Abstract
Krûppel-like factor-1 (KLF1) is an essential erythroid-specific transcription factor [1, 2]. A number of studies have shown up to ~700 genes are poorly expressed when KLF1 is absent [3-6]. This global loss of expression is responsible for failure of effective red blood cell production in KLF1 knockout mice, and partly responsible for congenital dyserythropoietic anemia type IV (CDA-IV) observed in humans with dominant mutations in the DNA-binding domain of KLF1 [7]. Recently an ENU-generated mouse model of neonatal anemia, ‘nan’, was also reported to harbour a mutation in the second zinc-finger of KLF1 [8]. Remarkably, the ‘nan’ mutation (E339D) resides at exactly the same amino acid which results in human CDA IV (= E325 in humans). Unlike loss of function point mutations in KLF1, this mutation leads to a more severe phenotype than the KLF1 null allele, suggesting it is an unusual dominant mutation [9].
To investigate how this mutation might cause disease, we introduced tamoxifen-inducible versions of KLF1 and KLF1nan into an erythroid cell line derived from Klf1-/- fetal liver cells [10]. We performed ChIP-seq to determine differences in genome occupancy in vivo, and identified novel sites occupied by EKLF-E339D but not by wild type KLF1. Using de novo motif discovery [11], we find KLF1nan binds a slightly degenerate CACC box element (CCMNGCCC) in comparison with wild type KLF1 (CCMCRCCC). This specificity is novel with respect to any known TFs, so we think it represents a sequence specificity not normally encoded in mammals. Ectopic binding to non-erythroid gene promoters is accompanied by aberrant gene expression as determined by 4sU labelling and deep sequencing of tamoxifen-induced primary nuclear RNAs. We find a 4-fold greater number of genes induced by KLF1-nan compared with wild type KLF1 which is consistent with degenerate genome occupancy. We compared the KLF1-nan dependent genes with RNA-seq performed in primary fetal liver for KLF1+/nan versus KLF1+/- mice. We confirmed aberrant binding using EMSA and surface plasmon resonance (SPR) using recombinant GST-Klf1 zinc finger domains expressed in E.coli. The degenerate motif is consistent with structural models of how the second zinc finger of KLF1 specifically interacts with its binding site [12, 13]. We are undertaking structural studies to confirm this modelling. Together RNA-seq, ChIP-seq and SPR studies have provided a novel explanation for how mutations in KLF1 result in dominant anemia in mice and man. To our knowledge this mechanism, whereby a transcription factor DNA-binding domain mutation leads to promiscuous binding, activation of an aberrant transcriptional program and subsequent derailing of co-ordinated differentiation, is novel.
References:
1.Perkins, A.C., A.H. Sharpe, and S.H. Orkin. Nature, 1995. 375(6529): p. 318-22.
2.Nuez, B., et al., Nature, 1995. 375(6529): p. 316-8.
3.Pilon, A.M., et al., Mol Cell Biol, 2006. 26(11): p. 4368-77.
4.Drissen, R., et al., Mol Cell Biol, 2005. 25(12): p. 5205-14.
5.Hodge, D., et al., Blood, 2006. 107(8): p. 3359-70.
6.Tallack, M.R., et al., Genome Res, 2012. 22(12):2385-98
7.Arnaud, L., et al., Am J Hum Genet. 87(5): p. 721-7.
8.Siatecka, M., et al., Proc Natl Acad Sci U S A. 2010. 107(34):15151-6
9.Heruth, D.P., et al., Genomics, 2010. 96(5): p. 303-7.
10.Coghill, E., et al., Blood, 2001. 97(6): p. 1861-1868.
11.Bailey, T.L., et al., Nucleic Acids Res, 2009. 37(Web Server issue): p. W202-8.
12.Schuetz, A., et al., Cell Mol Life Sci, 2011. 68(18): p. 3121-31.
13.Oka, S., et al., Biochemistry, 2004. 43(51): p. 16027-35.
Disclosures
No relevant conflicts of interest to declare.
Regions outside the DNA-binding domain are critical for proper in vivo specificity of an archetypal zinc finger transcription factor