GATA-1 and Oct-1 Are Required for Expression of the Human α-Hemoglobin-stabilizing Protein Gene*

α-Hemoglobin-stabilizing protein (AHSP) is an erythroid protein that binds and stabilizes α-hemoglobin during normal erythropoiesis and in pathological states of α-hemoglobin excess. AHSP has been proposed as a candidate gene in some Heinz body hemolytic anemias and as a modifier gene in the β-thalassemia syndromes. To gain additional insight into the molecular mechanisms controlling the erythroid-specific expression of the AHSP gene and provide the necessary tools for further genetic studies of these disorders, we have initiated identification and characterization of the regulatory elements controlling the human AHSP gene. We mapped the 5′-end of the AHSP erythroid cDNA and cloned the 5′-flanking genomic DNA containing the putative AHSP gene promoter. In vitro studies using transfection of promoter/reporter plasmids in human tissue culture cell lines, DNase I footprinting analyses and gel mobility shift assays, identified an AHSP gene erythroid promoter with functionally important binding sites for GATA-1- and Oct-1-related proteins. In transgenic mice, a reporter gene directed by a minimal human AHSP promoter was expressed in bone marrow, spleen, and reticulocytes, but not in nonerythroid tissues. In vivo studies using chromatin immunoprecipitation assays demonstrated hyperacetylation of the promoter region and occupancy by GATA-1. The AHSP promoter is an excellent candidate region for mutations associated with decreased AHSP gene expression.

type of ␤-thalassemia because of the cytotoxic effects of excessive free ␣-hemoglobin. In a murine model, AHSP deficiency leads to well-compensated hemolysis with Heinz body formation, reticulocytosis, and increased apoptosis of erythroid precursors (5). When AHSP-deficient mice were bred to ␤-thalassemia mice, loss of AHSP increased the severity of the thalassemia (5). These studies also suggested that AHSP might play a role in unexplained Heinz body hemolytic anemia in humans.
Identification and characterization of the regulatory elements that control AHSP gene expression have important implications for normal erythropoiesis and the pathogenesis of hemolytic disorders. AHSP may have therapeutic potential for ␤-thalassemia patients by decreasing ␣-globin precipitation and ameliorating clinical severity. Because AHSP is synthesized abundantly in erythroid cells, identification of the regulatory elements directing this high level, tissue-specific expression may provide important tools in directing other erythroid-specific genes in gene transfer applications.
AHSP is a highly expressed and erythroid-specific protein (1,7,8). It is present throughout primitive and definitive erythropoiesis of the mouse and accumulates to high levels in late erythroid cells. Its pattern of expression follows that of the ␣-globin gene throughout erythropoiesis (8). Identified as a target of the critical erythroid transcription factor GATA-1 (1), the AHSP gene is highly and rapidly induced by GATA-1.
In this report, we describe the identification and characterization of the human AHSP gene promoter. Our results demonstrate that the AHSP gene promoter requires GATA-1-and Oct-1-binding proteins to direct high level expression in erythroid cells in vitro. The minimal promoter, Ϫ170 to ϩ269, includes the 5Ј-flanking DNA and intron 1. In transgenic mice, a reporter gene directed by the AHSP minimal promoter directed expression exclusively in erythroid cells. The results demonstrate that the region from Ϫ170 to ϩ269 contains the sequences necessary for expression in erythroid cells and that this expression is dependent on GATA-1.

MATERIALS AND METHODS
Genomic Cloning-A human genomic library in PACs was screened by PCR amplification with oligonucleotide primers, 5Ј-GAGAT-TCACGCACCCTCAAGAGTGTG-3Ј (P s , sense) and 5Ј-GCA-GAACGCTGAACT CCTTCAATCC-3Ј (P A , antisense) (Fig. 1A), as described (9). These primers correspond to the 5Ј-end of the AHSP cDNA and amplify a 299-bp product from human genomic DNA.
Mapping the AHSP Gene Transcription Initiation Site-The transcription initiation site of the erythroid AHSP cDNA was determined using a primer extension assay as described with primer 5Ј-GGCTG-TATATGTCTCACCCACACTCTTG-3Ј and 20 g of total K562 or HeLa cell RNA or 1 g of tRNA (11).
Transient Transfection Analyses-AHSP promoter fragments Ϫ902/ ϩ32, Ϫ479/ϩ32, Ϫ170/ϩ32, Ϫ904/ϩ269, Ϫ479/ϩ269, and Ϫ170/ϩ269 were amplified by PCR using the human AHSP genomic DNA clone as template. These fragments were subcloned upstream of the firefly luciferase reporter gene in the plasmid pGL2B (Promega). Integrity of all test plasmids was confirmed by sequencing. Plasmids were purified using Qiagen columns (Qiagen, Inc., Chatsworth, CA), and at least two preparations of each plasmid were tested in triplicate. K562 and Hela cell transfections were performed as described (12).
Preparation of Nuclear Extracts-Nuclear extracts were prepared from K562 cells by hypotonic lysis followed by high salt extraction of nuclei as described by Andrews and Faller (13).
DNase I Footprinting in Vitro-The probes for DNase I footprinting were produced by PCR amplification of a subcloned AHSP genomic fragment Ϫ170/ϩ269 as template and a pair of oligonucleotide primers, 5Ј-GCGCTCGAGGGCTCTTGCCTTCTTGCATTTC-3Ј (sense) and 5Ј-CCGAAGCTTCTGGGTAGAGAAAAGGGTAGA-3Ј (antisense). One oligonucleotide was 5Ј-end-labeled with [ 32 P]ATP using polynucleotide kinase prior to use in PCR. Reaction mixes contained K562 cell nuclear extracts, 10,000 cpm of labeled probe, and 1 g of poly(dI-dC). After digestion with DNase I, samples were electrophoresed in 6% denaturing polyacrylamide gels, the gels dried, and subjected to autoradiography.
Electrophoretic Mobility Shift Analyses (EMSA)-Binding reactions were carried out as described (12). Competitor oligonucleotides were added at molar excesses of 100-fold. Antibodies to GATA-1 and Oct-1 were obtained from Santa Cruz Biotechnologies (Santa Cruz, CA). Reaction mixes were electrophoresed through 6% nondenaturing polyacrylamide gels in 0.5ϫ Tris-borate-EDTA at 21°C at 200 watts for 2 h. Gels were dried and subjected to autoradiography.
Preparation of Promoter Reporter Plasmids for Transgenic Mice-A 439-bp AHSP gene promoter fragment from Ϫ170/ϩ269 was ligated to a 2266-bp EcoRV/AatII fragment containing the human A ␥-globin gene (15) to generate AHSP/ A ␥. This plasmid construct was sequenced to confirm that the AHSP promoter was correctly fused to the ␥-globin gene. A 2348-bp AHSP promoter/ A ␥-globin gene fragment was excised from this plasmid with KpnI and HindIII for microinjection.
Generation and Analyses of Transgenic Mice-Transgenic mice were generated as described (15,16). Transgene copy number was determined by comparing the ␥-globin signals from Southern blot analysis of transgenic mouse and K562 DNA using a Molecular Dynamics Phos-phorImager. Statistical analysis of copy number and expression data were analyzed by linear regression using GraphPad Prism version 2.0 software.
Ribonuclease Protection Analysis of Transgene Expression-Expression of the human promoter/ A ␥-globin reporter transgene was analyzed using an RNase protection assay (RPA). Probe synthesis, hybridization, electrophoresis, autoradiography, and analyses were performed as described (12).
Expression of Human ␥-Globin Protein in the Erythrocytes of Transgenic Mice-Detection and measurement of ␥-globin protein in red blood cells was performed as described (15).

Cloning of Chromosomal Gene Isolation and Analysis of Recombinant
Clones-The PCR-based screening of the human genomic BAC library yielded one PCR-positive clone ϳ130 kb in length. DNA fragments that amplified with the screening primer pair were purified and subcloned into plasmid vectors. Restriction enzyme analysis, Southern blotting, and limited nucleotide sequencing identified the DNA fragments as AHSP gene-specific. Comparison of cDNA and genomic DNA sequences allowed determination of the genomic structure of the AHSP gene, a three exon gene spread over ϳ1 kb of DNA. The entire first exon and the 3Ј-end of the third exon contain untranslated sequence. The ag:gt rule is not violated at any splice junction.
Mapping the Human AHSP Erythroid mRNA Transcription Initiation Site and Identification of 5Ј-cDNA Sequences-To identify the 5Ј-end of the human AHSP cDNA, primer extension was performed using total RNA from K562 and HeLa cells. These experiments identified a single transcription initiation site in erythroid cells (Fig. 1B) and predicted the presence of an additional 32 bp in the mRNA upstream of the 5Ј-end of the previously reported sequence obtained from cDNA cloning. These additional 32 bp of upstream 5Ј-untranslated sequence were obtained by 5Ј-RACE. Sequences obtained by 5Ј-RACE were verified by comparison to corresponding genomic DNA sequences (Fig.  1C). The sequences around the transcription start site, ACA ϩ1 CTTG closely match transcription initiation recognition sequences, YYA ϩ1 NWYY (17). The first ATG is in exon 2; no ATGs are present in Relative luciferase activity was expressed as that obtained from the test plasmids versus the activity obtained from the promoterless plasmid pGL2B plasmid taking into account the transfection efficiency. The data are means Ϯ S.D. of at least six independent transfection experiments. The functionally important GATA-1 site at Ϫ98 to Ϫ104 is marked by a star. B, mutations in consensus DNA-protein binding sites, marked with an X, were introduced into the AHSP promoter/luciferase reporter plasmids and transfected into K562 cells and activity determined as described.

FIGURE 3. A, in vitro
DNase I footprinting of the human AHSP promoter. In vitro DNase I footprinting of the human AHSP gene promoter was performed using K562 extracts as described in the text. Two protected sites were identified, one in the 5Ј-flanking genomic DNA corresponding to a GATA-1 consensus binding site, and one in intron 1 corresponding to an Oct-1 consensus binding sequence. B, electrophoretic mobility shift assays of the GATA-1 site of the human AHSP gene promoter. Gel mobility shift assays using a probe corresponding to the GATA-1 consensus binding site of the human AHSP promoter identified by in vitro footprinting were performed using K562 nuclear extracts. Wild-type and mutant AHSP GATA-1 probes and a control GATA-1 probe were used. A GATA-1 antibody was added to the reaction mixtures where indicated.
the 5Ј-untranslated sequences. Taken together, these data suggest that this sequence is at or very near the 5Ј-end of the human AHSP erythroid cDNA.
The nucleotide sequence of the 5Ј-flanking genomic DNA upstream of the human AHSP cDNA transcription start site is shown in Fig. 3. Consensus TATA or CCAAT sequences are absent. Characteristic of an erythroid gene promoter, multiple consensus binding sequences for GATA-1 and a single Sp1/CACCC site are present.
An AHSP Gene Promoter Fragment Is Active in Erythroid Cells-Plasmids containing 5Ј-flanking putative AHSP promoter sequences linked to a luciferase reporter gene were transiently transfected into K562 cells. Relative luciferase activity was determined 48 h after transfection and compared with the activity obtained with pGL2B, a negative control, the promoterless plasmid, and pGL2P, a positive control, the luciferase reporter gene under control of the SV40 early promoter. Three AHSP promoter fragments, Ϫ904/ϩ32, Ϫ479/ϩ32, and Ϫ170/ϩ32, directed modest levels of luciferase activity over background compared with the control SV40 promoter plasmid ( Fig. 2A).
Similar to the human ␣-spectrin gene, GATA-1, and the erythropoietin receptor, where the addition of intron 1 sequences markedly increase promoter strength (14), we hypothesized that the addition of intron 1 would increase AHSP promoter expression. 234 bp of contiguous sequence encoding the 5Ј-untranslated region and intron 1 were added to each promoter fragment upstream of the luciferase reporter gene. Fragments Ϫ904/ϩ269, 479/ϩ269, and Ϫ170/ϩ269 all directed higher levels of luciferase activity in transfected K562 cells ( Fig. 2A). When transiently transfected into HeLa cells, none of the AHSP promoter reporter plasmids directed significant levels of luciferase expression ( Fig. 2A).
The AHSP Erythroid Promoter Contains Binding Sites for GATA-1 and Oct-1 Proteins-To identify binding sites for transcription factors within the core AHSP promoter, DNase I footprinting analysis of the Ϫ170/ϩ269 region with nuclear extracts from K562 cells was performed (Fig. 3A). Protected regions at Ϫ98 to Ϫ104 (5Ј-flanking) and ϩ238 to ϩ250 (intron 1) were identified, corresponding to consensus binding sequences for GATA-1 and Oct-1 proteins, respectively.
To determine if nuclear proteins could bind the footprinted GATA-1 site at Ϫ98 to Ϫ104 in the AHSP gene promoter in vitro, doublestranded oligonucleotide probes containing the corresponding AHSP promoter or control GATA-1 sequences (TABLE ONE) (18) were prepared and used EMSA with K562 (erythroid) extracts. The AHSP GATA-1 probe yielded a single complex that migrated identically to a complex formed with a control GATA-1 probe. These complexes were effectively competed both by an excess of unlabeled AHSP GATA-1 oligonucleotide, an excess of unlabeled control GATA-1 oligonucleotide, and a monoclonal antibody against GATA-1 protein (Fig. 3B). When a double-stranded oligonucleotide with mutation of the AHSP GATA-1 consensus sequence (GATA to GTTA) (19) was used as competitor in EMSA, complex formation was not abolished (not shown). These data indicate that GATA-1 binds in vitro to the AHSP gene promoter at Ϫ98 to Ϫ104.
Double-stranded oligonucleotide probes containing the corresponding AHSP promoter or control Oct-1 sequences (TABLE ONE) (20,21) were prepared and used in EMSA with K562 cell extracts. The AHSP Oct-1 probe yielded a large complex on EMSA (Fig. 4) that migrated at the same location as a complex obtained with a control Oct-1 probe.  Both AHSP and control Oct-1 complexes were effectively competed by an excess of unlabeled AHSP Oct-1 oligonucleotide, unlabeled control Oct-1 oligonucleotide, and Oct-1 antiserum (Fig. 4). When a doublestranded oligonucleotide with mutation of the AHSP Oct-1 consensus sequence (ATGTAAAT to CTGCAACC) (20,21) was used in EMSA, complex formation was abolished (Fig. 4). These data indicate that Oct-1-binding proteins bind in vitro to the AHSP gene promoter.
GATA-1 and Oct-1 Are Major Activators of the Human Erythroid AHSP Gene Promoter-To assess the relative importance of the Ϫ98 to Ϫ104 GATA-1 and ϩ238 to ϩ250 Oct-1 transcription factor binding sites in promoter function, mutations were introduced into these sites individually and in combination in the minimal Ϫ170/ϩ269 AHSP promoter/luciferase reporter plasmid. Wild-type and mutant plasmids were transiently transfected into K562 cells and luciferase activity assayed after 24 h (Fig. 2B). Mutation of the GATA-1 consensus sequence (site 1, GATA to GTTA) (19) decreased luciferase activity by 33% compared with wild type. Mutation of the Oct-1 site in intron 1 (ATGTAAAT to CTGCAACC) (20,21) reduced promoter activity by 60%. Mutation of both the GATA-1 site and the Oct-1 site reduced promoter activity to background.
Transactivation of the AHSP Gene Promoter by GATA-1 in Heterologous Cells-The Ϫ170/ϩ269 human AHSP gene minimal promoter luciferase plasmid was transiently transfected into HeLa cells. Luciferase activity was identical to background (Fig. 5). Cotransfection of 1 g of this AHSP gene promoter plasmid with increasing amounts of a GATA-1 cDNA expression plasmid (22) resulted in increasing luciferase activity with increasing amounts of GATA-1 plasmid (Fig. 5). However, cotransfection of 1 g of the Ϫ170/ϩ269 AHSP gene promoter plasmid with increasing amounts of an Oct-1 cDNA expression plasmid (IMAGE 4622256 in pcDNA3) did not result in any change in luciferase activity with increasing amounts of Oct-1 plasmid (Fig. 5).

Chromatin Immunoprecipitation Analysis of the Minimal Promoter
Region-Histone modifications within the minimal promoter region of the AHSP gene were examined using a ChIP assay with anti-diacetyl histone H3 and anti-tetraacetyl histone H4 antibodies (23)(24)(25). The core histones H3 and H4 were hyperacetylated in this region in chromatin from K562 cells (Fig. 6A) relative to the necdin gene. Because the AHSP gene 5Ј-flanking region was hyperacetylated in erythroid cells and it contains a functional GATA-1 binding site, we performed ChIP using anti-GATA-1 and the same PCR primers as above (Fig. 6B). These studies demonstrated that GATA-1 occupied the 5Ј-flanking region of the AHSP in chromatin from erythroid (K562) cells in vivo.
Transgenic Mice Express the AHSP/ A ␥-Globin Transgene in Erythroid Cells-We created five lines of transgenic mice containing 1-11 copies of the minimal Ϫ170/ϩ269 AHSP promoter fused to the human A ␥-globin gene (Fig. 7A). Using our previously described dual riboprobe that detects exon 2 of the human ␥-globin gene and exon 2 of the murine ␣-globin gene (Fig. 7A), we directly compared the level of human A ␥-globin and murine ␣-globin mRNA levels in tissues from transgenic animals by RNase protection.
All five AHSP/ A ␥-globin transgenic lines expressed ␥-globin mRNA in adult reticulocytes (TABLE TWO and Fig. 7B). The mean level of A ␥-globin compared with the level of murine ␣-globin mRNA was ϳ0.046 Ϯ 0.5. There was no correlation between transgene copy number and mRNA level, indicating the level of transgene mRNA was influenced by position effects.
To determine the distribution of human ␥-globin protein in the red cells of transgenic animals, an anti-human ␥-globin monoclonal antibody was used in fluorescence-activated cell sorting analyses. The pattern of ␥-globin expression in erythrocytes of two transgenic lines was uniform, i.e. present in 100% of cells (TABLE THREE, Fig.  7C) In two other transgenic lines, human ␥-globin expression was variegated. In the fifth line, expression was too low for accurate analyses.
The level of transgene expression was examined in tissues of mice from two transgenic lines perfused with saline immediately prior to sacrificing. RNase protection detected ␥-globin mRNA only in reticulocytes, adult bone marrow, and spleen but not in nonerythroid tissues (TABLE THREE). FIGURE 7. Human AHSP promoter/ A ␥-globin transgenic mice. A, AHSP promoter/ A ␥-globin transgene (top). A 439-bp AHSP gene promoter fragment (Ϫ170 to ϩ269) was fused to the human A ␥-globin gene (Ϫ4 to ϩ1906) to create the transgene construct shown. Hybrid human A ␥-globin/ mouse ␣-globin riboprobe (bottom). An Sp6 riboprobe containing sequences for both exon 2 of the human A ␥-globin gene and exon 2 of the murine ␣-globin gene was prepared to ensure that both the human A ␥-globin and murine ␣-globin sequences are labeled to equal specific activity in ribonuclease protection assays. B, detection of AHSP/ A ␥-globin mRNA in reticulocytes of transgenic mice containing the AHSP/human ␥-globin transgene. 1.0 g of RNA from adult reticulocytes was hybridized to a 32 P-labeled antisense riboprobe, which protects exon 2 of the AHSP/ A ␥-globin transgene (top band) and exon 2 of the mouse ␣-globin gene (lower band), digested with RNase, electrophoresed in an 8% nondenaturing gel, dried and subjected to autoradiography. C, expression of human ␥-globin protein in erythrocytes of transgenic mice. Fluorescence intensity from a fluorescein isothiocyanate-conjugated monoclonal antibody against human ␥-globin was correlated with the number of erythrocytes counted. In each panel, the thick line represents a transgenic mouse and the thin line represents a nontransgenic, littermate control. Dual peaks represent variegated expression of human ␥-globin. The top left panel shows results from a control mouse with uniform (i.e. 100% of erythrocytes) ␥-globin erythrocyte expression. The top right panel shows results from a control mouse with variegated ␥-globin erythrocyte expression. Panels A and D demonstrate variegated ␥-globin expression and panels B and E demonstrate uniform ␥-globin expression in erythrocytes from these AHSP transgenic mouse lines.

DISCUSSION
There are numerous consensus GATA-1 binding sites in the AHSP gene, including 5 in the minimal Ϫ170/ϩ269 promoter region, 3 in the 5Ј-flanking region and 2 in intron 1. In the AHSP minimal promoter, the functionally important GATA-1 site at Ϫ98 to Ϫ104 and a second site at Ϫ47 to Ϫ52 are conserved between mouse and man. The Ϫ47 to Ϫ52 GATA-1 site was not protected in in vitro footprinting (Fig. 3) and did not bind GATA-1 in electrophoretic mobility shift assays (not shown). Transactivation of the AHSP promoter by GATA-1 in nonerythroid cells is consistent with the observation that AHSP is induced by GATA-1.
The second region protected in DNase I footprinting corresponded to an Oct-1 consensus site in the AT-rich region of AHSP intron 1. EMSA and transfection assays demonstrated the functional importance of Oct-1 in vitro. Oct-1 is a ubiquitously expressed member of the POU domain family of transcription factors known to regulate numerous tissue-specific and ubiquitous genes as transcriptional activators or repressors (20,21). Recent gene ablation studies have demonstrated a critical role for Oct-1 in embryonic development and erythropoiesis (26). E12.5 fetal livers from Oct-1-deficient embryos demonstrated decreased numbers of TER-119-positive cells and decreased amounts of ␤-globin mRNA in erythroid cells.
The minimal AHSP promoter utilized in the transgenic mouse studies demonstrated that it contains the sequences necessary for erythroidspecific expression of AHSP. However, the lack of correlation between transgene mRNA levels and transgene copy number and the variegated expression seen in some lines suggesting that additional elements are necessary for authentic expression of the AHSP gene.
AHSP deficiency worsened the phenotype in ␤-thalassemic mice, leading to the suggestion that AHSP could be a modifier gene influencing the phenotype of human ␤-thalassemia syndromes (1, 6), which are marked by clinical heterogeneity (27). In a study of 120 thalassemia patients with hemoglobin E-thalassemia of varying clinical severity from Thailand, no mutations were identified in the exons of the AHSP gene or in the 5Ј-flanking DNA (28). Similar results were found in a population of ␤-thalassemic patients from Italy (29). However, in separate preliminary reports (30,4), thalassemic patients with discordant phenotypes were found to have decreased AHSP mRNA expression in reticulocytes and cultured erythroid precursors as determined by realtime, reverse transcription polymerase chain reaction. In one discordant thalassemic patient with decreased reticulocyte AHSP mRNA, reduced amounts of AHSP protein were observed in reticulocytes and cultured erythroid cells (4). Similar to the ␤-globin gene mutations causing thalassemia, it is possible that there are genetic differences at the AHSP locus, or in genetic or epigenetic regulators, that influence AHSP gene expression. We suggest that the Ϫ170/ϩ269 region be examined in thalassemia patients with varying severity, especially those with decreased AHSP.