Two tissue-specific factors bind the erythroid promoter of the human porphobilinogen deaminase gene

Maf nuclear oncoprotein recognizes sequences related to an AP-1 site and forms heterodimers with both Fos and Jun

The v-maf oncogene, identified from AS42 avian retrovirus, encodes a nuclear bZip protein. To elucidate the molecular mechanism of cell transformation induced by this oncogene, we determined the specific binding sequences of its product. Maf protein recognized two types of relatively long palindromic consensus sequences, TGCTGACTCAGCA and TGCTGACGTCAGCA, at roughly equal efficiency. The middle parts of these Maf-binding sequences completely match with two binding sequences for AP-1 transcription factor, i.e., phorbol 12-O-tetradecanoate-13-acetate (TPA)-responsive element (TRE) and cyclic AMP responsive element, suggesting partial overlapping of the target genes for Maf and AP-1. Furthermore, Maf efficiently formed heterodimers with the components of AP-1, Fos and Jun, through their leucine zipper structures, and these heterodimers show binding specificities distinct from those for Maf-Maf and Jun-Jun homodimers. Thus, a multiple combination of the dimers should generate a greatly expanded repertoire of transcriptional regulatory potential. DNA data base search for the Maf-binding consensus sequences suggested that some of the TRE-like cis elements reported previously may actually be the targets for Maf family proteins or their heterodimers with other bZip proteins.

Developmental regulation of the human embryonic beta-like globin gene is mediated by synergistic interactions among multiple tissue- and stage-specific elements

The stage-specific regulation of mammalian embryonic globin genes has been an experimentally elusive problem, in part because of the developmentally early timing of their expression. We have carried out a systematic analysis of truncation and internal deletion mutations within the 5'-flanking region of the human embryonic beta-like globin gene (epsilon) in erythroid and nonerythroid cell lines. Within a 670-bp region upstream from the constitutive promoter are multiple positive and negative control elements. Of these, a positive regulatory element (epsilon-PRE II) which is active only in embryonic erythroid cells is of particular interest. Remarkably, although it is inactive on its own, in the presence of other sequences located further upstream, it confers tissue- and developmental stage-specific expression on a constitutive epsilon-globin or heterologous promoter. The activity of epsilon-PRE II is also modulated by another positive regulatory domain located further downstream to direct erythroid cell-specific, but little or no embryonic stage-specific, transcription. A nuclear factor highly enriched in embryonic erythroid cells binds specifically within a 19-bp region of epsilon-PRE II. Nuclei from adult erythroid cells also contain a factor that binds to this region but forms a complex of faster electrophoretic mobility. We speculate that interactions between epsilon-PRE II and other upstream control elements play an important role in the developmental regulation of the human embryonic beta-like globin gene.

Molecular basis of acute intermittent porphyria: mutations and polymorphisms in the human hydroxymethylbilane synthase gene

Acute intermittent porphyria (AIP) is an autosomal dominant inborn error of metabolism that results from the half-normal activity of the third enzyme in the heme biosynthetic pathway, hydroxymethylbilane synthase (HMB-synthase). AIP is an ecogenetic condition, with life-threatening acute attacks precipitated by various factors including drugs, alcohol, fasting, and certain hormones. Biochemical diagnosis is problematic and the identification of mutations in the HMB-synthase gene provides accurate detection of presymptomatic heterozygotes, permitting avoidance of the acute precipitating factors. Two HMB-synthase isozymes are encoded by the HMB-synthase gene: one unique to erythroid cells and the other a housekeeping isozyme present in all cells. These two isozymes arise from a single gene by alternative splicing. The recent isolation of the cDNAs and entire genomic sequence encoding the HMB-synthase isozymes has facilitated the detection of diagnostically useful intragenic polymorphisms and disease-causing mutations. Of the 36 mutations identified to date, most caused the classic form of AIP. These mutations included small deletions and insertions, point mutations and RNA splice junction alterations and resulted in the half-normal activity of both the erythroid-specific and housekeeping isozymes. Most AIP mutations were private; however, certain mutations were frequently found in Dutch (R116W) and Swedish (W198X) AIP families. A variant form of AIP, in which patients have normal erythroid activity, but half-normal activity of the housekeeping isozyme, resulted from two mutations at the exon 1/intron 1 boundary, each altering splicing of the hepatic-specific transcript. In addition, 10 polymorphisms in the HMB-synthase gene have been identified that are useful for the diagnosis of presymptomatic AIP heterozygotes in families whose specific mutations have not been determined.

Maf nuclear oncoprotein recognizes sequences related to an AP-1 site and forms heterodimers with both Fos and Jun

The v-maf oncogene, identified from AS42 avian retrovirus, encodes a nuclear bZip protein. To elucidate the molecular mechanism of cell transformation induced by this oncogene, we determined the specific binding sequences of its product. Maf protein recognized two types of relatively long palindromic consensus sequences, TGCTGACTCAGCA and TGCTGACGTCAGCA, at roughly equal efficiency. The middle parts of these Maf-binding sequences completely match with two binding sequences for AP-1 transcription factor, i.e., phorbol 12-O-tetradecanoate-13-acetate (TPA)-responsive element (TRE) and cyclic AMP responsive element, suggesting partial overlapping of the target genes for Maf and AP-1. Furthermore, Maf efficiently formed heterodimers with the components of AP-1, Fos and Jun, through their leucine zipper structures, and these heterodimers show binding specificities distinct from those for Maf-Maf and Jun-Jun homodimers. Thus, a multiple combination of the dimers should generate a greatly expanded repertoire of transcriptional regulatory potential. DNA data base search for the Maf-binding consensus sequences suggested that some of the TRE-like cis elements reported previously may actually be the targets for Maf family proteins or their heterodimers with other bZip proteins.

The ubiquitous subunit of erythroid transcription factor NF-E2 is a small basic-leucine zipper protein related to the v-maf oncogene

Erythroid transcription factor NF-E2 is a tissue-restricted heterodimeric protein which recognizes an extended AP-1 motif [(T/C)TGCTGA(C/G)TCA(T/C)] found in the upstream locus control regions of the alpha- and beta-globin gene clusters. A cDNA clone encoding a cell-type-specific subunit of NF-E2, designated p45 NF-E2, has previously been characterized and shown to encode a basic-leucine zipper DNA-binding protein. Here we describe protein purification and cloning of cDNA that encodes the second basic-leucine zipper subunit of the native NF-E2 heterodimer. This polypeptide, designated p18, is widely expressed. It displays extensive homology to the v-maf oncogene product and a human retinal-specific protein, NRL. Unusual features in the basic region shared by v-Maf, NRL, and p18 place them in a distinct subfamily of AP-1-like proteins.

Developmental regulation of the human embryonic beta-like globin gene is mediated by synergistic interactions among multiple tissue- and stage-specific elements

The stage-specific regulation of mammalian embryonic globin genes has been an experimentally elusive problem, in part because of the developmentally early timing of their expression. We have carried out a systematic analysis of truncation and internal deletion mutations within the 5'-flanking region of the human embryonic beta-like globin gene (epsilon) in erythroid and nonerythroid cell lines. Within a 670-bp region upstream from the constitutive promoter are multiple positive and negative control elements. Of these, a positive regulatory element (epsilon-PRE II) which is active only in embryonic erythroid cells is of particular interest. Remarkably, although it is inactive on its own, in the presence of other sequences located further upstream, it confers tissue- and developmental stage-specific expression on a constitutive epsilon-globin or heterologous promoter. The activity of epsilon-PRE II is also modulated by another positive regulatory domain located further downstream to direct erythroid cell-specific, but little or no embryonic stage-specific, transcription. A nuclear factor highly enriched in embryonic erythroid cells binds specifically within a 19-bp region of epsilon-PRE II. Nuclei from adult erythroid cells also contain a factor that binds to this region but forms a complex of faster electrophoretic mobility. We speculate that interactions between epsilon-PRE II and other upstream control elements play an important role in the developmental regulation of the human embryonic beta-like globin gene.

Activation of the beta-globin promoter by the locus control region correlates with binding of a novel factor to the CAAT box in murine erythroleukemia cells but not in K562 cells

Four distinct factors in extracts from murine erythroleukemia (MEL) cells interacted with the human beta-globin gene promoter CAAT box: CP1, GATA-1, and two novel factors, denoted a and b, one of which is highly inducible in the MEL system. GATA-1 binding to the CAAT element was very unstable (half-life < 1 min), whereas bindings of a, b, and CP1 were comparatively stable, with half-lives of 18, 19, and 3.5 min, respectively. Stable transfections of MEL cells showed that in the presence of the beta-globin locus control region (LCR), the wild-type CAAT box, a mutant which bound to GATA-1 with increased stability over the normal sequences, and a mutant which bound a, b, and CP1 specifically could all stimulate transcription greater than ninefold over that induced by a null CAAT mutation in both uninduced and terminally differentiated MEL cells. A mutant which bound the a and b factors specifically gave only a twofold stimulation of promoter activity, and this lower activity correlated with a decrease in the stability of binding of the b protein. On the other hand, CP1 binding alone did not stimulate transcription. Taken together, these results suggest that in the context of the wild-type beta-globin CAAT element the b factor stimulates transcription directed by the LCR in MEL cells, although the LCR can also function through more stable GATA-1-binding sequences. However, in K562 cells, the wild-type beta-globin CAAT box alone was unable to stimulate gene expression directed by the LCR and high levels of transcription were obtained only upon inclusion of more upstream beta-globin promoter sequences. In contrast, a construct containing only the A gamma-globin CAAT box region did give high expression levels in K562 cells. Thus, there is a fundamental difference in the way the LCR functions in these two model systems in terms of its requirements at the promoter level.

Activation of the beta-globin promoter by the locus control region correlates with binding of a novel factor to the CAAT box in murine erythroleukemia cells but not in K562 cells

Four distinct factors in extracts from murine erythroleukemia (MEL) cells interacted with the human beta-globin gene promoter CAAT box: CP1, GATA-1, and two novel factors, denoted a and b, one of which is highly inducible in the MEL system. GATA-1 binding to the CAAT element was very unstable (half-life < 1 min), whereas bindings of a, b, and CP1 were comparatively stable, with half-lives of 18, 19, and 3.5 min, respectively. Stable transfections of MEL cells showed that in the presence of the beta-globin locus control region (LCR), the wild-type CAAT box, a mutant which bound to GATA-1 with increased stability over the normal sequences, and a mutant which bound a, b, and CP1 specifically could all stimulate transcription greater than ninefold over that induced by a null CAAT mutation in both uninduced and terminally differentiated MEL cells. A mutant which bound the a and b factors specifically gave only a twofold stimulation of promoter activity, and this lower activity correlated with a decrease in the stability of binding of the b protein. On the other hand, CP1 binding alone did not stimulate transcription. Taken together, these results suggest that in the context of the wild-type beta-globin CAAT element the b factor stimulates transcription directed by the LCR in MEL cells, although the LCR can also function through more stable GATA-1-binding sequences. However, in K562 cells, the wild-type beta-globin CAAT box alone was unable to stimulate gene expression directed by the LCR and high levels of transcription were obtained only upon inclusion of more upstream beta-globin promoter sequences. In contrast, a construct containing only the A gamma-globin CAAT box region did give high expression levels in K562 cells. Thus, there is a fundamental difference in the way the LCR functions in these two model systems in terms of its requirements at the promoter level.

Purification of the human NF-E2 complex: cDNA cloning of the hematopoietic cell-specific subunit and evidence for an associated partner

The human globin locus control region-binding protein, NF-E2, was purified by DNA affinity chromatography. Its tissue-specific component, p45 NF-E2, was cloned by use of a low-stringency library screen with murine p45 NF-E2 cDNA (N. C. Andrews, H. Erdjument-Bromage, M. B. Davidson, P. Tempst, and S. H. Orkin, Nature [London] 362:722-728, 1993). The human p45 NF-E2 gene was localized to chromosome 12q13 by fluorescent in situ hybridization. Human p45 NF-E2 and murine p45 NF-E2 are highly homologous basic region-leucine zipper (bZIP) proteins with identical DNA-binding domains. Immunoprecipitation experiments demonstrated that p45 NF-E2 is associated in vivo with an 18-kDa protein (p18). Because bZIP proteins bind DNA as dimers, we infer that native NF-E2 must be a heterodimer of 45- and 18-kDa subunits. Although AP-1 and CREB copurified with NF-E2, no evidence was found for heterodimer formation between p45 NF-E2 and proteins other than p18. Thus, p18 appears to be the sole specific partner of p45 NF-E2 in erythroid cells. Cloning of human p45 NF-E2 should permit studies of the role of NF-E2 in globin gene regulation and erythroid differentiation.

Purification of the human NF-E2 complex: cDNA cloning of the hematopoietic cell-specific subunit and evidence for an associated partner

The human globin locus control region-binding protein, NF-E2, was purified by DNA affinity chromatography. Its tissue-specific component, p45 NF-E2, was cloned by use of a low-stringency library screen with murine p45 NF-E2 cDNA (N. C. Andrews, H. Erdjument-Bromage, M. B. Davidson, P. Tempst, and S. H. Orkin, Nature [London] 362:722-728, 1993). The human p45 NF-E2 gene was localized to chromosome 12q13 by fluorescent in situ hybridization. Human p45 NF-E2 and murine p45 NF-E2 are highly homologous basic region-leucine zipper (bZIP) proteins with identical DNA-binding domains. Immunoprecipitation experiments demonstrated that p45 NF-E2 is associated in vivo with an 18-kDa protein (p18). Because bZIP proteins bind DNA as dimers, we infer that native NF-E2 must be a heterodimer of 45- and 18-kDa subunits. Although AP-1 and CREB copurified with NF-E2, no evidence was found for heterodimer formation between p45 NF-E2 and proteins other than p18. Thus, p18 appears to be the sole specific partner of p45 NF-E2 in erythroid cells. Cloning of human p45 NF-E2 should permit studies of the role of NF-E2 in globin gene regulation and erythroid differentiation.

Transcription of the histone H5 gene is regulated by three differentiation-specific enhancers

Histone H5, an early marker of the avian erythroid lineage, is expressed at low levels in early erythroid precursors and at higher levels in more mature cells. We show that the increase in H5 expression is due to transcriptional activation of the H5 gene following differentiation of precursor CFU(E). We have found and characterized two upstream enhancers, E1 (between -2233 and -1878 from the site of transcription initiation, +1) and E3 (between -1321 and -1163), and confirmed the presence of a downstream enhancer (C. D. Trainor, S. J. Stamler, and J. D. Engel, Nature [London] 328:827-830, 1987) E7 (between +846 and +1181) which are responsible for the increase in H5 gene transcription. The enhancers had a weak effect in nondifferentiated CFU(E) but a strong effect when the cells were induced to differentiate. Cooperation among the three enhancers, however, was not required for H5 gene activity in the differentiated cells. The enhancers contain binding sites for several ubiquitous and erythroid cell-specific nuclear proteins, including GATA-1, as demonstrated with GATA-1-specific antibodies. Although the GATA sites were required for enhancer function, the concentration of GATA-1, GATA-2, and GATA-3 decreased during cell differentiation, and overexpression of these factors had little effect on H5 transcription. Hence, the differentiation-specific effect of the enhancers is not mediated by changes in relative levels of the GATA factors. Functional analysis of the H5 promoter indicated that the requirement of several elements, including a GC box necessary for transcription enhancement, did not change during the early stages of CFU(E) differentiation. However, the UPE, a positive element in proliferating CFU(E) recognized by the transcription factor H4TF2, was dispensable in the differentiated cells. These results suggest that as the cells enter the final stages of differentiation, there is a reprogramming of the regulatory factors that control H5 transcription and that the enhancers rescue and increase the activity of the promoter.

Transcription of the histone H5 gene is regulated by three differentiation-specific enhancers

Histone H5, an early marker of the avian erythroid lineage, is expressed at low levels in early erythroid precursors and at higher levels in more mature cells. We show that the increase in H5 expression is due to transcriptional activation of the H5 gene following differentiation of precursor CFU(E). We have found and characterized two upstream enhancers, E1 (between -2233 and -1878 from the site of transcription initiation, +1) and E3 (between -1321 and -1163), and confirmed the presence of a downstream enhancer (C. D. Trainor, S. J. Stamler, and J. D. Engel, Nature [London] 328:827-830, 1987) E7 (between +846 and +1181) which are responsible for the increase in H5 gene transcription. The enhancers had a weak effect in nondifferentiated CFU(E) but a strong effect when the cells were induced to differentiate. Cooperation among the three enhancers, however, was not required for H5 gene activity in the differentiated cells. The enhancers contain binding sites for several ubiquitous and erythroid cell-specific nuclear proteins, including GATA-1, as demonstrated with GATA-1-specific antibodies. Although the GATA sites were required for enhancer function, the concentration of GATA-1, GATA-2, and GATA-3 decreased during cell differentiation, and overexpression of these factors had little effect on H5 transcription. Hence, the differentiation-specific effect of the enhancers is not mediated by changes in relative levels of the GATA factors. Functional analysis of the H5 promoter indicated that the requirement of several elements, including a GC box necessary for transcription enhancement, did not change during the early stages of CFU(E) differentiation. However, the UPE, a positive element in proliferating CFU(E) recognized by the transcription factor H4TF2, was dispensable in the differentiated cells. These results suggest that as the cells enter the final stages of differentiation, there is a reprogramming of the regulatory factors that control H5 transcription and that the enhancers rescue and increase the activity of the promoter.

Expression of ferrochelatase mRNA in erythroid and non-erythroid cells

Ferrochelatase, which catalyses the last step in haem biosynthesis, i.e. the insertion of Fe(II) into protophorphyrin IX, is present in all cells, but is particularly abundant in erythroid cells during haemoglobinization. Using mouse ferrochelatase cDNA as a probe two ferrochelatase transcripts, having lengths of 2.9 kb and 2.2 kb, were found in extracts of mouse liver, kidney, brain, muscle and spleen, the 2.9 kb transcript being more abundant in the non-erythroid tissues and the 2.2 kb transcript more predominant in spleen. In mouse erythroleukemia cells the 2.9 kb ferrochelatase transcript is also more abundant; however, following induction of erythroid differentiation by dimethyl sulphoxide there is a preferential increase in the 2.2 kb transcript, which eventually predominates. With mouse reticulocytes, the purest immature erythroid cell population available, over 90% of the total ferrochelatase mRNA is present as the 2.2 kb transcript. Since there is probably only one mouse ferrochelatase gene, the occurrence of two ferrochelatase transcripts could arise from the use of two putative polyadenylation signals in the 3' region of ferrochelatase DNA. This possibility was explored by using a 389 bp DNA fragment produced by PCR with synthetic oligoprimers having sequence similarity with a region between the polyadenylation sites. This fragment hybridized only to the 2.9 kb ferrochelatase transcript, indicating that the two transcripts differ at their 3' ends and suggesting that the 2.2 kb transcript results from the utilization of the upstream polyadenylation signal. The preferential utilization of the upstream polyadenylation signal may be an erythroid-specific characteristic of ferrochelatase gene expression.

Regulation of heme biosynthesis: distinct regulatory features in erythroid cells

Our previous research has demonstrated that in hemoglobin-synthesizing cells, as compared with nonerythroid cells, a step in iron transport from transferrin localized between the transferrin receptor and ferrochelatase is rate-limiting for the synthesis of heme. In this communication we report our more recent studies on the mechanisms involved in the regulation of the transferrin receptors and ferrochelatase in differentiating erythroid cells. Our studies indicate that transferrin receptor gene expression is regulated differently in hemoglobin synthesizing as compared with uninduced murine erythroleukemia (MEL) cells: 1) With nuclear run-on assays our experiments showed increased transferrin receptor mRNA transcription cells of MEL following induction of erythroid differentiation with dimethylsulfoxide (DMSO). 2) DMSO treatment of MEL cells does not increase iron-responsive element binding protein (IRE-BP) activity which is, however, increased in uninduced MEL cells by Fe chelators. 3) Following induction of MEL cells there is an increase in the stability of transferrin receptor mRNA whose level is only slightly affected by iron excess. Using murine ferrochelatase cDNA as a probe, two ferrochelatase transcripts having lengths of 2.9 kb and 2.2 kb were found in extracts of mouse liver, kidney, brain, muscle and spleen, the 2.9 kb transcript being more abundant in nonerythroid tissues and the 2.2 more predominant in spleen. In MEL cells, the 2.9 ferrochelatase transcript is also more abundant; however, following induction of erythroid differentiation by DMSO there is a preferential increase in the 2.2 kb transcript which eventually predominates. With mouse reticulocytes, the purest immature erythroid cell population available, over 90% of the total ferrochelatase mRNA is present as the 2.2 kb transcript. Our further experiments indicate that the 2.2 kb transcript results from the utilization of the upstream polyadenylation signal and suggest that the preferential utilization of the upstream polyadenylation signal may be an erythroid-specific characteristic of ferrochelatase gene expression. These results provide further evidence for the idea that iron metabolism and heme synthesis are controlled by distinct mechanisms in erythroid versus nonerythroid cells.

Mouse microcytic anaemia caused by a defect in the gene encoding the globin enhancer-binding protein NF-E2

The nuclear DNA-binding protein NF-E2 is thought to mediate the powerful erythroid enhancer activity of the alpha- and beta-globin locus control regions and participates in the control of genes encoding two enzymes of haem biosynthesis (porphobilinogen deaminase and ferrochelatase). The major component of NF-E2 is a 45K polypeptide (designated p45 NF-E2) that belongs to the basic region-leucine zipper family of transcription factors. This subunit of NF-E2 is specifically expressed in haematopoietic progenitor cells and differentiated cells of the erythroid, megakaryocyte and mast cell lineages. The gene encoding p45 NF-E2 (murine gene Nfe2) has been mapped to mouse chromosome 15 near the mutation microcytosis (mk). Homozygous mk mice have severe hypochromic microcytic anaemia as a result of decreased globin synthesis and defects in intestinal and erythroid iron absorption. Here we investigate whether the mk mutation lies within Nfe2 by characterizing the p45 NF-E2 gene and determining its DNA sequence in wild-type and mk alleles. The mk allele carries a missense mutation that causes substitution of valine by alanine at amino acid 173 of the p45 NF-E2 protein. Expression of p45 NF-E2 messenger RNA was detected in erythroid tissues of normal mice and in the duodenum of normal and severely anaemic beta-thalassaemic (Hbbd-th3/Hbbd-th3) mice. We propose that the mk mutation results in an impaired form of NF-E2 which fails to regulate both globin production and iron metabolism properly.

Erythroid transcription factor NF-E2 is a haematopoietic-specific basic-leucine zipper protein

Expression of globin genes in developing erythroid cells is controlled by upstream locus control regions. Activity of these regions in vivo requires an erythroid-specific nuclear factor (NF-E2) that binds AP-1-like recognition sites. Its tissue-specific component (p45 NF-E2) has been characterized by complementary DNA cloning as a new basic region-leucine zipper protein which dimerizes with a ubiquitous partner to form native NF-E2.

Mouse GATA-4: a retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and heart

We report the cDNA cloning and characterization of mouse GATA-4, a new member of the family of zinc finger transcription factors that bind a core GATA motif. GATA-4 cDNA was identified by screening a 6.5-day mouse embryo library with oligonucleotide probes corresponding to a highly conserved region of the finger domains. Like other proteins of the family, GATA-4 is approximately 50 kDa in size and contains two zinc finger domains of the form C-X-N-C-(X17)-C-N-X-C. Cotransfection assays in heterologous cells demonstrate that GATA-4 trans activates reporter constructs containing GATA promoter elements. Northern (RNA) analysis and in situ hybridization show that GATA-4 mRNA is expressed in the heart, intestinal epithelium, primitive endoderm, and gonads. Retinoic acid-induced differentiation of mouse F9 cells into visceral or parietal endoderm is accompanied by increased expression of GATA-4 mRNA and protein. In vitro differentiation of embryonic stem cells into embryoid bodies is also associated with increased GATA-4 expression. We conclude that GATA-4 is a tissue-specific, retinoic acid-inducible, and developmentally regulated transcription factor. On the basis of its tissue distribution, we speculate that GATA-4 plays a role in gene expression in the heart, intestinal epithelium, primitive endoderm, and gonads.

Mouse GATA-4: a retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and heart

We report the cDNA cloning and characterization of mouse GATA-4, a new member of the family of zinc finger transcription factors that bind a core GATA motif. GATA-4 cDNA was identified by screening a 6.5-day mouse embryo library with oligonucleotide probes corresponding to a highly conserved region of the finger domains. Like other proteins of the family, GATA-4 is approximately 50 kDa in size and contains two zinc finger domains of the form C-X-N-C-(X17)-C-N-X-C. Cotransfection assays in heterologous cells demonstrate that GATA-4 trans activates reporter constructs containing GATA promoter elements. Northern (RNA) analysis and in situ hybridization show that GATA-4 mRNA is expressed in the heart, intestinal epithelium, primitive endoderm, and gonads. Retinoic acid-induced differentiation of mouse F9 cells into visceral or parietal endoderm is accompanied by increased expression of GATA-4 mRNA and protein. In vitro differentiation of embryonic stem cells into embryoid bodies is also associated with increased GATA-4 expression. We conclude that GATA-4 is a tissue-specific, retinoic acid-inducible, and developmentally regulated transcription factor. On the basis of its tissue distribution, we speculate that GATA-4 plays a role in gene expression in the heart, intestinal epithelium, primitive endoderm, and gonads.

CCACC-binding or simian-virus-40-protein-1-binding proteins cooperate with human GATA-1 to direct erythroid-specific transcription and to mediate 5' hypersensitive site 2 sensitivity of a TATA-less promoter

Previous studies have shown that a -112 to +78 DNA fragment from the erythroid promoter of the human porphobilinogen deaminase (PBGD) gene has erythroid-specific activity. This PBGD-(-112 to +78) promoter contains a CCACC binding site (position -100), a GATA binding site (position -70) and an initiator element around the cap site. Using a cotransfection assay, we find that the human factor GATA-1 trans-activates the PBGD-(-112 to +78) promoter in non-erythroid cells. We show that, if trans-activation is abolished by mutations that destroy either the -100 CCACC binding or the -70 GATA binding sites, replacement of the -100 CCACC binding site by a simian-virus-40-protein-1 (Sp1) binding site maintains both the erythroid-specific activity of this promoter and the human GATA-1 trans-activation. Thus, human GATA-1 acts on the PBGD promoter in association with Sp1 or CCACC binding proteins. This PBGD-(-112 to +78) promoter is activated 20-fold by a cis-linked 5' hypersensitive site 2 (5'HS-2) of the human beta-globin locus control region. This activation depends on the -70 GATA and -100 CCACC or Sp1 binding sites. When a longer -714 to +78 fragment of the PBGD promoter is used, the -70 GATA mutant still displays erythroid-specific activity and is cis-activated by the 5'HS-2 enhancer, while the -100 CCACC mutant is completely inactive in the absence or in the presence of the 5'HS-2 enhancer. Thus, the -100 CCACC binding site is indispensable for the correct activity and sensitivity of the human PBGD promoter to the 5'HS-2 enhancer, whereas the -70 GATA binding site can functionally be replaced by upstream cis-acting elements.

A small single-"finger" peptide from the erythroid transcription factor GATA-1 binds specifically to DNA as a zinc or iron complex

Sequence-specific DNA binding has been demonstrated for a synthetic peptide comprising only one of the two "finger"-like domains of the erythroid transcription factor GATA-1 (also termed Eryf-1, NF-E1, or GF-1). Quantitative analysis of gel-retardation assays yields a specific association constant of 1.2 x 10(8) M, compared with values of about 10(9) M for the full-length natural GATA-1 protein. By the use of peptides of various lengths, it was possible to delineate the smallest region necessary for specific binding. A single C-terminal finger of the double-finger motif is necessary but not sufficient for sequence-specific interaction. Basic amino acids located C-terminal to the finger (some more than 20 amino acids away) are also essential for tight binding. In addition to demonstrating that zinc is important for the formation of an active binding complex, we show that other ions, notably Fe2+, can fulfill this role. Our results make it clear that the GATA-1 metal binding motif is quite distinct from that found in the steroid hormone family and that GATA-1 is a member of a separate class of DNA binding proteins.

Erythroid regulatory elements

Erythroid differentiation leads to the production of red blood cells that contain a high level of hemoglobin. This level is mainly regulated by globin gene transcription during development and differentiation. Although numerous cis-acting sequences are involved in transcriptional activity of globin genes, combinations of three motifs, CCACC, SP1 and GATA represent the core elements of their regulatory sequences. These combinations are also found in promoters and/or enhancers of non-globin genes specifically expressed in the late stages of erythroid differentiation. The CCACC and SP1 sequences bind proteins that do not display erythrocytic specificity, and the GATA sequences bind a family of transacting factors recently cloned. The GATA family members are distinctive for a highly homologous DNA binding domain that exists in two zinc fingers reminiscent of those of the glucocorticoid receptor. None of the GATA family members displays only erythroid specificity, but gene disruption followed by rescue indicates that GATA-1 is necessary for terminal erythroid differentiation throughout development. The GATA/SP1 and GATA/CCACC associations are present in positive, negative or inducible regulatory sequences suggesting that other elements control the fine tuning of erythroid gene expression. NF-E2, which is a major transcriptional activator, members of the ets family which are implicated in the early stages of erythropoiesis and finally c-erbA which directly regulates a set of erythroid-specific genes are proteins that bind these latter regulatory motifs.

The regulation of human globin gene expression

The haemopoietic system provides a well-characterized and accessible system for studying the mechanisms of developmental regulation and differentiation in higher eukaryotes. Our current understanding of the steps involved in the early stages of differentiation are poorly understood but a great deal is now known about the mechanisms by which globin expression is regulated in cells committed to the erythroid lineage. Many of the critical cis-acting sequences and some of the important trans-acting factors involved have been identified and current work is focusing on how these interact to produce high levels of tissue-specific and developmentally regulated expression of the human globin genes.

Enhancer-dependent transcription of the epsilon-globin promoter requires promoter-bound GATA-1 and enhancer-bound AP-1/NF-E2

We analyzed epsilon-globin transcription in erythroid cells and in erythroid extracts to determine the requirements for enhancer-dependent expression of this gene. Mutations that abolished GATA-1 binding at a single position in the promoter prevented interaction with enhancers, whereas elimination of a second more distal promoter GATA-1 site had no effect. Deletion or mutation of the GATA-1 sites in either the human beta-globin locus control region DNase-hypersensitive site II enhancer or the chicken beta A/epsilon-globin enhancer did not diminish the ability of the enhancers to interact with the promoter. In contrast, mutation of the AP-1/NF-E2 sites in these enhancers resulted in elimination of enhancement. In vitro transcription of these constructs was promoter dependent and was not sensitive to abolition of GATA-1 binding in the promoter, consistent with the role of GATA-1 solely as a mediator of the enhancer effect. Thus, GATA-1 regulates the response of the epsilon-globin gene to enhancers through a specific site in the promoter and requires enhancer AP-1/NF-E2 binding to transduce the enhancer effect on transcription.

Enhancer-dependent transcription of the epsilon-globin promoter requires promoter-bound GATA-1 and enhancer-bound AP-1/NF-E2

We analyzed epsilon-globin transcription in erythroid cells and in erythroid extracts to determine the requirements for enhancer-dependent expression of this gene. Mutations that abolished GATA-1 binding at a single position in the promoter prevented interaction with enhancers, whereas elimination of a second more distal promoter GATA-1 site had no effect. Deletion or mutation of the GATA-1 sites in either the human beta-globin locus control region DNase-hypersensitive site II enhancer or the chicken beta A/epsilon-globin enhancer did not diminish the ability of the enhancers to interact with the promoter. In contrast, mutation of the AP-1/NF-E2 sites in these enhancers resulted in elimination of enhancement. In vitro transcription of these constructs was promoter dependent and was not sensitive to abolition of GATA-1 binding in the promoter, consistent with the role of GATA-1 solely as a mediator of the enhancer effect. Thus, GATA-1 regulates the response of the epsilon-globin gene to enhancers through a specific site in the promoter and requires enhancer AP-1/NF-E2 binding to transduce the enhancer effect on transcription.

Murine erythroleukemia cells contain two distinct GATA-binding proteins that have different patterns of expression during cellular differentiation

GATA-1 is a major transcription factor of the erythroid lineage that has been implicated in the induced expression of a variety of red cell-specific genes during terminal differentiation of murine erythroleukemia cells. Although the GATA-1 protein is present at nearly equal levels before and after differentiation of murine erythroleukemia cells, in this study it was found that in the early commitment stages of the differentiation program there is a transient decrease in the GATA-1 mRNA and DNA binding activity levels due to a temporary block in transcription of the gene. Moreover, using a whole cell extraction procedure it was discovered that murine erythroleukemia cells contain a second GATA binding activity (denoted GATA-rel) which appears to be distinct from the GATA-1 factor based on its non-reactivity to two GATA-1 antisera. This protein has a limited tissue specificity, as it could not be detected in extracts from CHO, NIH 3T3, or COS cells. Similarly to the GATA-1 DNA-binding activity, the GATA-rel activity decreased during the early stages of differentiation. However, unlike GATA-1, GATA-rel activity did not return to pre-induced levels at later times. These results suggest that changes in gene expression during erythroid terminal differentiation may involve an interplay on levels of different GATA-binding factors.

Transcription of the hypersensitive site HS2 enhancer in erythroid cells

In the human genome, the erythroid-specific hypersensitive site HS2 enhancer regulates the transcription of the downstream beta-like globin genes 10-50 kilobases away. The mechanism of HS2 enhancer function is not known. The present study employs RNA protection assays to analyze the transcriptional status of the HS2 enhancer in transfected recombinant chloramphenicol acetyltransferase (CAT) plasmids. In erythroid K562 cells in which the HS2 enhancer is active, the HS2 sequence directs the synthesis of long enhancer transcripts that are initiated apparently from within the enhancer and elongated through the intervening DNA into the cis-linked CAT gene. In nonerythroid HL-60 cells in which the HS2 enhancer is inactive, long enhancer transcripts are not detectable. Splitting the HS2 enhancer between two tandem Ap1 sites abolishes the synthesis of a group of long enhancer transcripts and results in loss of enhancer function and transcriptional silencing of the cis-linked CAT gene. In directing the synthesis of RNA through the intervening DNA and the gene by a tracking and transcription mechanism, the HS2 enhancer may (i) open up the chromatin structure of a gene domain and (ii) deliver enhancer binding proteins to the promoter sequence where they may stimulate the transcription of the gene at the cap site.

In vitro and in vivo protein--DNA interactions on the rat erythroid-specific L' pyruvate kinase gene promoter

The rat L-type pyruvate kinase gene possesses two alternative tissue-specific promoters, located 472 bp apart; the upstream L' promoter is erythroid-specific and the downstream L promoter is hepatocyte-specific. The erythroid-specific L' promoter is strongly active in fetal liver at day 17 of gestation, while its activity rapidly decreases thereafter. A L' promoter fragment spanning from nucleotide -320 to +10 with respect to the cap site is able to direct a weak but erythroid-specific transcription in a cell-free system. We have used DNAse I footprinting and gel mobility shift assays to characterize and identify the binding of nuclear factors from both 17-day-old fetal liver and adult liver nuclear extracts to a 320 bp fragment of the 5' flanking region of the gene in vitro. Two clusters of erythroid-specific interactions were found. The proximal cluster consists of two GATA-1 binding sites at -50 bp and -65 bp from the transcription initiation site, immediately downstream of a CACC motif and two G/C-rich elements. The distal cluster of cis-elements, located 130 bp upstream, corresponds to two GATA-1 sequences. These two sequences overlap NF1 motifs interacting with ubiquitous NF1 transcriptional factors in presence of adult hepatic extracts. Furthermore, we have examined in vivo protein-DNA interactions by DMS footprinting in livers of 17-day-old rat fetuses and adult rats. We found that the sites characterized in vitro are occupied in vivo. Therefore, in adult liver the L' promoter, although inactive, nevertheless interacts with ubiquitous factors.

Functional erythroid promoters created by interaction of the transcription factor GATA-1 with CACCC and AP-1/NFE-2 elements

We have investigated interactions between the erythroid transcription factor GATA-1 and factors binding two cis-acting elements commonly linked to GATA sites in erythroid control elements. GATA-1 is present at all stages of erythroid differentiation, is necessary for erythropoiesis, and binds sites in all erythroid control elements. However, minimal promoters containing GATA-1 sites are inactive when tested in erythroid cells. Based on this observation, two erythroid cis elements, here termed CACCC and AP-1/NFE-2, were linked to GATA sites in minimal promoters. None of the elements linked only to a TATA box created an active promoter, but GATA sites linked to either CACCC or AP-1/NFE-2 elements formed strong erythroid promoters. A mutation of T to C at position -175 in the gamma-globin promoter GATA site, associated with hereditary persistence of fetal hemoglobin (HPFH), increased expression of these promoters in both fetal and adult cells. A construct bearing the beta-globin CACCC element was more active in adult and less active in fetal erythroid cells, when compared with the gamma-globin CACCC element. These studies suggest that erythroid control elements are formed by the interactions of at least three transcription factors, none of which functions alone.

Structural analysis of human pyruvate kinase L-gene and identification of the promoter activity in erythroid cells

The human pyruvate kinase (PK) L-gene is organized in 12 exons over 9.5 kilobases, and the first and second exons are specifically transcribed to the R- and L-type PK mRNA. The 5'-flanking region upstream the first exon has two CAC boxes and four GATA motifs within 250 bp from the translational initiation codon. Comparison with the rat L-gene revealed four well-conserved elements in the region. The transient transfection demonstrated that the 270-bp upstream region was a powerful promoter in K562 cells, whereas deletion of the distal 150-bp sequences, which included three GATA motifs, resulted in drastic reduction of the activity. When the hypersensitive site 2 of the human beta-globin gene locus was joined to the promoter, the activity of the proximal 120-bp region was enhanced. We concluded that the proximal 120-bp region had a basal promoter activity and that the distal 150-bp region acted as an enhancer in erythroid cells.

In vivo protein-DNA interactions at hypersensitive site 3 of the human beta-globin locus control region

The expression of beta-globin genes in developing erythroid cells is dependent on distant, upstream regulatory sequences, known as the locus control region (LCR), which are marked in chromatin by DNase I hypersensitive sites (HS-1 to HS-4). Linkage of the beta-globin gene complex LCR or fragments surrounding core regions of 200-300 base pairs to the human beta-globin gene permits consistent, high-level expression of the transgene in mice. To define the array of nuclear factors interacting with beta-LCR HS-3, we have performed in vivo dimethyl sulfate footprinting of the active HS-3 core in erythroid cells by a modified procedure that permits assessment of protein-DNA contacts at adenine, as well as guanine, residues. In vivo protein occupancy differs considerably from that predicted from previous in vitro binding analyses. In vivo footprinting detects protein binding at four sites recognized by the erythroid transcription factor GATA-1, at two CACC/GT motifs, and at a single AP-1/NF-E2 site. The regulatory elements occupied in vivo in HS-3 appear similar to those described previously in globin gene promoters and 3' enhancers. These findings suggest that the distinctive properties of the HS-3 region may be attributable to the organization of these occupied motifs and the consequent protein interactions, rather than to the binding of unique LCR regulatory factors.

In vivo footprinting of the human alpha-globin locus upstream regulatory element by guanine and adenine ligation-mediated polymerase chain reaction

A major regulatory element required for expression of the human alpha-globin genes is located 40 kb upstream of the embryonic zeta-globin gene. To understand how this and other locus control region (LCR) elements contribute to high-level expression in erythroid cells, we have performed high-resolution, in vivo dimethyl sulfate footprinting. In addition, we have modified the dimethyl sulfate-based ligation-mediated polymerase chain reaction in vivo footprinting procedure to permit the assessment of interactions at guanine and adenine residues, rather than guanines alone. In vivo footprinting of the human alpha-LCR element carried on chromosome 16 in a mouse erythroleukemia cell environment revealed protein occupancy at GATA-1, AP-1/NF-E2, and CACC/GGTGG motifs, specific differences compared with in vitro protein binding, and distinct changes in one region upon dimethyl sulfoxide-induced cellular maturation. No protein contacts were detected in nonexpressing hepatoma cells. In addition, we have demonstrated that two AP-1 motifs in the alpha-LCR element which are occupied in vivo bind purified mouse NF-E2 protein in vitro. Our data suggest that three proteins, GATA-1, NF-E2, and unknown CACC/GGTGG factors, are minimally required as DNA-binding proteins for the function of LCR-like elements. The juxtaposition and interaction of these factors with each other, and with accessory proteins not directly in contact with DNA, are likely to account for the relative position independence of the upstream globin regulatory elements.

In vivo footprinting of the human alpha-globin locus upstream regulatory element by guanine and adenine ligation-mediated polymerase chain reaction

A major regulatory element required for expression of the human alpha-globin genes is located 40 kb upstream of the embryonic zeta-globin gene. To understand how this and other locus control region (LCR) elements contribute to high-level expression in erythroid cells, we have performed high-resolution, in vivo dimethyl sulfate footprinting. In addition, we have modified the dimethyl sulfate-based ligation-mediated polymerase chain reaction in vivo footprinting procedure to permit the assessment of interactions at guanine and adenine residues, rather than guanines alone. In vivo footprinting of the human alpha-LCR element carried on chromosome 16 in a mouse erythroleukemia cell environment revealed protein occupancy at GATA-1, AP-1/NF-E2, and CACC/GGTGG motifs, specific differences compared with in vitro protein binding, and distinct changes in one region upon dimethyl sulfoxide-induced cellular maturation. No protein contacts were detected in nonexpressing hepatoma cells. In addition, we have demonstrated that two AP-1 motifs in the alpha-LCR element which are occupied in vivo bind purified mouse NF-E2 protein in vitro. Our data suggest that three proteins, GATA-1, NF-E2, and unknown CACC/GGTGG factors, are minimally required as DNA-binding proteins for the function of LCR-like elements. The juxtaposition and interaction of these factors with each other, and with accessory proteins not directly in contact with DNA, are likely to account for the relative position independence of the upstream globin regulatory elements.

Structure of the human ferrochelatase gene. Exon/intron gene organization and location of the gene to chromosome 18

We have determined the structure of the human ferrochelatase gene after isolation and characterization of lambda phage clones mapping discrete regions of the cDNA. This gene was assigned to human chromosome 18 at region q21.3, by fluorescent in situ hybridization. The gene contains a total of 11 exons and has a minimum size of about 45 kb. The exon/intron boundary sequences conform to consensus acceptor (GTn) and donor (nAG) sequences, and the exons in the gene appear to encode functional protein domains. A major site of the transcription initiation, determined by S1 nuclease mapping, was assigned to an adenine base 89 bases upstream from the adenine base of the translation initiation ATG. The promoter region contains a potential binding site for Sp1, NF-E2 and erythroid-specific transcriptional factor GATA-1, but not a typical TATAA or CCAAT sequence. Analysis of primer extension showed that the transcription starts at the same position between hepatoma HepG2 and erythroleukemia K562 cell mRNA, thereby suggesting that there can be a single transcript in erythroid and non-erythroid cells.

Sequence and expression of GLN3, a positive nitrogen regulatory gene of Saccharomyces cerevisiae encoding a protein with a putative zinc finger DNA-binding domain

The GLN3 gene of Saccharomyces cerevisiae is required for the activation of transcription of a number of genes in response to the replacement of glutamine by glutamate as source of nitrogen. We cloned the GLN3 gene and constructed null alleles by gene disruption. GLN3 is not essential for growth, but increased copies of GLN3 lead to a drastic decrease in growth rate. The complete nucleotide sequence of the GLN3 gene was determined, revealing one open reading frame encoding a polypeptide of 730 amino acids, with a molecular weight of approximately 80,000. The GLN3 protein contains a single putative Cys2/Cys2 zinc finger which has homology to the Neurospora crassa NIT2 protein, the Aspergillus nidulans AREA protein, and the erythroid-specific transcription factor GATA-1. Immunoprecipitation experiments indicated that the GLN3 protein binds the nitrogen upstream activation sequence of GLN1, the gene encoding glutamine synthetase. Neither control of transcription nor control of initiation of translation of GLN3 is important for regulation in response to glutamine availability.

The promoter of the CD11b gene directs myeloid-specific and developmentally regulated expression

Human CD11b/CD18 (complement receptor type 3) is a member of the beta 2 integrin subfamily which also includes the heterodimers CD11a/CD18 and CD11c/CD18. The CD11 molecules and the common CD18 are the products of different genes that exhibit distinct though overlapping patterns of tissue- and developmental-specific expression. Whereas expression of CD11b and CD11c is almost exclusively restricted to cells of the myeloid lineage, that of CD11a and CD18 is panleukocytic. To begin to understand the mechanisms by which expression of these gene products is restricted to leukocytes and leukocyte subpopulations and to elucidate the mechanisms by which their expression is coordinated, we have cloned and characterized the promoter region of the CD11b gene. A single transcription initiation site has been identified and the region extending 242 base pairs upstream and 71 base pairs downstream of this site has been shown to be sufficient to direct tissue-, cell-, and development-specific expression in vitro, which mimics that of the CD11b gene in vivo. Within this region there are potential binding sites for transcription factors known to be involved in hematopoietic-specific and phorbol ester-inducible gene expression. Further analysis of this region of the CD11b gene and comparison with the promoters of the CD11a, CD11c, and CD18 genes should lead to significant insights into the molecular mechanisms by which these genes are regulated during hematopoietic development and upon activation.