Direct competition between DNA binding factors highlights the role of Krüppel-like Factor 1 in the erythroid/megakaryocyte switch

Abstract

The Krüppel-like factor (KLF) family of transcription factors play critical roles in haematopoiesis. KLF1, the founding member of the family, has been implicated in the control of both erythropoiesis and megakaryopoiesis. Here we describe a novel system using an artificial dominant negative isoform of KLF1 to investigate the role of KLF1 in the erythroid/megakaryocytic switch in vivo. We developed murine cell lines stably overexpressing a GST-KLF1 DNA binding domain fusion protein (GST-KLF1 DBD), as well as lines expressing GST only as a control. Interestingly, overexpression of GST-KLF1 DBD led to an overall reduction in erythroid features and an increase in megakaryocytic features indicative of a reduced function of endogenous KLF1. We simultaneously compared in vivo DNA occupancy of both endogenous KLF1 and GST-KLF1 DBD by ChIP qPCR. Here we found that GST-KLF1 DBD physically displaces endogenous KLF1 at a number of loci, providing novel in vivo evidence of direct competition between DNA binding proteins. These results highlight the role of KLF1 in the erythroid/megakaryocyte switch and suggest that direct competition between transcription factors with similar consensus sequences is an important mechanism in transcriptional regulation.

Figure 1

The GST-KLF1 DBD construct is capable of binding to the canonical KLF1 binding site and has a biological effect in MELs. (A) Schematic demonstrating the experimental design of the two different constructs nucleofected into MEL cells. (B) Western blots from MEL lines showing protein expression of GST and GST-KLF1 DBD for each clone. Actin is presented as a loading control. (C) EMSAs from representative MEL clones indicating GST-KLF1 DBD is capable of binding the canonical KLF1 consensus sequence (β-globin CACCC probe) in vitro. (D) Western blots from MEL clones with an antibody that recognises both endogenous KLF1 and GST-KLF1 DBD. β-Actin is presented as a loading control. (E) Western blots from MEL clones with an antibody that recognises KLF3, indicating reduced expression of endogenous KLF3 in cells overexpressing GST-KLF1 DBD compared to GST only clones. β-Actin is presented as a loading control. n = 4 for each construct.

Figure 2

MEL cells overexpressing GST-KLF1 DBD display a less pronounced erythrocyte phenotype. (A) Representative flow cytometry plots of MEL + GST and MEL + GST-KLF1 DBD cells stained with antibodies against CD71 and TER119. (B) Histogram of flow cytometry for TER119 in a representative GST and GST-KLF1 DBD clone. (C) Statistical analysis of flow cytometry data for erythrocyte markers from all clones (D) MEL clones induced with DMSO. (E) Absolute levels of globin expression were analysed by qPCR in MEL + GST expressing clones compared to MEL + GST-KLF1 DBD expressing clones induced with DMSO. Expression levels were normalised to 18S rRNA. n = 4 for each construct. Error bars shown represent standard error of the mean, p values indicate the difference between the means, *p < 0.05, **p < 0.01 (paired Student’s two-tailed t test).

Figure 3

MEL cells overexpressing GST-KLF1 DBD display a more pronounced megakaryocyte phenotype. (A) Representative flow cytometry plots of MEL + GST and MEL + GST-KLF1 DBD cells stained with antibodies against CD42a and CD41. (B) Histogram of flow cytometry for CD41 in a representative GST and GST-KLF1 DBD clone. (C) Statistical analysis of flow cytometry data for erythrocyte markers from all clones. (D) Absolute levels of megakaryocyte related gene expression were analysed by real time qPCR in un-induced GST expressing clones compared to GST-KLF1 DBD expressing clones. Expression levels were normalised to 18S rRNA. n = 4 for each construct. Error bars shown represent standard error of the mean, p values indicate the difference between the means, *p < 0.05, **p < 0.01 (paired Student’s two-tailed t test).

Figure 4

The GST-KLF1 DBD protein displaces endogenous KLF1 at a number of KLF1 target genes in vivo. (A) Schematic depicting the experimental design of the ChIP experiment. (B) ChIPs were performed using α-KLF1, α-GST and α-IgG in both in MEL + GST and MEL + GST-KLF1 DBD clones, (n = 4 for each group per IP). Data are represented as the fold-change enrichment of IP over input, where the negative control for each IP, the Klf1 promoter for GST, and the Klf8 +33 kb locus for both KLF1 and IgG, was set to 1. The Klf3 1b and Klf8 promoters have been included as positive controls while the Klf1 promoter, and Klf3 +10 kb and Klf8 +33 kb genomic regions are shown as negative controls. Error bars shown represent standard error of the mean, p values indicate the difference between the means, *p < 0.05, **p < 0.01 (paired Student’s two-tailed t test).

Figure 5

GST-KLF1 DBD can act as a dominant negative at some KLF1 target sites to displace endogenous KLF1 binding, thereby inhibiting DNA activation. (A) In the absence of GST-KLF1 DBD, endogenous KLF1 is able to bind and activate its target genes normally. (B) At some KLF1 target sites, for instance the Klf3 1b promoter, the GST-KLF1 DBD protein is able to out-compete endogenous KLF1, either by sheer amount of protein present, or a higher affinity for the site. This results in reduced activation of these target genes. (C) At some sites, GST is unable to bind due to it not being able to interact efficiently with other KLF1 partners and SRC family members. In this case, activation of KLF1 target genes in unaffected.