Mammals express a variety of erythroid and nonerythroid ankyrin-1 isoforms generated by alternate mRNA splicing and by expression from promoters upstream of 3 known alternate first exons. ANK-1 Exon 1B is located 138 kb 5′ of exon 2 and is expressed only in neuronal and muscle cells. Exon 1E is located 39 kb 5′ of exon 2 and is expressed only in erythroid cells. Exon 1A is located 27 kb 5′ of exon 2 and is expressed in many cell types. We have previously shown that the ANK-1E promoter is flanked by DNase I Hypersensitive Sites (HS), one immediately upstream of the RNA initiation sites (5′HS) and a pair of closely spaced HS 5kb downstream (3′HS1 and 3′HS2). To determine the location of additional HS in the ANK-1 locus, we designed PCR primers spaced ~250 bp apart that span a 200 kb region from exon 2 to 60 kb upstream of Exon 1B for use in a high throughput DNase I HS assay. We identified the HS surrounding Exon 1E, as well as HS that flank Exons 1A and 1B. In both the Exon 1A and Exon 1B promoters, the 5′HS is located immediately upstream of the mRNA initiation sites and the 3′HS is located 4–7 kb downstream. An additional pair of HS were identified 70 kb upstream of exon 2 between the ANK-1B and ANK-1E promoters. This region contains the 5′ ends of at least 5 human ESTs. We used 5′ RACE to show that the homologous region in the mouse is transcribed and splices to exon 2. This putative promoter is designated ANK-1C. All 4 ANK-1 promoters lack consensus promoter sequences involved in the binding of the transcription initiation complex, TFIID, including TATA, InR, DPE or DCE elements. We have recently identified a novel consensus sequence that binds TFIID: (T/G)(G/C)(G/C)GGTGAG. This sequence is present multiple times in all 4 ANK-1 promoters as well as in 22% of >4000 mammalian promoters lacking TFIID-binding consensus sequences, strong evidence of functional significance. To understand the relationship of the flanking HS to the ANK-1 promoters we used the activation of ANK-1E promoter in erythroid cells as a model and have undertaken a molecular dissection of the elements in the ANK-1E region. Using transgenic mice and K562-based assays we have shown that both ANK-1E 5′HS and 3′HS2 are barrier elements that prevent gene silencing. In K562 cells, ANK-1E 3′HS1 increases expression only when located adjacent to the ANK-1E or thymidine kinase promoters (p=0.0009), but not in SY5Y neuronal cells (p=0.35). DNase I footprinting, gel shift, and reporter gene assays demonstrated that 3′HS1 binds the erythroid-specific transcription factor NF-E2. Mutation of the NF-E2 binding site abolished the ability of 3′HS1 to increase gene expression (p=0.08) in K562 cells. Chromatin Conformation Capture (3C) analysis demonstrated the formation of a 5 kb erythroid-specific chromatin loop that brings 5′HS into close proximity with 3′HS1/2. In agreement with the 3C results, Chromatin Immune Precipitation analysis demonstrated a hub in which Brg-1 and CTCF (associated with barrier elements), NF-E2, GATA-1 and RNA Pol II occupy both 5′HS and 3′HS1/2, despite the lack of consensus sites for NF-E2 in 5′HS, or GATA-1 or RNA initiation sites in 3′HS1/2. Our current model is that the formation of the erythroid-specific ANK-1E chromatin loop is mediated by the binding of the erythroid-specific transcription factors GATA-1 to 5′HS and NF-E2 to 3′HS1.

Disclosure: No relevant conflicts of interest to declare.

Author notes


Corresponding author