Chromatin Remodeling of the Mouse AHSP Gene Requires EKLF.

Alpha-Hemoglobin Stabilizing Protein (AHSP) is an erythroid-specific protein that binds to α-globin, preventing precipitation of α-hemoglobin tetramers. Our interest has focused on how the AHSP gene is specifically expressed in erythroid cells, and we have investigated the roles of cis acting DNA sequences, the transcription factor EKLF and chromatin structure on AHSP gene expression. We have previously shown that the AHSP gene has a single mRNA initiation site followed by a non-coding exon. Putative promoter sequences from −904, −479 or −170 were active in luciferase reporter assays only when the constructs contained the downstream 269 bp containing exon 1 and intron 1. The −904/+269, −479/+269 and −170/+269 constructs gave 53.3+2.0 to 122.1+8.8 -fold increased levels of luciferase expression in K562 cells compared to plasmids without exon 1 and intron 1 (p<0.001; Gallagher et al., BLOOD 102, 267a, 2003). In vitro DNase I footprinting and EMSA assays revealed two regions (−75 to −67 and +153 to +164) that bound GATA-1. Analysis of 5 lines of transgenic mice with between 1 and 11 copies of the −170/+269 promoter fused to the human γ-globin gene demonstrated position independent expression (5/5 lines express) of γ-globin mRNA, at levels that were 4.6% the level of mouse α-globin mRNA per transgene copy. There was no correlation between transgene copy number and the level of γ-globin mRNA and 3/5 lines exhibited variegated expression. We concluded that sequences upstream of −170 or downstream of +269 are required for authentic expression from the AHSP promoter. To examine the role of EKLF in AHSP expression, we used subtractive hybridization, microarray and RNase protection analysis to compare AHSP mRNA levels in fetal liver cells from wild type and EKLF−/− mice. EKLF −/− fetal liver cells had 9 -fold less AHSP mRNA than wild type fetal liver cells, which we propose would exacerbate the moderate β- thalassemia described in EKLF −/− mice. Based on the observation that EKLF associates with an erythroid chromatin remodeling complex (Armstrong et al. Cell 95, 93–104, 1998), we hypothesized that EKLF was involved in chromatin remodeling at the AHSP locus. We assayed for DNase I Hypersensitive sites (HS) in chromatin from 13.5 day wild type and EKLF−/− mouse fetal liver nuclei. We demonstrated a strong DNase I HS between −200 and −400 of the AHSP gene, just upstream of the minimal −170 promoter, that was not present in chromatin from EKLF−/− fetal liver cells. To examine histone acetylation across the 3.5 kb AHSP locus we performed Chromatin Immune Precipitation analysis on wild type and EKLF −/− fetal liver chromatin using 13 different primer pairs (~300 bp intervals). In wild type chromatin there are two regions where histones H3 and H4 were hyperacetylated relative to a control region from the mouse α-globin gene promoter. The 5′ region corresponded to the DNase I HS at −400 to −200, while the second region maps 3′ to the AATAAA signal in the AHSP gene. Histones H3 and H4 were also acetylated in the interval between the hyperacetylated regions, while the chromatin upstream and downstream (~1 kb in each direction) of these regions was hypoacetylated. In contrast, all sites were hypoacetylated in chromatin from EKLF−/− fetal liver cells, correlating with the severe reduction in AHSP gene expression. We conclude that EKLF is required for remodeling the chromatin of the AHSP locus and that EKLF could be a modifier gene for the thalassemia syndromes.