Increased gamma-globin expression in a nondeletion HPFH mediated by an erythroid-specific DNA-binding factor

Erythroid differentiation in chimaeric mice blocked by a targeted mutation in the gene for transcription factor GATA-1

The zinc-finger transcription factor GATA-1 (previously known as GF-1, NF-E1 or Eryf 1 binds to GATA consensus elements in regulatory regions of the alpha- and beta-globin gene clusters and other erythroid cell-specific genes. Analysis of the effects of mutations in GATA-binding sites in cell culture and in binding assays in vitro, as well as transactivation studies with GATA-1 expression vectors in heterologous cells, have provided indirect evidence that this factor is involved in the activation of globin and other genes during erythroid cell maturation. GATA-1 is also expressed in megakaryocytes and mast cells, but not in other blood cell lineages or in non-haemopoietic cells. To investigate the role of this factor in haematopoiesis in vivo, we disrupted the X-linked GATA-1 gene by homologous recombination in a male (XY) murine embryonic stem cell line and tested the GATA-1-deficient cells for their ability to contribute to different tissues in chimaeric mice. The mutant embryonic stem cells contributed to all non-haemopoietic tissues tested and to a white blood cell fraction, but failed to give rise to mature red blood cells. This demonstrates that GATA-1 is required for the normal differentiation of erythroid cells, and that other GATA-binding proteins cannot compensate for its absence.

Regulated expression of globin chains and the erythroid transcription factor GATA-1 during erythropoiesis in the developing mouse

Erythropoiesis in vertebrates is characterized by sequential changes in erythropoietic site, erythroblast morphology, and hemoglobin synthesis. We have examined the expression of globin chains and the major erythroid transcription factor GATA-1 (previously known as GF-1/NF-E1/Eryf 1) from days 7.5 to 17.5 of mouse development. mRNAs for embryonic (epsilon y2, beta H1, and zeta) and adult (alpha and beta) globin chains were quantitated by RNase protection assays. Switching of globins within the alpha-globin cluster (alpha and zeta) was not strictly coordinated with that within the beta-globin cluster (epsilon y2, beta H1, and beta). Regulation of globin switches during development was primarily transcriptional. Of particular note, we found two developmental switches (beta H1 to epsilon y2 and epsilon y2 to beta) in the mouse, more analogous than previously thought to shifts found in human development. The erythroid transcription factor GATA-1, believed to be a principal regulator of genes expressed in erythroid cells, first appeared in the embryo in yolk sac at the time of blood island formation and remained at a low level during embryonic erythropoiesis (8 to 11 days) relative to that found later in fetal liver (12 to 15 days). The rise in GATA-1 mRNA in fetal liver paralleled and preceded the rapid accumulation of adult beta-globin RNA. RNase protection assays and a GATA-1-specific peptide antiserum were used to establish that a single GATA-1 polypeptide is expressed throughout mouse development. Overall, these findings suggest that the levels of this erythroid transcription factor during development may contribute to the differential gene activation characteristic of definitive versus primitive erythropoiesis.

Retroviral transfer of a human fetal globin gene carrying the -202 G gamma beta (+)-HPFH mutation into the human erythroleukemia line, KMOE

The presence of point mutations at position -202 relative to the mRNA Cap site of both human fetal gamma-globin genes is linked with elevated fetal globin levels in adults. The question addressed in this study is whether the -202 mutation affects gamma-globin gene expression in the same manner as the -117 hereditary persistence of fetal hemoglobin (HPFH) A gamma-globin mutation. The -117 mutation was found to cause over-expression and confer inducibility of a retrovirally transferred gamma-globin gene in cytosine arabinoside (araC)-treated KMOE cells in an earlier study. In this study, fetal globin genes driven by either the normal G gamma or -202 HPFH G gamma-globin promoter were retrovirally transferred into human erythroid KMOE cells. The -202 HPFH mutation did not cause over-expression or confer inducibility of the transferred gamma-globin gene in araC-treated KMOE cells. Thus, the -202 HPFH mutation affects gamma-globin gene expression by a different mechanism than the -117 HPFH mutation. Furthermore, this study provides evidence against a general increasing of gamma-globin gene expression as might be expected from the -202 mutation altering binding of a ubiquitous factor such as Sp1.

Progressive inactivation of the expression of an erythroid transcriptional factor in GM- and G-CSF-dependent myeloid cell lines

The transcriptional binding protein NFE-1 (also called GF-1 and Ery-f1) is thought to play a necessary, but not sufficient, role in the regulation of differentiation-related gene expression in a subset of hematopoietic lineages (erythroid, megakaryocytic, and basophil-mast cell). In order to clarify the mechanism which underlies the lineage-specificity of the NFE-1 expression, as well as the relationship between the expression of this factor and growth factor responsiveness, we have evaluated the capacity of erythropoietin (Epo)-, granulomonocytic (GM)-colony stimulating factor (CSF)-, and granulocyte (G)-CSF-dependent subclones derived from the interleukin 3 (IL-3)-dependent cell line 32D, to express 1) NFE-1 mRNA, 2) NFE-1-related nuclear proteins, and 3) chloramphenicol acetyl transferase (CAT) activity when transfected with a CAT gene under the control of NFE-1 cognate sequences. NFE-1 mRNA was found to be expressed not only in cells with mast cell (IL-3-dependent 32D) and erythroid (Epo-dependent 32D Epo1) phenotypes, but also in cells with predominantly granulocyte/macrophage properties, such as the GM-CSF- (early myelomonocytic) and G-CSF- (myelocytic) dependent subclones of 32D. However, a gradient of expression, correlating with the lineage, the stage of differentiation, and the growth factor responsiveness of the cell lines, was found among the different subclones: Epo greater than or equal to IL-3 greater than GM-CSF greater than G-CSF. Binding experiments demonstrated NFE-1 activity in all cell lines except the G-CSF-dependent line. Function of the NFE-1 protein was assessed by the expression of the CAT gene linked to the SV40 promoter and a mutant (-175 T----C) HPFH gamma-globin promoter. High level CAT expression was seen only in the Epo1 cells although low level expression was also seen in the parent 32D. These results demonstrate that the specificity of the expression of NFE-1 for the erythroid--megakaryocytic--mast cell lineages is obtained by progressive inactivation of its expression in alternative lineages.

Characterization of murine erythropoietin receptor genes

We have isolated and characterized the murine genomic and complementary DNAs encoding erythropoietin (Epo) receptor from Epo-responsive and unresponsive mouse erythroleukemia cells. Two classes of Epo receptor cDNAs were isolated from Epo-responsive cells. One is a 55,000 Mr membrane-bound Epo receptor, and the other is a 29,000 Mr soluble Epo receptor lacking the transmembrane and cytoplasmic domains. As a result of alternative splicing, two insert sequences containing termination codons are produced, and the encoded polypeptide diverges four amino acids upstream from the transmembrane domain, adding 20 new amino acids before terminating. Amino acid sequence of the Epo receptor cDNA isolated from Epo-responsive cells was identical with that of Epo-unresponsive cells, indicating that Epo-responsiveness does not depend upon the primary structure of the Epo receptor (binding) protein. Analysis of 6.6 x 10(3) base-pairs (kb) genomic DNA segments covering complete Epo receptor gene and promoter regions revealed that potential regulatory elements (NF-E1, GF-1 or Eryf 1) for erythroid-specific and differentiation stage-specific gene expression are located in the promoter and 3' noncoding regions.

Function of transfected globin promoters and the globin locus activator in K562 erythroleukemia cells

We have examined the importance of cis-acting regulatory elements within the human gamma-globin gene promoter and the globin locus activating region in K562 cells. A gamma-globin or beta-globin promoter fragments were fused with the neomycin phosphotransferase gene in a plasmid-based vector (gamma-neo or beta-neo) and transiently transfected by electroporation into K562 cells. Correctly initiated gamma-neo or beta-neo transcripts were detected with an S1 nuclease protection assay that was internally controlled for transfection efficiency and RNA content. We first optimized the conditions for electroporation and then determined that a gamma-globin promoter fragment extending from -299 and +36 was active in the assay but that a beta-globin promoter extending from -375 to +46 was inactive. Deletion of the gamma-globin promoter to -199 did not affect promoter function, but deletion to -160 reduced promoter strength to 70% of that of control. Additional deletion to position -130 reduced promoter strength to 19% of the control value, and to position -61, 8.7% of the control value. Three gamma-globin promoters containing mutations associated with hereditary persistence of fetal hemoglobin (HPFH), -202 C----G, -196 C----T and -117 G----A, were not overexpressed in K562 cells, consistent with the hypothesis that these promoters are not overexpressed in fetal erythroblasts, only in adult red cells. When the beta-globin locus activating region (LAR) was added to a wild-type or an HPFH gamma-neo plasmid, the abundance of correctly initiated gamma-neo transcripts increased dramatically. However, beta-neo expression could not be activated by the LAR in K562 cells. These studies should allow us to further dissect the interactive roles of globin promoters and enhancers in K562 cells.

Regulated expression of globin chains and the erythroid transcription factor GATA-1 during erythropoiesis in the developing mouse

Erythropoiesis in vertebrates is characterized by sequential changes in erythropoietic site, erythroblast morphology, and hemoglobin synthesis. We have examined the expression of globin chains and the major erythroid transcription factor GATA-1 (previously known as GF-1/NF-E1/Eryf 1) from days 7.5 to 17.5 of mouse development. mRNAs for embryonic (epsilon y2, beta H1, and zeta) and adult (alpha and beta) globin chains were quantitated by RNase protection assays. Switching of globins within the alpha-globin cluster (alpha and zeta) was not strictly coordinated with that within the beta-globin cluster (epsilon y2, beta H1, and beta). Regulation of globin switches during development was primarily transcriptional. Of particular note, we found two developmental switches (beta H1 to epsilon y2 and epsilon y2 to beta) in the mouse, more analogous than previously thought to shifts found in human development. The erythroid transcription factor GATA-1, believed to be a principal regulator of genes expressed in erythroid cells, first appeared in the embryo in yolk sac at the time of blood island formation and remained at a low level during embryonic erythropoiesis (8 to 11 days) relative to that found later in fetal liver (12 to 15 days). The rise in GATA-1 mRNA in fetal liver paralleled and preceded the rapid accumulation of adult beta-globin RNA. RNase protection assays and a GATA-1-specific peptide antiserum were used to establish that a single GATA-1 polypeptide is expressed throughout mouse development. Overall, these findings suggest that the levels of this erythroid transcription factor during development may contribute to the differential gene activation characteristic of definitive versus primitive erythropoiesis.

Stopping the biologic clock for globin gene switching

The developmental switch from production of fetal (gamma) to adult (beta) globin occurs on a normally set biologic clock which proceeds even if expression of the adult (beta) globin genes is defective and produces little or no protein, as in the beta-thalassemias. Preventing or reversing the globin gene switch could provide a way of keeping the abnormal globin genes "silent" and maintaining expression of the fetal globin gene. We have identified a class of agents which, when present in elevated plasma concentrations during gestation, inhibits the gamma----beta-globin gene switch in developing humans. Further investigation has shown that butyric acid and related compounds can increase gamma-globin and decrease beta-globin expression in cultured erythroid cells of patients with beta-thalassemia. Butyrate compounds were therefore infused in an in vivo fetal animal model, and the globin switch was inhibited and even reversed in some fetal lambs. Histone hyperacetylation, which maintains active chromatin structure, and an effect on the gamma-globin promoter appear to be mechanisms of action involved. These data suggest that inhibiting expression of abnormal beta-globin genes by pharmacologic means may in the future be possible for treatment of individuals with beta-globin disorders.

Transcriptional activation and DNA binding by the erythroid factor GF-1/NF-E1/Eryf 1

The murine, erythroid DNA-binding protein GF-1 (also known as NF-E1, Eryf 1), a 413-amino acid polypeptide with two novel finger domains of the Cx-Cx variety, recognizes a consensus GATA motif present in cis elements of the majority of erythroid-expressed genes. We have performed a structure-function analysis of this protein to evaluate its potential as a transcriptional activator and to examine the role of the finger domains in DNA binding. Using a cotransfection assay, we find that GF-1 is a potent transcriptional activator with several activation domains but that this is revealed only in heterologous cells and with reporters containing minimal promoters onto which either a single or multiple GATA-binding sites are placed. The two fingers of GF-1 are functionally distinct and cooperate to achieve specific, stable DNA binding. The amino finger is necessary only for full specificity and stability of binding, whereas the carboxyl finger is required for binding. The role of each finger is more pronounced with some GATA-binding sites than with others, suggesting a diversity of interactions between GF-1 and different target sites. The complex activation and DNA-binding properties of GF-1 are likely to contribute to the ability of this single protein to participate widely in gene expression throughout erythroid development.

The T----C substitution at -198 of the A gamma-globin gene associated with the British form of HPFH generates overlapping recognition sites for two DNA-binding proteins

Defects in the developmental changes of human hemoglobin production characterized by the continued expression of fetal globin during adult life are classified as hereditary persistence of fetal hemoglobin (HPFH). Among the various molecular lesions associated with this phenotype, the non-deletion forms with point mutations in the promoter region are thought to provide mechanistic clues for gamma-globin gene regulation. The natural occurrence of four different base substitutions mapping within six nucleotides of a homopurine.homopyrimidine motif in the upstream promoter region demarcate a potential control element. To assess its importance for transcriptional activity, we compared the -202 (C----G), -198 (T----C) and -196 (C----T) HPFH mutations with the normal sequence in binding studies with nuclear proteins from erythroid and non-erythroid cells. Wildtype DNA and HPFH mutations at -202 or -196 showed only a weak protein interaction of unclear functional significance. In contrast, -198 (T----C) generated overlapping, high-affinity binding sites for two ubiquitous nuclear proteins. One cognate protein was identified as the transcription factor Sp1. The second one was termed NF-G.C as it interacted strongly with the homopolymer poly(dG).poly(dC). The generation of additional recognition sites for trans-acting factors by the -198 HPFH mutation correlated with a modest increase in promoter activity in vitro specifically with nuclear extracts from erythroid cells. The activation appears to be mediated by binding of Sp1, but it requires interaction with an erythroid-specific factor, most likely GF-1. Templates containing the -196 HPFH mutation showed a transcriptional activity identical to wildtype. This suggests that despite the topological proximity of the mutations, the HPFH phenotype may be established by different mechanisms.

Activity and tissue-specific expression of the transcription factor NF-E1 multigene family

NF-E1, a DNA-binding protein that recognizes the general consensus motif WGATAR, is the first tissue-specific factor to be identified in erythroid cells. Using a probe from the murine GF-1 (NF-E1) cDNA clone, we isolated three homologous chicken cDNAs: One of these corresponds to an mRNA (NF-E1a) that is abundantly and exclusively expressed in erythroid cells; a second mRNA (NF-E1b) is also expressed in all developmental stages of erythroid cells but is additionally found in a limited subset of other chicken tissues; mRNA representative of a third gene (NF-E1c) is expressed only in definitive (adult) erythrocytes within the red cell lineage but is also abundantly expressed in T lymphocytes and brain. All NF-E1 proteins are highly conserved within the DNA-binding domain and bind to the consensus motif with similar affinities in vitro; they are also all stimulatory trans-acting factors in vivo. The factors differ quantitatively in their ability to trans-activate reporter genes in which the number and position of cognate binding sites is varied relative to the transcriptional initiation site. These data suggest that the NF-E1 consensus motif directs a broader and more complicated array of developmental transcriptional regulatory processes than has been assumed and that NF-E1c may play a unique regulatory role in the developing chicken brain and in T lymphocytes.

A nonerythroid GATA-binding protein is required for function of the human preproendothelin-1 promoter in endothelial cells

Endothelin-1 (ET-1) is a 21-amino-acid peptide synthesized by endothelial cells that has potent vasoconstrictor activity. Human ET-1 is derived from a 212-amino-acid prepropeptide, termed preproendothelin-1 (PPET-1). To identify cis-acting sequences essential for PPET-1 gene transcription, bovine aortic endothelial (BAE) cells were transfected with plasmids containing 5'-flanking sequences of the human PPET-1 gene fused to the human growth hormone gene as a reporter. Deletional analysis of these fusion plasmids showed that the sequence spanning positions -141 to -127 of the human PPET-1 promoter is required for full transcription activity. Introduction of clustered point mutations into this region of the promoter reduced transcription activity. Gel shift analysis, methylation interference, protein-DNA cross-linking, and oligonucleotide competition studies revealed that BAE cell nuclear extract contains a 47-kilodalton DNA-binding protein recognizing the core motif TATC (GATA) located at positions -135 to -132 of the PPET-1 promoter. The size and specificity of this DNA-binding protein resemble GF-1, a previously described transcription factor of erythroid cells that binds to the same core motif. Gel shift analysis indicated that GF-1 and the DNA-binding protein interacting with the PPET-1 promoter have different tissue distributions; the former is restricted to a subset of hematopoietic cells, and the latter is found in various cell types, including BAE, NIH 3T3, and HeLa cells. By using an antiserum to the C-terminal region of GF-1, the two proteins were also found to be antigenically distinct. When a growth hormone fusion plasmid containing the proximal 141 nucleotides of the PPET-1 promoter was transfected into a variety of cell types, these was preferential expression in cells of endothelial origin. We conclude that a nuclear factor with binding specificity for a GATA motif similar to that of the transcriptional activator GF-1 is necessary for the efficient and cell-specific expression of the human PPET-1 gene.

A novel C-T transition within the distal CCAAT motif of the G gamma-globin gene in the Japanese HPFH: implication of factor binding in elevated fetal globin expression

Hereditary persistence of fetal hemoglobin (HPFH) is a condition characterized by the continued expression of the fetal globin gene in adulthood. Both deletional and nondeletional forms have been described. We studied one Japanese family with two different nondeletional forms of HPFH. Analysis of polymorphic restriction sites in the beta-globin gene cluster suggested that one affecting both G gamma and A gamma globin expression in two members of the family could be associated with unknown conditions not linked to the beta-globin gene loci. Characterization by the polymerase chain reaction (PCR) of another form producing a G gamma-HPFH phenotype in two other members demonstrated a novel C-T transition at the nucleotide -114 within the distal CCAAT motif of the G gamma-globin gene. Using gel retardation assays on various nuclear extracts, we also demonstrated that this novel mutation abolishes the binding of the ubiquitous CCAAT binding factor, CP1 to the distal CCAAT motif of the gamma-globin gene but does not affect the binding of any erythroid specific factor, thereby suggesting a possible role for CP1 in the developmental regulation of fetal globin expression.

A nonerythroid GATA-binding protein is required for function of the human preproendothelin-1 promoter in endothelial cells

Endothelin-1 (ET-1) is a 21-amino-acid peptide synthesized by endothelial cells that has potent vasoconstrictor activity. Human ET-1 is derived from a 212-amino-acid prepropeptide, termed preproendothelin-1 (PPET-1). To identify cis-acting sequences essential for PPET-1 gene transcription, bovine aortic endothelial (BAE) cells were transfected with plasmids containing 5'-flanking sequences of the human PPET-1 gene fused to the human growth hormone gene as a reporter. Deletional analysis of these fusion plasmids showed that the sequence spanning positions -141 to -127 of the human PPET-1 promoter is required for full transcription activity. Introduction of clustered point mutations into this region of the promoter reduced transcription activity. Gel shift analysis, methylation interference, protein-DNA cross-linking, and oligonucleotide competition studies revealed that BAE cell nuclear extract contains a 47-kilodalton DNA-binding protein recognizing the core motif TATC (GATA) located at positions -135 to -132 of the PPET-1 promoter. The size and specificity of this DNA-binding protein resemble GF-1, a previously described transcription factor of erythroid cells that binds to the same core motif. Gel shift analysis indicated that GF-1 and the DNA-binding protein interacting with the PPET-1 promoter have different tissue distributions; the former is restricted to a subset of hematopoietic cells, and the latter is found in various cell types, including BAE, NIH 3T3, and HeLa cells. By using an antiserum to the C-terminal region of GF-1, the two proteins were also found to be antigenically distinct. When a growth hormone fusion plasmid containing the proximal 141 nucleotides of the PPET-1 promoter was transfected into a variety of cell types, these was preferential expression in cells of endothelial origin. We conclude that a nuclear factor with binding specificity for a GATA motif similar to that of the transcriptional activator GF-1 is necessary for the efficient and cell-specific expression of the human PPET-1 gene.

Positive and negative elements regulate human interleukin 3 expression

The human interleukin 3 (IL-3) promoter is comprised of several cis-acting DNA sequences that modulate T-cell expression of IL-3. These are located within 315 nucleotides upstream of the mRNA start site. Transient expression of reporter genes linked to serially deleted sequences of the IL-3 promoter has allowed mapping of two activator sequences and an interposed repressor sequence. The proximal regulatory region is specific to IL-3 and prerequisite for efficient transcription. Its effect is enhanced by a second, more distal activating sequence consisting of an AP-1 binding site. Between the two activators lies a transcriptional silencer, which is a potent repressor in the absence of the AP-1 site. DNA-nuclear protein binding experiments demonstrate specific complex formation within each of these functional regions. Thus, both positive and negative regulatory elements appear to control expression of the human IL-3 gene in activated T cells.

Human globin locus activation region (LAR): role in temporal control

A region of DNA located far upstream of the human beta-globin locus is critically involved in the regulation of the beta-globin gene family. Recent experiments in transgenic mice suggest that switching from fetal to adult globin gene expression during human development results from competition among individual globin gene family members for interaction with sequences in this region. The phenotypes of patients with defined hemoglobinopathies support this hypothesis.

Disruption of a C/EBP binding site in the factor IX promoter is associated with haemophilia B

Haemophilia B (or Christmas disease) is an inherited, X-linked bleeding disorder caused by mutations in the gene for clotting factor IX. There is a rare class of patients, exemplified by haemophilia B Leyden, who suffer from haemophilia B as children but improve after puberty. In these patients, plasma factor IX concentrations are less than 10% of normal during childhood, but after puberty they gradually rise to between 40 and 80% of normal. Mutations clustered around the main transcription start point (defined as +1 (ref.2)) have been reported in seven of these patients (at -20 (refs 1, 3, 4); -6 (refs 5, 6) and +13 (refs 7, 8)). To determine how these mutations interfere with factor IX expression, we have assayed for transcription factors binding to this area and have identified a nuclear factor-1 liver (NF1-L) binding site (-99 to -76) and a binding site for the CCAAT/enhancer binding protein (C/EBP) (+1 to +18). We show that the A----G mutation at +13 prevents the binding of C/EBP to this site. Furthermore, we show that C/EBP is capable of transactivating a cotransfected normal factor IX promoter but not the mutant promoter. This is the first natural mutation to be reported which disrupts a C/EBP binding site and is an illustration of the importance of this transcription factor in humans.

Protein-DNA interactions upstream from the human A gamma globin gene

We have used DNAase I footprinting and the gel mobility shift assay to study proteins which bind to promoter elements located between -140 and -382 upstream of the human A gamma globin gene. Footprints are found with both erythroid and nonerythroid nuclear extracts at three sites: from -294 to -264, -242 to -227, and -189 to -172 from the transcription initiation site. An erythroid-specific footprint is identified from -194 to -189. We demonstrate that two known transcription factors, the ubiquitous octamer-binding protein OTF-1 and the erythroid regulatory factor NFE-1, bind to the -194 to -172 region and that their footprints overlap. Binding of OTF-1 to this region is reduced by a mutation at -175 associated with a form of non-deletion hereditary persistence of fetal hemoglobin. We conclude that OTF-1 may compete with NFE-1 for the -175 binding site, possibly functioning as a repressor of gamma globin transcription.

Expression of an erythroid transcription factor in megakaryocytic and mast cell lineages

The nuclear factor GF-1 (also known as NF-E1, Eryf-1; refs 1-3 respectively) is important in regulation of the transcription of globin and other genes that are specifically expressed in erythroid cells. We have previously shown that GF-1 of both mouse and human origin is a 413-amino-acid polypeptide with two novel zinc-finger domains whose expression is restricted to erythroid cells. Using in situ hybridization of mouse bone marrow cells and northern blot analysis of purified cell populations and permanent cell lines, we show here that GF-1 is expressed in two other hematopoietic lineages, megakaryocytes and bone marrow-derived mast cells. Our findings are consistent with results from hematopoietic progenitor culture which suggest a relationship between erythroid, megakaryocytic and mast cell lineages, and imply that GF-1 is expressed in committed multipotential cells and their progeny. Hence, the mere presence of this transcription factor is unlikely to be sufficient to programme differentiation of a single haematopoietic lineage. GF-1 may regulate the transcription of not only erythroid genes, but also many genes characteristic of megakaryocytes and mast cells, or genes shared among these lineages.

Megakaryocytic and erythrocytic lineages share specific transcription factors

Erythroid-specific genes contain binding sites for NF-E1 (also called GF-1 and Eryf-1; refs 1-3 respectively), the principal DNA-binding protein of the erythrocytic lineage. NF-E1 expression seems to be restricted to the erythrocytic lineage. A closely related (if not identical) protein is found in both a human megakaryocytic cell line and purified human megakaryocytes; it binds to promoter regions of two megakaryocytic-specific genes. The binding sites and partial proteolysis profile of this protein are indistinguishable from those of the erythroid protein; also, NF-E1 messenger RNA is the same size in both the megakaryocytic and erythroid cell lines. Furthermore, point mutations that abolish binding of NF-E1 result in a 70% decrease in the transcriptional activity of a megakaryocytic-specific promoter. We also find that NF-E2, another trans-acting factor of the erythrocytic lineage, is present in megakaryocytes. Transcriptional effects in both lineages might then be mediated in part by the same specific trans-acting factors. Our data strengthen the idea of a close association between the erythrocytic and the megakaryocytic lineages and could also explain the expression of markers specific to the erythrocytic and megakaryocytic lineages in most erythroblastic and megakaryoblastic permanent cell lines.

Roles of fetal G gamma-globin promoter elements and the adult beta-globin 3' enhancer in the stage-specific expression of globin genes

The human fetal G gamma-globin and adult beta-globin genes are expressed in a tissue- and developmental stage-specific pattern in transgenic mice: the G gamma gene in embryonic cells and the beta gene in fetal and adult erythroid cells. Several of the cis-acting DNA sequences thought to be responsible for these patterns of expression are located 5' to the G gamma-globin gene and 3' to the beta-globin gene. To further define the locations and functional roles of these elements, we examined the effects of 5' truncations on the expression of the G gamma-globin gene, as well as the ability of G gamma-globin upstream sequences to alter the developmental regulation of a beta-globin gene, as well as the ability of G gamma-globin upstream sequences to alter the developmental regulation of a beta-globin gene. We found that sequences between -201 and -136 are essential for expression of the G gamma-globin gene, whereas those upstream of -201 have little effect on the level or tissue or stage specificity of G gamma-globin expression. The G gamma-globin upstream sequences from -201 to -136 were, furthermore, capable of activating a linked beta-globin gene in embryonic blood cells; however, a G gamma-globin fragment from -383 to -206 was similarly active in this assay, and the complete fragment from -383 to -136 was considerably more active than either of the smaller fragments, suggesting the presence of multiple cis-acting elements for embryonic blood cells. Our data also suggested the possibility of a negative regulatory element between -201 and -136. These results are discussed in relation to several DNA elements in the G gamma-globin upstream region, which have been shown to bind nuclear factors in erythroid cells. Finally, we observed that removal of the beta-globin 3'-flanking sequences, including the 3' enhancer, from the G gamma-globin upstream-beta-globin hybrid gene resulted in a 25-fold reduction in expression in embryonic blood cells. This suggests that the beta-globin 3' enhancer is potentially active at the embryonic stage and thus cannot be solely responsible for the fetal or adult specificity of the beta-globin gene.

A factor binding GATAAG confers tissue specificity on the promoter of the human zeta-globin gene

We describe the characterisation of cis-acting sequences which control the tissue specific expression of the human zeta globin gene. An extensive search for enhancer sequences in the vicinity of this gene proved negative. Instead our data demonstrate that the minimal promoter of the zeta gene is itself tissue specific. Sequences close to and possibly including the -100 CACCC and -70 CCAAT boxes display some erythroid specificity. However the principal tissue specific element is a GATAA sequence at -120 directly adjacent to the minimal promoter. Specific deletion of GATAA reduces zeta promoter activity 5 fold in erythroid but not non-erythroid cells. We also demonstrate that an erythroid specific factor binds to this GATAA sequence. Furthermore this factor forms a complex with the transcription factor CP1 which we show interacts with the zeta CCAAT box. We present evidence that the zeta GATAA binding factor is equivalent to GF1 recently purified and cloned by Tsai et al [1]. The erythroid specific GATAA sequence has been found in the promoters and enhancers of a number of erythroid specific genes. Similarly we show here that the zeta globin gene relies on a GATAA sequence in its promoter to specify its expression in erythroid cells.

Roles of fetal G gamma-globin promoter elements and the adult beta-globin 3' enhancer in the stage-specific expression of globin genes

The human fetal G gamma-globin and adult beta-globin genes are expressed in a tissue- and developmental stage-specific pattern in transgenic mice: the G gamma gene in embryonic cells and the beta gene in fetal and adult erythroid cells. Several of the cis-acting DNA sequences thought to be responsible for these patterns of expression are located 5' to the G gamma-globin gene and 3' to the beta-globin gene. To further define the locations and functional roles of these elements, we examined the effects of 5' truncations on the expression of the G gamma-globin gene, as well as the ability of G gamma-globin upstream sequences to alter the developmental regulation of a beta-globin gene, as well as the ability of G gamma-globin upstream sequences to alter the developmental regulation of a beta-globin gene. We found that sequences between -201 and -136 are essential for expression of the G gamma-globin gene, whereas those upstream of -201 have little effect on the level or tissue or stage specificity of G gamma-globin expression. The G gamma-globin upstream sequences from -201 to -136 were, furthermore, capable of activating a linked beta-globin gene in embryonic blood cells; however, a G gamma-globin fragment from -383 to -206 was similarly active in this assay, and the complete fragment from -383 to -136 was considerably more active than either of the smaller fragments, suggesting the presence of multiple cis-acting elements for embryonic blood cells. Our data also suggested the possibility of a negative regulatory element between -201 and -136. These results are discussed in relation to several DNA elements in the G gamma-globin upstream region, which have been shown to bind nuclear factors in erythroid cells. Finally, we observed that removal of the beta-globin 3'-flanking sequences, including the 3' enhancer, from the G gamma-globin upstream-beta-globin hybrid gene resulted in a 25-fold reduction in expression in embryonic blood cells. This suggests that the beta-globin 3' enhancer is potentially active at the embryonic stage and thus cannot be solely responsible for the fetal or adult specificity of the beta-globin gene.

Cell type-specific protein-DNA interactions in the human zeta-globin upstream promoter region: displacement of Sp1 by the erythroid cell-specific factor NF-E1

The protein-DNA interactions of the upstream promoter region of the human embryonic zeta-globin gene in nuclear extracts of erythroid K562 cells and nonerythroid HeLa cells were analyzed by DNase I footprinting, gel mobility shift assay, methylation interference, and oligonucleotide competition experiments. There are mainly two clusters of nuclear factor-binding sites in the zeta promoter. The proximal cluster spans the DNA sequence from -110 to -60 and consists of binding sites for CP2, Sp1, and NF-E1. NF-E1 binding is K562 specific, whereas CP2 binding is common to both types of cells. Overlapping the NF-E1- and CP2-binding sites is a hidden Sp1-binding site or CAC box, as demonstrated by binding studies of affinity-purified Sp1. In the distal promoter region at -250 to -220, another NF-E1-binding site overlaps a CAC box or Sp1-binding site. Extract-mixing experiments demonstrated that the higher affinity of NF-E1 binding excluded the binding of Sp1 in the K562 extract. NF-E1 factors could also displace prebound Sp1 molecules. Between the two clusters of multiple-factor-binding sites are sequences recognized by other factors, including zeta-globin factors 1 and 2, that are present in both HeLa and K562 extracts. We discuss the cell type-specific, competitive binding of multiple nuclear factors in terms of functional implications in transcriptional regulation of the zeta-globin gene.

A naturally occurring gamma globin gene mutation enhances SP1 binding activity

Transcription of the human fetal globin genes in erythroid cells is tightly regulated during different stages of development and differentiation. Two naturally occurring mutations 202 base pairs upstream of the duplicated gamma globin genes are associated with incorrectly regulated gamma globin gene gene expression; elevated levels of fetal globin are synthesized during adult life. A C-to-G base substitution upstream of the G gamma-globin gene is highly correlated with a dramatic increase in gene expression. It increases the similarity of the region to the consensus Sp1 recognition site. We determined that the mutated DNA had a 5- to 10-fold-higher affinity for Sp1 than did normal gamma globin gene sequence. We also observed a reduction in normal factor-binding activity. A different substitution at -202, C to T, upstream of the A gamma-globin gene was associated with a more moderate increase in fetal globin expression. This mutation decreased the similarity of the sequence to an Sp1 recognition site. We determined that it did not result in enhanced Sp1 binding but did alter normal factor binding. We suggest that these changes in nuclear protein-binding properties detected in vitro are responsible for the enhanced gamma globin gene expression found in -202 G gamma beta + patients with hereditary persistence of fetal hemoglobin.

Structure and evolution of a human erythroid transcription factor

Vertebrate erythroid cells contain a tissue-specific transcription factor referred to as Eryf 1 (ref. 1), GF-1 (ref. 2) or NF-E1 (ref. 3), for which binding sites are widely distributed in the promoters and enhancers of the globin gene family, and of other erythroid-specific genes. Aberrant binding of the human factor to a mutant site has been implicated in one form of hereditary persistence of fetal haemoglobin (HPFH; ref. 2). The complementary DNAs for both the chicken cEryf 1 (ref. 11) and mouse mEryf 1 (ref. 12) encoding genes have recently been cloned. We report here the cloning of the cDNA for the human Eryf 1 encoding gene. The central third of the hEryf 1 cDNA, containing two 'finger' motifs, is almost identical to that of chicken or mouse. The amino-and carboxy-terminal thirds of the human protein are similar to those of mouse, but are strikingly different from the corresponding domains in chicken. The evidence indicates that these erythroid regulatory factors evolved from a common precursor composed of two distinct kinds of repeated domains, which subsequently evolved at greatly different rates.

A naturally occurring gamma globin gene mutation enhances SP1 binding activity

Transcription of the human fetal globin genes in erythroid cells is tightly regulated during different stages of development and differentiation. Two naturally occurring mutations 202 base pairs upstream of the duplicated gamma globin genes are associated with incorrectly regulated gamma globin gene gene expression; elevated levels of fetal globin are synthesized during adult life. A C-to-G base substitution upstream of the G gamma-globin gene is highly correlated with a dramatic increase in gene expression. It increases the similarity of the region to the consensus Sp1 recognition site. We determined that the mutated DNA had a 5- to 10-fold-higher affinity for Sp1 than did normal gamma globin gene sequence. We also observed a reduction in normal factor-binding activity. A different substitution at -202, C to T, upstream of the A gamma-globin gene was associated with a more moderate increase in fetal globin expression. This mutation decreased the similarity of the sequence to an Sp1 recognition site. We determined that it did not result in enhanced Sp1 binding but did alter normal factor binding. We suggest that these changes in nuclear protein-binding properties detected in vitro are responsible for the enhanced gamma globin gene expression found in -202 G gamma beta + patients with hereditary persistence of fetal hemoglobin.

The major human erythroid DNA-binding protein (GF-1): primary sequence and localization of the gene to the X chromosome

Genes expressed in erythroid cells contain binding sites for a cell-specific nuclear factor, GF-1 (NF-E1, Eryf 1), believed to be an important transcriptional regulator. Previously we characterized murine GF-1 as a 413-amino acid polypeptide containing two cysteine-cysteine regions reminiscent of zinc-finger DNA-binding domains. By cross-hybridization to the finger domain of murine GF-1 we have isolated cDNA encoding the human homolog. Peptide sequencing of purified human GF-1 confirmed the authenticity of the human cDNA. The predicted primary sequence of human GF-1 is highly similar to that of murine GF-1, particularly in the DNA-binding region. Although the DNA-binding domains of human, murine, and chicken proteins are remarkably conserved, the mammalian polypeptides are strikingly divergent from the avian counterpart in other regions, most likely those responsible for transcriptional activation. By hybridization to panels of human-rodent DNAs we have assigned the human GF-1 locus to Xp21-11. The localization of the gene to the X chromosome has important implications for hereditary persistence of fetal hemoglobin syndromes unlinked to the beta-globin cluster and for genetic experiments designed to test the role of the factor in erythroid cell gene expression.

Cell type-specific protein-DNA interactions in the human zeta-globin upstream promoter region: displacement of Sp1 by the erythroid cell-specific factor NF-E1

The protein-DNA interactions of the upstream promoter region of the human embryonic zeta-globin gene in nuclear extracts of erythroid K562 cells and nonerythroid HeLa cells were analyzed by DNase I footprinting, gel mobility shift assay, methylation interference, and oligonucleotide competition experiments. There are mainly two clusters of nuclear factor-binding sites in the zeta promoter. The proximal cluster spans the DNA sequence from -110 to -60 and consists of binding sites for CP2, Sp1, and NF-E1. NF-E1 binding is K562 specific, whereas CP2 binding is common to both types of cells. Overlapping the NF-E1- and CP2-binding sites is a hidden Sp1-binding site or CAC box, as demonstrated by binding studies of affinity-purified Sp1. In the distal promoter region at -250 to -220, another NF-E1-binding site overlaps a CAC box or Sp1-binding site. Extract-mixing experiments demonstrated that the higher affinity of NF-E1 binding excluded the binding of Sp1 in the K562 extract. NF-E1 factors could also displace prebound Sp1 molecules. Between the two clusters of multiple-factor-binding sites are sequences recognized by other factors, including zeta-globin factors 1 and 2, that are present in both HeLa and K562 extracts. We discuss the cell type-specific, competitive binding of multiple nuclear factors in terms of functional implications in transcriptional regulation of the zeta-globin gene.

Structure of the 5' flanking region of the gene encoding human glycophorin A and analysis of its multiple transcripts

Glycophorin A (GPA), the major sialoglycoprotein of human erythrocytes, is the carrier for blood group MN antigens and a receptor for viruses, bacteria and parasites. (1) Three distinct GPA mRNAs (1.0, 1.7 and 2.2 kb) have been previously identified in erythroid tissues by Northern-blot analysis. It is shown here by sequence analysis of several human fetal liver cDNAs, and by transcription start point (tsp) determination using primer extension analysis, that the production of the multiple GPA mRNAs is governed by poly(A) site choice generating 3'-untranslated regions of different length, and not by the tsp heterogeneity, since all messages exhibit the same cap site (tsp). (2) The structural gene encoding GPA has been recently cloned [Vignal et al., Eur. J. Biochem. 184 (1989) 337-344; Kudo and Fukuda, Proc. Natl. Acad. Sci. USA 86 (1989) 4619-4623] and we have now determined the sequence of a DNA genomic fragment upstream from the tsp. This fragment does not contain the typical TATA and CAAT boxes found in a number of tissue-specific genes, but contains typical motifs like the CACC, nuclear factor erythroid 1 and 2 elements, which have been identified recently in several erythroid-specific promoters, therefore suggesting that transcription of these genes might be regulated by the same or analogous factors.

Increased Sp1 binding mediates erythroid-specific overexpression of a mutated (HPFH) gamma-globulin promoter

The -198 T----C mutation in the promoter of the A gamma-globin gene increases 20-30 fold the expression of this gene in adult erythroid cells of patients (Hereditary Persistence of Fetal Hemoglobin, HPFH). We show here that this mutation creates a strong binding site, resembling a CACCC box, for two ubiquitous nuclear proteins, one of which is Sp1. The mutated promoter is four to five-fold more efficient than a normal gamma-globin promoter in driving expression of a CAT reporter plasmid when transfected into erythroid cells. The overexpression of the mutant is abolished by the introduction of an additional mutation disrupting the new binding site. No overexpression of the mutant is observed in non-erythroid cells, indicating that the ubiquitous factors bound on the mutated sequence must cooperate with erythroid specific factors.

Basal expression of the histone H5 gene is controlled by positive and negative cis-acting sequences

Sequences from -3500 to +1365 of the chicken histone H5 gene have been analyzed for the presence of cis-acting elements in H5 expressing (transformed CFU-E) and non-expressing cells (fibroblasts). The region from -3500 to -115 had little effect on transcription. Proximal upstream sequences contain a negative element (UNE, -115 to -95), capable to also repress the activity of the heterologous HSV tk promoter, and two positive elements, a consensus GC-box (-83 to -74) and a proximal element (UPE, -54 to -38). The sequence of the UPE is highly related to the histone H4 subtype-specific element and it has been conserved in the duck H5 and the human and mouse H1(0) genes at equivalent positions. Although the effect of the UNE, GC-box and UPE was not tissue-specific, sequences from -38 to +77 appear to confer a degree of tissue specificity to the promoter. An activating erythroid-specific element (DE) was found downstream of the H5 gene (+1042 to +1185). The activity of the DE was modest but independent of position and orientation and required the presence of the promoter proximal elements. The DE harbors the sequence AGATAA that is recognized by a protein factor, presumably the same that binds to other erythrocyte-specific enhancers. The low activity of DE in the CFU-E may be related to the low concentration of the AGATAA-binding factor in the differentiation-blocked cells.

The erythroid-specific transcription factor Eryf1: a new finger protein

The erythroid-specific transcription factor Eryf1 binds to DNA sites within regulatory regions of every member of both the alpha- and beta-globin families in chicken. The distribution of these sites suggests that Eryf1 may serve as a general "switch" factor for erythroid development. We have cloned the cDNA for Eryf1 and show that the corresponding mRNA is present in all erythroid lineages, but is absent from non-erythroid cells. We demonstrate that the cDNA encodes the specific Eryf1 binding activity found in erythrocytes. Eryf1 is a basic 38 kd protein containing a pair of highly similar "fingers" with the motif Cys-x-x-Cys-x17-Cys-x-x-Cys. The amino acid sequences of these regions bear no resemblance to those found in other regulatory proteins with a similar arrangement of cysteine residues. Our evidence suggests, furthermore, that transition metal ions are unusually tightly bound, or may not be necessary for the sequence-specific DNA binding of Eryf1.

The deletion of the distal CCAAT box region of the A gamma-globin gene in black HPFH abolishes the binding of the erythroid specific protein NFE3 and of the CCAAT displacement protein

Non-deletion Hereditary Persistence of Fetal Hemoglobin (HPFH) is characterized by great elevation of the synthesis, in adult age, of fetal hemoglobin (HbF), of either the A gamma or G gamma type. Strong genetic evidence indicates point mutations in the G gamma- or A gamma-globin promoter as responsible for overexpression of the mutated gene. Here we report that a 13 nucleotides deletion in the CCAAT box region of the A gamma-globin promoter, associated with greater than 100 fold overexpression of the gene, abolishes the in vitro binding of the ubiquitous factors CP1 and CDP (CCAAT displacement protein) and of the erythroid specific protein NFE3. Loss of NFE3 binding is consistent with a similar effect of the -117 G greater than A HPFH mutation, suggesting a possible role of NFE3 as a negatively acting factor. In addition, loss of CDP binding indicates that this alteration might also contribute to the HPFH phenotype in this particular case, suggesting possible heterogeneity of the mechanisms causing HPFH.

Increased erythroid-specific expression of a mutated HPFH gamma-globin promoter requires the erythroid factor NFE-1

The -175 T greater than C mutation in the promoter of the A gamma- or G gamma-globin gene causes a 50-100 fold increase of the expression of the respective gene in adult erythroid cells (Hereditary Persistence of Fetal Hemoglobin). We show here that this mutation increases 3-9 fold the expression of a gamma-CAT reporter plasmid transfected into the erythroid cells K562, but not that of the same plasmid in non erythroid cells. The overexpression of the mutant is abolished by the mutation of the binding site for the erythroid specific factor NFE1; inactivation of the adjacent binding site for the ubiquitous factor OTF1 does not cause overexpression of the normal gamma-globin promoter. Previous results demonstrated that the -175 mutation slightly increases the in vitro binding of NFE1 and almost abolishes that of OTF1; the present functional data indicate that altered binding of NFE1, but not of OTF1, is responsible for the observed overexpression of the mutated promoter.

Mutations in two regions upstream of the A gamma globin gene canonical promoter affect gene expression

Two regions upstream of the human fetal (A gamma) globin gene, which interact with protein factors from K562 and HeLa nuclear extracts, have functional significance in gene expression. One binding site (site I) is at a position -290 to -267 bp upstream of the transcription initiation site, the other (site II) is at -182 to -168 bp. Site II includes the octamer sequence (ATGCAAAT) found in an immunoglobulin enhancer and the histone H2b gene promoter. A point mutation (T----C) at -175, within the octamer sequence, is characteristic of a naturally occurring HPFH (hereditary persistence of fetal hemoglobin), and decreases factor binding to an oligonucleotide containing the octamer motif. Expression assays using a A gamma globin promoter-CAT (chloramphenicol acetyl transferase) fusion gene show that the point mutation at -175 increases expression in erythroid, but not non-erythroid cells when compared to a wild-type construct. This correlates with the actual effect of the HPFH mutation in humans. This higher expression may result from a mechanism more complex than reduced binding of a negative regulator. A site I clustered-base substitution gives gamma-CAT activity well below wild-type, suggesting that this factor is a positive regulator.

Cloning of cDNA for the major DNA-binding protein of the erythroid lineage through expression in mammalian cells

Genes expressed in erythroid cells contain binding sites for a cell-specific factor believed to be an important regulator for this haematopoietic lineage. Using high-level transient expression in mammalian cells, we have identified complementary DNA encoding the murine protein. The factor, a new member of the zinc-finger family of DNA-binding proteins, is restricted to erythroid cells at the level of RNA expression and is closely homologous between mouse and man.

DNA sequences regulating human globin gene transcription in nondeletional hereditary persistence of fetal hemoglobin

Strong genetic evidence supports the idea that point mutations in the promoter of gamma-globin genes overexpressed in adult age [hereditary persistence of fetal hemoglobin (HPFH)] are responsible for the observed phenotype. DNA binding sites for ubiquitous and/or erythroid specific nuclear proteins correlate in location with the positions of point mutations responsible for HPFH. The analysis of the effects of one of these mutations (-175 T greater than C) on in vitro binding of nuclear proteins and on the activity of the mutated promoter in transfection assays indicates that altered binding of the erythroid-specific protein NFE-1 may be responsible for increased activity of the mutated promoter. Other HPFH mutations close to the distal CCAAT box (-117 G greater than A and 13 nucleotide deletions, -114 to -102) have complex effects on in vitro binding of nuclear proteins; their only common effect is the loss of binding of the erythroid-specific factor NFE3. If mechanisms generating the HPFH phenotype are homogeneous, NFE3 might be a negatively acting factor; alternatively, heterogeneous mechanisms might operate and HPFH might additionally be related to loss of binding to the distal CCAAT box region of either NFE1 (-117 HPFH) or of the ubiquitous CCAAT displacement protein-CDP (13 nucleotides deletion). Finally, it is also proposed that increased activity of the HPFH promoters may secondarily cause decreased expression of the delta- and beta-globin genes in cis possibly by competition between gamma- and beta-globin promoters for interaction with common regulatory elements.