the murine Bklf locus showing the relative positions of the two alternative first exons, 1a and 1b. Exons are denoted by gray boxes. The ATG start codon is also shown.

FIG. 3.

The sequence of Bklf exon 1a (A) and exon 1b (B) and their flanking genomic regions. Note that approximately 8.3 kb of intron 1 has been omitted. Exons are written in boldface type, and their transcriptional start sites, as determined by results shown in Fig. 2A and B, are denoted by triangles. For each exon, the 5′-most position of these start sites has been designated position +1. For exon 1b, an upstream transcriptional start site that was infrequently detected by 5′ RACE cloning is indicated with a boldface underline. For exon 1a, other putative transcription start sites are underlined. These transcription start sites were inferred from full-length cDNA entries in GenBank (DDBJ accession numbers AK007959 [42], AK157576, AK143838, and AK010713 [14, 41]; GenBank accession numbers BC116938 and BC119214 [77]) and from queries of the DBTSS (clone names BY338164, BY724103, BY707959, BY742654, BY732683, BY732370, BY188922, BY062971, and BB628801 [45, 80]). Motifs that precisely fit the Eklf binding consensus of 5′-NCNCNCCC-3′ (or its reverse complement) are boxed. The putative Gata-binding site in promoter 1b is underlined.

FIG. 4.

The downstream Bklf promoter is active predominantly in hematopoietic tissue. The levels of the two Bklf transcripts in murine tissues were determined by quantitative real-time RT-PCR. (A) Levels of Bklf transcripts generated from the upstream promoter and containing exon 1a. (B) Levels of Bklf mRNA arising from the downstream promoter and containing exon 1b. All values were normalized to 18S rRNA levels. WAT, white adipose tissue.

FIG. 5.

Both Bklf promoters are less active in E14.5 Eklf^−/− murine fetal liver. Analysis of the two Bklf transcripts in E14.5 littermate Eklf^+/+, Eklf^+/−, and Eklf^−/− fetal livers was performed by quantitative real-time RT-PCR. (A) Bklf exon 1a transcripts. (B) Bklf exon 1b transcripts. (C) Total Bklf transcripts: Bklf-specific primers were designed to amplify a region spanning the exon 4/5 boundary. Bklf mRNA levels were normalized to 18S rRNA levels.

FIG. 6.

Eklf transactivation assays of the Bklf promoters in SL-2 cells. Zero, 10, 50, or 100 ng pPac-Eklf (supplemented to 100 ng with pPac) was cotransfected with 1 μg pGL4.10[luc2] reporter vector containing fragments of Bklf promoter 1a (−189 to +89 or −532 to +89) (A) or varied lengths (−225 to +61, −129 to +61, −121 to +61, or −35 to +61) of Bklf promoter 1b (B). Firefly luciferase activity has been expressed as activation above that observed for cells transfected with each reporter vector and pPac alone. (A and B) Cell lysates were assayed for firefly luciferase activity 48 h posttransfection. Reporter activity has been normalized to Renilla luciferase levels. *, P < 0.05 compared to pGL4.10[luc2]-empty cotransfected with 10, 50, or 100 ng pPac-Eklf as appropriate (paired Student's t tests). **, P < 0.0005 and ***, P < 0.05 compared to pGL4.10[luc2]-Bklfprom1b(−129+61) cotransfected with 50 ng and 100 ng pPac-Eklf, respectively (paired Student's t tests). (C) EMSA analysis demonstrating that Eklf binds in vitro a CACCC box present in Bklf promoter 1b. Nuclear extracts were obtained from mock-transfected (lane 1) and Eklf-overexpressing (lanes 2 to 3) COS cells. Nuclear extracts were allowed to bind to a radiolabeled probe containing the Bklf promoter 1b CACCC box in the presence of either anti-Eklf (lane 3) or preimmune serum (lanes 1 and 2). RLU, relative light units.

FIG. 7.

Bklf exon 1b and Klf10 transcripts are upregulated in the presence of Eklf. (A) Klf mRNA levels in B1.6 cells that were harvested 48 h following induction with either 100 nM tamoxifen or with ethanol (0.0001%) as a control. (B) Klf10 transcript levels in Eklf^+/+, Eklf^+/−, and Eklf^−/− fetal livers. All mRNA levels were determined by quantitative real-time RT-PCR and normalized to 18S rRNA levels.

FIG. 8.

Bklf and Klf10 are directly activated by Eklf-ER. (A and B) B1.6 cells were exposed to cycloheximide (CHX) 30 min before induction with tamoxifen and were harvested for RNA 8 h thereafter. Bklf exon 1b levels (A) and Klf10 mRNA levels (B) were determined by quantitative real-time RT-PCR. (C and D) B1.6 cells were harvested at numerous time points after tamoxifen induction in the absence of cycloheximide. Again, Bklf exon 1b levels (C) and Klf10 mRNA levels (D) were determined by quantitative real-time RT-PCR. In all cases, 18S rRNA levels were used for normalization. *, **, and ***, P < 0.05 (paired Student's t test, compared to ethanol control or 0-h time point as appropriate).

FIG. 9.

Eklf ChIP at the two Bklf promoters. ChIP material was extracted from B1.6 cells that were induced with tamoxifen or ethanol (control). Chromatin was immunoprecipitated using an Eklf antibody (denoted by “+”) or preimmune serum (−). Primers used for real-time PCR quantification were targeted against Bklf promoter 1a (A), Bklf promoter 1b (B), a control region 10 kb upstream of the Bklf locus (C), the nectin promoter (D), and the β-major globin promoter (E).