Analysis of both the cis-regulatory sequences which control globin gene switching as well as the trans-acting factors which bind to these sequences to elicit a differential, developmentally regulated response has lent insight into the general mechanisms responsible for tissue-specific gene regulation. We show here that the chicken adult β-globin gene promoter sequences are intimately involved in competitive interaction with the β/ε-globin enhancer to regulate differentially ε-versus β-globin gene transcription. Secondly, we show that the family of GATA transcription factors directs gene regulation in a variety of discrete cell types, and describe potential cellular target genes for each member of the GATA factor family, as well as potential mechanisms whereby multiple GATA factors expressed in a single cell might be used to elicit differential transcriptional activities.

During vertebrate ontogeny, erythroid cells undergo a series of morphological and biosynthetic modifications in parallel with changes in lineage and sites of erythropoiesis (Bruns and Ingram, 1973). Transcriptional regulation of this process, particularly with respect to the α- and β-like globin genes, has been extensively studied as a paradigm for cellular commitment, differentiation and embryonic development. In chickens, the cis-linked, ρ-like globin genes are arranged in the order 5′-ρ -PH-P-ε-3′ (Dolan et al., 1981) and their expression is restricted to either the primitive or definitive erythroid compartments. Primitive cells form by 36 hours of development and express embryonic β - and ε-globin (Bruns and Ingram, 1973). Beginning at day 5, erythroid cells of the definitive lineage rapidly replace these primitive cells and express adult βH- and β-globin (Bruns and Ingram, 1973). This changing pattern of globin isotypes during development is referred to as hemoglobin switching.

On the basis of transient transfection experiments, we proposed that the ε-to β-globin switch is mediated at the transcriptional level by mutually exclusive competition between the promoters of each gene for interaction with a single enhancer (Choi and Engel, 1988). Central to this model was the finding that correct developmental regulation of the ε-globin gene is observed only when it is linked in cis to both the shared β/ε enhancer and the adult gene; either duplication of the enhancer or deletion of the adult gene resulted in ε-globin expression in both primitive and definitive erythroid cells (Choi and Engel, 1988). Further analysis demonstrated that definitive suppression of ε-globin transcription requires a stage selector element (SSE) located between -112 and -12 bp of the adult promoter (Choi and Engel, 1988). This is in contrast to β-globin gene transcription, which is regulated autonomously, and is not dependent upon the linked embryonic gene for definitive-stage specificity (Choi and Engel, 1988). Biochemical analysis of proteins interacting with the enhancer and SSE strongly suggested that NF-E4 and pCTF (factors that bind the SSE only in definitive cells; Fig. IB) are the principle determinants controlling the two alternate promoter/enhancer interactions at different developmental stages (Engel et al., 1989; Gallarda et al., 1989a,b).

As a further test of this model, a cotransfection RNA/PCR assay was used to analyze the effects of individual SSE mutations on transcription of linked β - and ε-globin genes (Foley and Engel, 1992). Specific SSE mutations (see below) were introduced into a test construct containing the genomic β /ε-globin locus which was modified by insertion of short oligonucleotides into the third exon of each gene [p β ′ ε ′ (+).pP(-112)] (Fig. 1A); these inserted sequences serve to differentiate transcripts derived from the transfected and endogenous globin genes. The assay also employs (as a control for transfection and RNA/PCR efficiency) a second construct with genes marked by slightly larger inserts [p β ″ ε ″ (++)] (Fig. 1A).

Fig. 1.

RNA/PCR assay and P-globin SSE mutations. (A) Structures of the test [p β ′ ε ′ (+).p β (-112)] and control]p β ″ ε ″ (++)] constructs. Shaded and black regions represent exons and third exon inserts, respectively. The positions of enhancer elements are depicted by open boxes labeled E. Arrowheads separated by dashed lines represent β - and ε-globin specific PCR primers and the expected sizes of amplification products derived from cDNA templates are indicated. (B) Sequences of mutated SSEs used to replace the wild-type sequence [pP(— 112)] of the test construct. Mutations in individual transcription factor-binding sites are indicated in bold lettering. The identities of the transcription factors that are presumed to bind to these sequences are represented above the wild-type SSE (and are used to identify each mutation).

Fig. 1.

RNA/PCR assay and P-globin SSE mutations. (A) Structures of the test [p β ′ ε ′ (+).p β (-112)] and control]p β ″ ε ″ (++)] constructs. Shaded and black regions represent exons and third exon inserts, respectively. The positions of enhancer elements are depicted by open boxes labeled E. Arrowheads separated by dashed lines represent β - and ε-globin specific PCR primers and the expected sizes of amplification products derived from cDNA templates are indicated. (B) Sequences of mutated SSEs used to replace the wild-type sequence [pP(— 112)] of the test construct. Mutations in individual transcription factor-binding sites are indicated in bold lettering. The identities of the transcription factors that are presumed to bind to these sequences are represented above the wild-type SSE (and are used to identify each mutation).

This control construct contains a duplication of the p/ε enhancer to allow expression of both genes in definitive cells (Choi and Engel, 1988). Two PCR primer pairs were designed that specifically amplify both test and control transcripts from either the marked β-or ε-globin genes (Fig. 1A). RNA/PCR (incorporating a[32P]dCTP) was then performed on total RNA isolated from definitive erythroid cells transiently transfected with the test and control constructs. Subsequent analysis by denaturing polyacrylamide gel elec-trophoresis demonstrated that the ratios between β - and ε-globin test and control signals were constant within the exponential phase of amplification and corresponded to the template ratios present in the starting RNA samples (data not shown; Foley and Engel, 1992).

Since the SSE is required for adult β-globin gene expression as well as for suppression of ε-globin transcription in definitive cells (Choi and Engel, 1988), we individually mutated all four characterized protein-binding sites in this region (Fig. 1B). These modified SSEs were introduced into the test construct and adult p-globin expression was analyzed in definitive cells as described above. Mutation of the TATA box was found to decrease β-globin transcription by 10-fold, while mutation of the distal AP-2 site reduced expression by 2-fold (Fig. 2). Disruption of either the NF-E4 or pCTF-binding sites significantly decreased transcription to within approximately 2-fold of the level observed after deletion of the entire SSE (Fig. 2). Given that NF-E4 and pCTF are definitive stage-specific factors (Gallarda et al., 1989a), these results are consistent with their being principally responsible for β-globin gene activation in the definitive lineage. While pCTF is expressed in definitive cells both before and after overt β-globin transcription, NF-E4 is restricted to the mature definitive stage (Gallarda et al., 1989a). It therefore seems likely that NF-E4 may serve as a key control point regulating β-globin induction during erythroid cell maturation.

Fig. 2.

SSE mutations result in reciprocal changes in β-versus ε-globin expression. The test construct and its derivatives were cotransfected with the control construct into HD3 cells (Beug et al., 1982) and analyzed by RNA/PCR as described in the text. SSE mutations are identified as in Fig. 1, while β P(-820) and β P(-15) refer to deletions of the β-globin promoter to-820 and-15 bps, respectively. The β P(NF-E4/ε P) construct is not discussed in the text. The results presented are the fold changes in β ′ β ″ (β) and ε ′ /ε ″ (ε) ratios relative to that of the wild-type SSE and represent the means (±1 s.d.) from three independent experiments. Increased and decreased expression are indicated by solid and hatched bars, respectively.

Fig. 2.

SSE mutations result in reciprocal changes in β-versus ε-globin expression. The test construct and its derivatives were cotransfected with the control construct into HD3 cells (Beug et al., 1982) and analyzed by RNA/PCR as described in the text. SSE mutations are identified as in Fig. 1, while β P(-820) and β P(-15) refer to deletions of the β-globin promoter to-820 and-15 bps, respectively. The β P(NF-E4/ε P) construct is not discussed in the text. The results presented are the fold changes in β ′ β ″ (β) and ε ′ /ε ″ (ε) ratios relative to that of the wild-type SSE and represent the means (±1 s.d.) from three independent experiments. Increased and decreased expression are indicated by solid and hatched bars, respectively.

The SSE mutations were also analyzed for their effects on expression of the linked ε-globin gene in definitive cells. Mutations in the TATA, NF-E4 and PCTF sites, as well as deletion of the entire SSE, resulted in a 10-to 20-fold activation of ε-globin expression, while the AP-2 mutation resulted in a 4-fold increase (Fig. 2). Comparison of the changes in β-versus ε-globin transcription for each individual mutated site clearly indicates that in definitive cells the expression of each gene is reciprocally related to the other and dependent upon the level of SSE activity (Fig. 2).

These experiments demonstrate that the key predictions of the promoter competition model are correct, and we can draw several conclusions from these results: firstly, that definitive stage-specific transcription of the p-globin gene is dependent on expression of the SSE-binding factor NF-E4; secondly, that β-globin expression is absent in primitive cells due to the lack of NF-E4 and pCTF; and thirdly, that the SSE element is required for suppression of E-globin transcription in definitive cells. In the first two instances, NF-E4 and β CTF appear to act as transcriptional activators directly on the β-globin promoter; in the third instance, however, they likely function indirectly by mediating preferential association between the enhancer and the adult promoter, thereby preventing ε-globin gene transcription. These results strongly support the existence of a competitive promoter-enhancer equilibrium.

We originally cloned (from a chicken cDNA library) a family of transcription factors capable of binding to a DNA motif with the core sequence GATA and showed that each member of this family is a potent transcriptional activator in vivo (Yamamoto et al., 1990). The three family members (cGATA-1, cGATA-2 and cGATA-3) exhibit greater than 90% amino acid homology to one another within the DNA-binding domain, but are less highly conserved outside of this region. Indeed, the duplicated zinc finger motif that forms the DNA-binding domain is highly conserved among all members of the GATA family from all species examined so far (Zon et al., 1991).

The cGATA factors exhibit distinct tissue and temporal patterns of expression (Yamamoto et al., 1990; M. W. Leonard, K.-C. Lim and J. D. Engel, unpublished observations). The various GATA family members have overlapping but distinct patterns of expression; GATA-1 expression is restricted to erythroid cells, mast cells and megakaryocytes (Tsai et al., 1989; Martin et al., 1990; Romeo et al., 1990). GATA-2 is expressed in a wide variety of cell types and GATA-3 is abundantly expressed in T-lymphocytes, mature erythrocytes and the developing brain (Yamamoto et al., 1990). In some cases, multiple GATA factors are coexpressed and may potentially bind the same regulatory sites of downstream target genes. For example, GATA-1, -2 and -3 are all expressed in definitive chicken erythroid cells (Yamamoto et al., 1990); similarly, GATA-2 and GATA-3 are expressed in identical neurons within the chicken central nervous system (J. M. Korn-hauser, M. W. Leonard, M. Yamamoto, J. H. LaVail, K. E. Mayo and J. D. Engel, unpublished observations).

(A)Binding site specificity of the cGATA factor family

The GATA family of transcription factors are all related by their very similar zinc finger DNA-binding domains. The consensus recognition sequence for these factors, WGATAR (Evans et al., 1988; Wall et al., 1988; W = A/T, R-A/G), contains inherent ambiguity, and although each family member, cGATA-1, -2 and -3, has been shown to bind to the same GATA site found in the mouse α-globin promoter (TGATAA), we proposed that each factor might have a slightly different, distinguishable binding specificity still encompassed by the GATA core consensus. The determination of the precise recognition sequences for each factor may be critical to understanding the role these factors play, individually as well as relative to each other, in the tissues in which they are coexpressed. Defining which factor is bound to any given promoter or enhancer region will be necessary to discern which of the GATA factors has a functional role at that given site. In addition to overlapping expression patterns, the cell types with highest expression levels differ for the various GATA factors. It is anticipated that the target genes for GATA-3 in T-cells would differ from those of GATA-1 in red blood cells. Because the WGATAR-binding consensus for the GATA factors was compiled on the basis of erythroid-specific genes, the recognition element of the GATA factors mediating transcriptional regulation in non-erythroid cells could well be different, and must, therefore, be directly assessed.

In order to investigate the binding specificities of the GATA factors, an oligonucleotide was synthesized with a central (GAT) core sequence bordered by randomized nucleotides: 3 nucleotides 5 ′ (positions −1, −2 and-3) and 4 nucleotides 3 ′ (positions 1, 2, 3 and 4) from the fixed GAT sequence. The double-stranded oligonucleotide was radiolabeled and then used as a probe in electrophoretic gel mobility shift assays (EGMSA) with purified, bacterially expressed glutathione-S-transferase fusion proteins (Smith and Johnson, 1988) of each cGATA-1,-2 and-3. Selected sites with a high affinity for each GATA protein were then recovered by gel isolating the reduced mobility band and elution from the gel. The selected oligonucleotides were then amplified by PCR. This process was repeated to complete four rounds of selection, and the pool of oligonucleotides was cloned and individual sites were sequenced (Blackwell and Weintraub, 1990).

The frequency of encountering either G, A, T or C was determined for each randomized nucleotide position. Fig. 3 compares, for the three chicken GATA proteins, the most highly favored nucleotide at each position, and the frequencies with which it is recovered in a number of sequenced, selected sites. All the GATA proteins strongly selected for GATAA, with position 1 being more highly selected than position 2. This result was quite unanticipated, since an A or G (at position +2) is part of the traditional WGATAR consensus (Evans et al., 1988; Wall et al., 1988).

Fig. 3.

Optimum binding sites determined for the cGATA-1, -2 and -3 transcription factors. The consensus preferred binding site for each of the cloned chicken GATA factors is shown. Sites were selected in three rounds on the basis of binding by bacterially expressed, purified glutathione S-transferase fusion proteins of cDNA clones encoding cGATA-1, -2 and -3 (Yamamoto et al., 1990). Selected sites were cloned and sequenced, and then analyzed for the frequency of encountering each nucleotide at each randomized position. Nucleotides shown are those most frequently recovered at each position when sequencing the following number of independently recovered binding sites: cGATA-1, 22 clones; cGATA-2, 52 clones, cGATA-3, 79 clones.

Fig. 3.

Optimum binding sites determined for the cGATA-1, -2 and -3 transcription factors. The consensus preferred binding site for each of the cloned chicken GATA factors is shown. Sites were selected in three rounds on the basis of binding by bacterially expressed, purified glutathione S-transferase fusion proteins of cDNA clones encoding cGATA-1, -2 and -3 (Yamamoto et al., 1990). Selected sites were cloned and sequenced, and then analyzed for the frequency of encountering each nucleotide at each randomized position. Nucleotides shown are those most frequently recovered at each position when sequencing the following number of independently recovered binding sites: cGATA-1, 22 clones; cGATA-2, 52 clones, cGATA-3, 79 clones.

Beyond the 5-nucleotide similarity, at the positions further from the core, the individual factors begin to show different optimal sites. At position-1, GATA-2 and GATA-3 selected a T as frequently as an A, consistent with the GATA consensus; however, GATA-1 strongly preferred an A at that position. Not surprisingly, the further away from the central GAT core, the weaker the selection for a particular nucleotide at that position. Each of the factors seems to have unique preferences at positions −2 and +3; these positions might then be those anticipated to be the most important in discriminating amongst the three proteins. The furthest positions analyzed, −3 and +4, showed relatively weak selection, but, once again, there are subtle, consistent differences between the three proteins.

Surprisingly, the canonical GATA site, WGATAR, was not found to be the most favored site for any of the GATA factors. At the +2 (R) position, not only is G not selected, it seems to be selected against; none of the sequenced binding sites (of >150) obtained with any of the three factors had a G at this position, whereas T and C were found, albeit very infrequently. The WGATAR consensus was determined to include A/G at the +2 position because there are GATAG sites found in the chicken p-globin enhancer (Evans et al., 1988), the chicken α-globin enhancer (Knezetic and Felsenfeld, 1989), the chicken βH promoter (Perkins et al., 1989) and the human Aγ globin promoter (Martin et al., 1989); however, the site in the chicken β-globin enhancer is the only one of these which has been shown to be functionally significant by mutational analysis (Reitman and Felsenfeld, 1988).

(B) Cloning and characterization of the mGATA-3 gene

Functionally important consensus GATA-binding sites have been identified in transcriptional regulatory regions of a number of murine and human T-cell specific genes (Ko et al., 1991) and so we sought to determine whether T-lymphocytes from species other than chickens similarly express high levels of a GATA-binding factor. We isolated murine and human homologues of cGATA-3 cDNA (mGATA-3 and hGATA-3 respectively; Ko et al., 1991), and demonstrated that all three factors are highly conserved in their amino acid sequences (greater than 90% overall identity between hGATA-3 and cGATA-3 compared with approximately 40% identity between hGATA-1 and cGATA-1; Trainor et al., 1990), have similar tissue distributions (being abundantly expressed in T-lymphocytes and brain) and are capable of activating transcription in vivo (Ko et al., 1991).

As an initial step towards understanding how this complex pattern of GATA-3 expression is achieved, we have isolated and characterized the mGATA-3 locus. Four overlapping clones isolated from a genomic DNA library encompass the entire coding region of mGATA-3 with an additional 18 kb of 5′ and 15 kb of 3′ flanking sequences. The gross organization of the mGATA-3 gene (Fig. 4) is similar to that of mGATA-1 (Tsai et al., 1991) and cGATA-1 (Hannon et al., 1991). Thus, the gene comprises six exons, the first being entirely untranslated and the second containing the initiation methionine codon. Each of the duplicated C-X2-C-X17-C-X2-C zinc fingers that define the GATA family is located in a separate exon (exons 4 and 5) and the carboxy terminus of the protein, 3′ untranslated region and poly adenylation site are located in exon 6.

Fig. 4.

Organization of the mGATA-3 gene. Recombinant λbacteriophage DNAs describing the mGATA-3 locus (top line, markers every 5 kbp) are shown. The vertical bars within the clones describe the position of restriction sites for EcoRI or the single Not\ site (N). The sequence organization of the gene, the splice boundaries and the chromosomal location of mGATA-3, are shown in abbreviated form below the genomic clones.

Fig. 4.

Organization of the mGATA-3 gene. Recombinant λbacteriophage DNAs describing the mGATA-3 locus (top line, markers every 5 kbp) are shown. The vertical bars within the clones describe the position of restriction sites for EcoRI or the single Not\ site (N). The sequence organization of the gene, the splice boundaries and the chromosomal location of mGATA-3, are shown in abbreviated form below the genomic clones.

Primer extension and RNase protection assays were performed to determine the transcription initiation site (data not shown). Transcription of the mGATA-3 gene initiates at a unique site 188 nucleotides upstream of the first intron, in contrast to the situation for mGATA-1 and cGATA-1, where extensive mRNA 5′-end heterogeneity was found with multiple transcription initiation sites mapping over approximately 75 nucleotides. As is the case for mGATA-1, the mGATA-3 promoter lacks an identifiable TATA box (although motifs with the sequence TTAA and TAGAA are located at −27 and −37, respectively) and no consensus for the transcription initiator element (Smale and Baltimore, 1989) is present. A number of consensus binding sites for a variety of transcription factors (particularly SP1 and AP2) are present within the proximal 310 bp 5’ to the cap site, but the functional importance of individual elements for mGATA-3 promoter activity has yet to be ascertained. No consensus GATA factor-binding sites exist within this region.

Quantitative RT/PCR was used to assess the level of GATA-3 expression in transformed and primary T-lymphocytes from a number of sources. cGATA-3 is expressed in v-rel transformed pre-B/T cells, is approximately 5-fold less abundant in RP9 mature B-cells and greater than 10fold more abundant in MSB mature T-cells (M. W. Leonard, K.-C. Lim and J. D. Engel, unpublished observations). hGATA-3 and mGATA-3 are similarly expressed at some 50-to 100-fold greater abundance in normal mature T-lymphocytes than in B-cells. Furthermore, FACS sorted CD4-CD8-, CD4+CD8+, CD4+ and CD8+ murine thymocyte fractions all express high levels of mGATA-3 (the lowest expression being in the CD4+CD8+ cells and the highest (8-fold higher) in the CD4+ single positive population; M. W. Leonard, D. Kouissis, F. Grosveld and J. D. Engel, unpublished observations). Taken together these data strongly suggest that the GATA-3 transcription factor is likely to play an important role in the regulation of T-cell specific transcription from the earliest stages of T-lymphocyte development.

(C) GATA factors that conditionally regulate transcription

As mentioned above, one enigmatic aspect of GATA factor expression is that multiple members of the family, which ostensibly share very similar DNA-binding site specificity, are expressed in the same cells during erythroid and neuronal development (Yamamoto et al., 1990). The observed changing ratios of cGATA-1,-2 and -3 factors with respect to one another during erythroid differentiation (Yamamoto et al., 1990; M. W. Leonard, K.-C. Lim and J. D. Engel, unpublished observations) suggested that the ratio of these factors might play a role in determining the decision pathway in erythroid progenitor cells: i.e., that expression of one GATA factor at greater or lesser abundance might lead to an altered propensity for regeneration or differentiation in the erythroid developmental program. In order to address this question empirically, conditional mutants of the cGATA factors were prepared. By fusing the hormone-binding domain of the human estrogen receptor (hER; Kumar et al., 1986) to the carboxy termini of the chicken GATA factors (Yamamoto et al., 1990; Fig. 5A), we hoped to generate cGATA regulatory proteins whose activities could be conditionally regulated by the addition or removal of the ligand β-estradiol (E2).

Fig. 5.

Conditional GATA factors. (A) Constructions. Full-length cGATA factor cDNAs, lacking termination codons, were generated using PCR. These were then joined in frame to a 0.95 kbp BamHl/EcoRI cDNA fragment encoding the hormone-binding domain (amino acids 282 to 595) of the human estrogen receptor (hER) using a PCR-introduced BamHI restriction site at the 3′ end of each of the cGATA cDNA clones. The introduction of the BamHI linker resulted in the insertion of 3 amino acids (Pro-Asp-Pro) in the joining region of the chimeric proteins between the last encoded amino acid of the cGATA factors and Ser282 of hER. The resulting chimeric cDNAs (1.9 kb cGATA-1/ER, 2.35 kb cGATA-2/ER and 2.45 kb cGATA-3/ER) were then subcloned into the eukaryotic expression vector TFAneo, in which synthesis of the hybrid cGATA/ER mRNAs is directed by the RSV LTR. (B) cGATA/ER chimeric proteins trans-activate GATA-directed reporter plasmids in hormone dependent manner. The optimal ratio of activator (wild-type cGATA or chimeric cGATA/ER) and reporter (C3β3GH) plasmids was cotransfected into QT6 quail fibroblast cells, as described (Yamamoto et al., 1990). Sixteen hours post-transfection, β-estradiol (10−6 M) or an estrogen antagonist (ICI 164,384 at IO−6 M or tamoxifen at 10’7 M) in ethanol was added to the cells. Thirty-six hours later, the media was assayed for secreted human growth hormone using the Allegro hGH kit (Nichols Institute Diagnostics). Trans-activation activity was calculated from at least two independent experiments as described (Yamamoto et al., 1990).

Fig. 5.

Conditional GATA factors. (A) Constructions. Full-length cGATA factor cDNAs, lacking termination codons, were generated using PCR. These were then joined in frame to a 0.95 kbp BamHl/EcoRI cDNA fragment encoding the hormone-binding domain (amino acids 282 to 595) of the human estrogen receptor (hER) using a PCR-introduced BamHI restriction site at the 3′ end of each of the cGATA cDNA clones. The introduction of the BamHI linker resulted in the insertion of 3 amino acids (Pro-Asp-Pro) in the joining region of the chimeric proteins between the last encoded amino acid of the cGATA factors and Ser282 of hER. The resulting chimeric cDNAs (1.9 kb cGATA-1/ER, 2.35 kb cGATA-2/ER and 2.45 kb cGATA-3/ER) were then subcloned into the eukaryotic expression vector TFAneo, in which synthesis of the hybrid cGATA/ER mRNAs is directed by the RSV LTR. (B) cGATA/ER chimeric proteins trans-activate GATA-directed reporter plasmids in hormone dependent manner. The optimal ratio of activator (wild-type cGATA or chimeric cGATA/ER) and reporter (C3β3GH) plasmids was cotransfected into QT6 quail fibroblast cells, as described (Yamamoto et al., 1990). Sixteen hours post-transfection, β-estradiol (10−6 M) or an estrogen antagonist (ICI 164,384 at IO−6 M or tamoxifen at 10’7 M) in ethanol was added to the cells. Thirty-six hours later, the media was assayed for secreted human growth hormone using the Allegro hGH kit (Nichols Institute Diagnostics). Trans-activation activity was calculated from at least two independent experiments as described (Yamamoto et al., 1990).

As anticipated, the presence or absence of estradiol did not influence the ability of the three wild-type cGATA factors to rranx-activate the C3βGH reporter plasmid (Fig. 5B; Yamamoto et al., 1990). The cGATA/ER proteins, on the other hand, irans-activate the reporter plasmid only in the presence of E2 (Fig. 5B). cGATA-1/ER was the best transcriptional activator in this assay and, perhaps surprisingly, activates the reporter plasmid to a greater extent than the parent factor, cGATA-1. The level of trans-activation by cGATA-2/ER is approximately half that of cGATA-2. cGATA-3/ER appears to be the poorest activator of the chimeric proteins, although its ability to stimulate transcription is quite comparable to that of the parental wildtype cGATA-3 factor. In the presence of the estrogen antagonist tamoxifen, the degree of tran.s-activation of cGATA-1/ER is reduced to a level comparable to that of native cGATA-1, whereas treatment of cells with antagonist ICI results in rather poor trans-activation by cGATA-1/ER. cGATA-2/ER and cGATA-3/ER are both completely inactive in the presence of either antagonist.

While the irans-activating activities of cGATA-2/ER and cGATA-3/ER do not differ markedly from their wild-type counterparts, cGATA-1/ER trans-activates a reporter construct significantly better than the native cGATA-1 factor itself. As there is a hormone-responsive trans-activation domain within the hormone-binding domain of the estrogen receptor (Webster et al., 1988; Lees et al., 1989), it is possible that this hER rrans-activation domain contributes to the observed increase in /ra/r.v-activation potential of this chimera.

By fusing individual cGATA transcription factor cDNAs to a segment of cDNA encoding the hormone-binding domain of the human estrogen receptor, we generated cGATA/ER chimeric proteins. Using an artificial promoter containing GATA factor-binding sites, we showed that cGATA/ER chimeric proteins are indeed hormone responsive in cotransfection, trans-activation assays. In the absence of E2, the chimeric GATA/ER proteins fail to activate transcription; only the ligand-bound species are able to modulate gene expression.

(D) hGATA-3 regulates HIV-1 transcription

Human immunodeficiency virus type 1 (HIV-1) is the etiologic agent of acquired immune deficiency syndrome (AIDS), and infects human CD4+ T cells and myeloid cells of the monocytic lineage using the CD4 receptor for viral entry (see Cullen and Green, 1989; McCune, 1991; Vaish-nav and Wong-Staal, 1991). The retroviral long terminal repeat (LTR) of HIV-1 has been suggested to play a number of roles in the regulation of the life cycle of the virus after it infects cells (Rosen et al., 1985). The viral LTR U3 region contains a variety of czs-acting regulatory sequences responsible for modulating viral gene expression. Little is known, at present, regarding the identities of factors that may govern the cell type-restricted expression pattern of HIV-1.

Since the expression profile of hGATA-3 is consistent with the documented tropism of HIV-1 replication, one might anticipate that hGATA-3 could play a role in the T cell-specific regulation of HIV-1 transcription. Indeed, initial inspection of the sequences in the viral LTR indicated the presence of four potential GATA factor-binding sites (Fig. 6A). Both strands of HIV-1 U3 region DNA were examined for their ability to bind to hGATA-3 by DNase I footprinting analysis (data not shown). The foot-printing results actually revealed six specific hGATA-3-binding sites, all localized within the 5’ domain of the LTR. The four consensus GATA sites were protected by hGATA-3 factor, as well as two additional non-canonical GATA-binding sites: GATTA (site 5) and GATGA (site 2; Fig. 6A).

Fig. 6.

(A) Nucleotide sequence of, and transcription factor-binding sites in, the HIV-1 LTR. Nucleotide sequence of the HIV-1 LTR from the CAP site (nucleotide +1) extending 5’ to the boundary of the U3 region is shown. The demonstrated and/or putative (consensus sequence) transcription factor-binding sites are underlined and the identity of the corresponding proteins are shown either below or above the sequence (Jones et al., 1986; Nabel and Baltimore, 1987; Hauber and Cullen, 1988; Jakobovits et al., 1988; Shaw et al., 1988; Harrich et al., 1989; Smith and Greene, 1989; Dasgupta et al., 1990; Kato et al., 1991; Berkhout and Jeang, 1992). The binding sites for hGATA-3 identified by in vitro footprinting are shown in outlined letters and are numbered 1 to 6 consecutively, beginning nearest to the HIV-1 CAP site. The numbers at left indicate the nucleotide number relative to the transcriptional start site. (B) GATA-binding sites and mutations. All six hGATA-3-binding sites and surrounding sequences are listed on the top line with the GATA sequence shown in bold. Each GATA site was mutated to a specific restriction enzyme site by PCR mutagenesis (underlined) as shown on the bottom line. (C) hGATA-3 /ran.s’-activates HIV-1 LTR-LTR U3, distance from HIV CAP site (+1) HIV-1 LTR constructs plus 5 μg of RSV/hGATA-3 (Ko et al., 1991). 40% of the cells from a confluent plate were used for each transfection. Cell lysates were then prepared by repeated freeze/thaw. Protein concentrations were determined and CAT assays were performed using aliquots of extract containing the equal quantities of recovered protein (Gorman et al., 1982). The result was quantified by determining the amount of 14C-chloramphenicoI produced in the enzymatic assay and the conversion was quantitated on a Molecular Dynamics Phosphorlmager. The results shown are calculated relative to the conversion by cotransfection with antisense hGATA-3, and are the average of four independent experiments. The transfections were done with the HIV/CAT constructs indicated at the left and with sense strand hGATA-3 (open bar) or antisense strand hGATA-3 (shaded bar). (D) HIV-1 LTR wild type and mutations. Line 1: the wild-type sequence of the HIV-1 LTR-453 to +80. The ‘noGATA’ construct (line 2) represents a plasmid in which each of the GATA-binding sites was mutated by PCR into unique restriction enzyme sites. All of the mutated LTRs were then subcloned 5 ′ to a promoterless chloramphenicol acetyltransferase (CAT) gene (pCATbasic; Promega) and were confirmed by DNA sequencing.

Fig. 6.

(A) Nucleotide sequence of, and transcription factor-binding sites in, the HIV-1 LTR. Nucleotide sequence of the HIV-1 LTR from the CAP site (nucleotide +1) extending 5’ to the boundary of the U3 region is shown. The demonstrated and/or putative (consensus sequence) transcription factor-binding sites are underlined and the identity of the corresponding proteins are shown either below or above the sequence (Jones et al., 1986; Nabel and Baltimore, 1987; Hauber and Cullen, 1988; Jakobovits et al., 1988; Shaw et al., 1988; Harrich et al., 1989; Smith and Greene, 1989; Dasgupta et al., 1990; Kato et al., 1991; Berkhout and Jeang, 1992). The binding sites for hGATA-3 identified by in vitro footprinting are shown in outlined letters and are numbered 1 to 6 consecutively, beginning nearest to the HIV-1 CAP site. The numbers at left indicate the nucleotide number relative to the transcriptional start site. (B) GATA-binding sites and mutations. All six hGATA-3-binding sites and surrounding sequences are listed on the top line with the GATA sequence shown in bold. Each GATA site was mutated to a specific restriction enzyme site by PCR mutagenesis (underlined) as shown on the bottom line. (C) hGATA-3 /ran.s’-activates HIV-1 LTR-LTR U3, distance from HIV CAP site (+1) HIV-1 LTR constructs plus 5 μg of RSV/hGATA-3 (Ko et al., 1991). 40% of the cells from a confluent plate were used for each transfection. Cell lysates were then prepared by repeated freeze/thaw. Protein concentrations were determined and CAT assays were performed using aliquots of extract containing the equal quantities of recovered protein (Gorman et al., 1982). The result was quantified by determining the amount of 14C-chloramphenicoI produced in the enzymatic assay and the conversion was quantitated on a Molecular Dynamics Phosphorlmager. The results shown are calculated relative to the conversion by cotransfection with antisense hGATA-3, and are the average of four independent experiments. The transfections were done with the HIV/CAT constructs indicated at the left and with sense strand hGATA-3 (open bar) or antisense strand hGATA-3 (shaded bar). (D) HIV-1 LTR wild type and mutations. Line 1: the wild-type sequence of the HIV-1 LTR-453 to +80. The ‘noGATA’ construct (line 2) represents a plasmid in which each of the GATA-binding sites was mutated by PCR into unique restriction enzyme sites. All of the mutated LTRs were then subcloned 5 ′ to a promoterless chloramphenicol acetyltransferase (CAT) gene (pCATbasic; Promega) and were confirmed by DNA sequencing.

We next investigated the possibility that hGATA-3 might irons-activate HIV-1 LTR-mediated transcription by transfection into HeLa cells, where only low levels of hGATA-2, but no hGATA-3, are expressed. The HIV-1 LTR (nucleotides-453 to +80, Fig. 6A) was subcloned into pCAT basic (Promega) to generate HIVwt/CAT; this reporter construct was cotransfected into HeLa cells with an activator plasmid constitutively expressing hGATA-3 protein (RSV/hGATA-3; Ko et al., 1991). As shown in Fig. 6C, hGATA-3 indeed stimulates HIVwt/CAT expression approximately six-fold. These data show that hGATA-3 activates HIV-1 transcription in vivo.

To examine whether the increase in expression was a direct effect of hGATA-3 binding to the identified GATA sites within the LTR of HIV-1, a series of mutations were generated by PCR in which the GATA sites (Fig. 6B) were individually changed into unique restriction enzyme recognition sequences and then ligated into a single HIV-1 LTR lacking all of the GATA-binding sites (noGATA; Fig. 6D). DNase I footprint analysis confirmed that the mutations indeed eliminated hGATA-3 binding to the GATA sites (data not shown). A 5 • deletion to-120 was also constructed in which only the enhancer region lying 3 • to all the GATA-binding sites remained (Enh; Fig. 6D); this construct still includes the NF-KB, Spl, TF-IID and LBP sites. The individual and multiple GATA-binding site mutations were then functionally analyzed for their effect upon the ability of hGATA-3 to direct transcription from the HIV-1 LTR. Mutagenesis of the individual GATA-binding sites resulted in only a slight decrease in the transcriptional activation by hGATA-3 (data not shown). However, when all of the mutations were combined into a single plasmid (noGATA/CAT; Fig. 6C), its ability to be irau.s-activated by cotransfected hGATA-3 was significantly reduced. These data taken together demonstrate that hGATA-3 stimulates HIV-1 transcription by binding to the GATA sites within the HIV-1 LTR U3 region.

In summary, the binding of hGATA-3 is required for a 6-fold increase in HIV-1 LTR-mediated transcriptional activation in non-lymphoid cells, and mutations which abolish all of the hGATA-3-binding sites within the LTR result in a quantitatively similar decrease in HIV-1 expression upon cotransfection into non-lymphoid cells (Fig. 6). The reduction in activity seen upon transfection of a mutant HIV-1 LTR bearing mutations in all of the GATA-binding sites into Jurkat cells (data not shown) strongly supports the hypothesis that hGATA-3 may be one of the factors mediating basal expression of HIV-1, and is indeed required for optimal expression of the virus.

(E) Expression of GATA-2 and GATA-3 in the brain

To determine if the expression of the GATA-2 and GATA-3 transcription factor genes is restricted to a subset of cells within the chicken brain, we performed in situ hybridization studies using 35S-labeled cGATA complementary RNA probes. At E3.5 and E4, the earliest embryonic stages examined, strong specific hybridization to GATA-2 and-3 mRNAs is detected at the rostro-ventral boundary of the mesencephalon in the region of the constriction between the mesencephalon (the developing optic lobe) and the diencephalon (Fig. 7). Specific expression of GATA-2 and-3 is also observed within the rostral optic tectum, and most of this hybridization is localized to the outermost portion of the neural epithelium. During development, formation of the tectal layers proceeds in a rostro-ventro-lateral to caudo-dorso-medial direction (LaVail and Cowan, 1971a), and so this rostral portion of the tectum which shows more intense labeling is more mature. GATA mRNAs are also detected along the ventral surface of the metencephalon (data not shown).

Fig. 7.

cGATA-2 and cGATA-3 expression in the chicken optic tectum. (A,B) Sagittal section of E4 embryo hybridized to a cGATA-2 probe; (C,D) sagittal section of E6 embryo hybridized with cGATA-3 probe; (E,F) coronal section of E6 embryo hybridized with cGATA-2 probe; (G,H) higher magnification view of a coronal section of E6 brain hybridized with cGATA-3 probe. Bright-fleld photomicrographs in A, C, E and G show the morphology of the tissue; the same fields are visualized by darkfield microscopy in B, D, F and H to show the autoradiographic silver grains indicating hybridization, di, diencephalon; mes, mesencephalon; met, metencephalon; tec, optic tectum; III, third ventricle; ret, retina; NE, neural epithelium; P, pial layer. Chicken embryos were frozen and stored at-80°C until sectioning. 20 μ m-thick coronal or sagittal sections were cut and mounted on gelatin/poly-L-lysine-coated slides. Plasmid cDNA clones encoding cGATA-2 or cGATA-3 (Yamamoto et al., 1990) were digested with restriction enzymes prior to in vitro transcription using either SP6 or T7 RNA polymerase, to generate mRNA-sense or-antisense 35S-labeled RNA probes, respectively. These probes did not contain the sequences encoding the DNA-binding domain to ensure that the hybridization was factor-specific.

Fig. 7.

cGATA-2 and cGATA-3 expression in the chicken optic tectum. (A,B) Sagittal section of E4 embryo hybridized to a cGATA-2 probe; (C,D) sagittal section of E6 embryo hybridized with cGATA-3 probe; (E,F) coronal section of E6 embryo hybridized with cGATA-2 probe; (G,H) higher magnification view of a coronal section of E6 brain hybridized with cGATA-3 probe. Bright-fleld photomicrographs in A, C, E and G show the morphology of the tissue; the same fields are visualized by darkfield microscopy in B, D, F and H to show the autoradiographic silver grains indicating hybridization, di, diencephalon; mes, mesencephalon; met, metencephalon; tec, optic tectum; III, third ventricle; ret, retina; NE, neural epithelium; P, pial layer. Chicken embryos were frozen and stored at-80°C until sectioning. 20 μ m-thick coronal or sagittal sections were cut and mounted on gelatin/poly-L-lysine-coated slides. Plasmid cDNA clones encoding cGATA-2 or cGATA-3 (Yamamoto et al., 1990) were digested with restriction enzymes prior to in vitro transcription using either SP6 or T7 RNA polymerase, to generate mRNA-sense or-antisense 35S-labeled RNA probes, respectively. These probes did not contain the sequences encoding the DNA-binding domain to ensure that the hybridization was factor-specific.

By day E6, GATA mRNA expression is more prominent within the optic tectum, although labeling is still visible in the developing diencephalon (Fig. 7). In the day E12 optic tectum, twelve distinct cell layers have been described (LaVail and Cowan, 1971a). GATA-2 and-3 mRNAs are found mainly in layers vi through ix, with the highest expression in layers vi and viii (data not shown). At this stage, the neuroepithelium (NE) has shrunk, and many neuronal progenitor cells have migrated peripherally. No GATA mRNA is detected in the NE at E12, suggesting that the cells expressing GATA-2 and-3 have indeed migrated out of this layer by this time. cGATA-2 and cGATA-3 are expressed in the same tectal cell layers. GATA-2 and-3 mRNAs continue to be expressed in the adult optic tectum. Throughout development the highest levels of cGATA mRNAs are in the mesencephalic region and the developing optic tectum. These data suggest that GATA mRNA expression is restricted to a limited number of cell lineages within the brain, specifically those of the mesencephalon, the mesencephalon-diencephalon junction region, and the ventral metencephalon.

These results suggest a role for cGATA-2 and-3 transcription factors in the regulated expression of specific genes in the developing chicken visual system. Qualitatively, we find that GATA mRNA expression is most prominent within, and is precisely localized to, discrete groups of cells in the developing brain which are physiologically associated with the visual system. These cells are generated during days 6 to 9 of embryogenesis (LaVail and Cowan, 1971b). Thus, the cGATA-2 and-3 genes become activated in neurons generated during this defined period of major neural and morphological organization, and continue to be expressed thereafter. The cells of the optic tectum expressing GATA mRNAs appear to be neuronal, although it is also possible that these factors may be expressed in non-neuronal cells. At all developmental stages examined, the spatial patterns of cGATA-2 and cGATA-3 expression in the brain appear to be identical.

The conservation between chicken and mouse GATA-3 amino acid sequence (Ko et al., 1991) prompted us to perform mRNA in situ hybridization to ask whether the analogous neural structure (the superior colliculus) in mice expresses GATA-2 and GATA-3. At day 14 of gestation, both mGATA-2 and-3 are expressed in the mesencephalic roof (this structure further differentiates to form the superior colliculus, the primary receptive center for the optic tracts). Thus it appears that the factors may play an important role in the developing vision systems of both species. However, in the mouse, mGATA-2 and-3 are expressed in different cell layers of the developing superior colliculus. mGATA-2 is expressed in less mature cells directly bordering the ventricular zone, while mGATA-3 is expressed in the outer (more-mature) cell layers where pre-neurons are migrating (Fig. 8). This is quite clearly different from the localized expression pattern of these factors seen in the chicken optic tectum, and the significance of this disparity is not yet clear.

Fig. 8.

mGATA-2 and mGATA-3 expression in the superior colliculus at E14.5. Adjacent sagittal sections of an E14.5 embryo were probed with riboprobes specific for each factor. (A) mGATA-3 is expressed in the outer cell layers of the mesecephalic roof. (B) mGATA-2 is expressed in the cell layer directly bordering the ventricular zone. Both factors also show hybridization to distinct cell layers in the tuberculum posterius.

Fig. 8.

mGATA-2 and mGATA-3 expression in the superior colliculus at E14.5. Adjacent sagittal sections of an E14.5 embryo were probed with riboprobes specific for each factor. (A) mGATA-3 is expressed in the outer cell layers of the mesecephalic roof. (B) mGATA-2 is expressed in the cell layer directly bordering the ventricular zone. Both factors also show hybridization to distinct cell layers in the tuberculum posterius.

Other areas of the murine central nervous system were also analyzed for expression of the two factors. At day 11 of gestation, both mGATA-2 and-3 are expressed in the motor neuron pool and sympathetic ganglia. The pattern is similar at day 14, with the exception that mGATA-2 expression in the motor neuron pool is greatly diminished (data not shown). Taken together these results suggest specific and distinct functions for the two transcription factors during murine neural development.

This work was supported by research grants from the NIH (HL 24415, HL 45168 and GM 28896) and a postdoctoral fellowship from the Leukemia Society of America (M. W. L.).

Berkhout
,
B.
and
Jeang
,
K.-T.
(
1992
).
Functional roles for the TATA promoter and enhancers in basal and tat-induced expression of the human immunodeficiency virus type 1 long terminal repeat
.
J. Virology
66
,
139149
.
Beug
,
H.. Palmieri
, S.,
Freudenstein
,
C.
,
Zentgraf
,
H.
and
Graf
,
T.
(
1982
).
Hormone-dependent terminal differentiation in vitro of chicken erythroleukemia cells transformed by ts mutants of avian erythroblastosis virus
.
Cell
28
,
907
919
.
Blackwell
,
T. K.
and
Weintraub
,
H.
(
1990
).
Differences and similarities in DNA-binding preferences of myoD and E2a protein complexes revealed by binding site selection
.
Science
250
,
1104
1110
.
Bruns
,
G. A. P.
and
Ingram
,
V. M.
(
1973
).
The erythroid cells and haemoglobins of the chick embryo
.
Phil. Trans. R. Soc. London Ser. B
.
266
,
225
305
.
Choi
,
O.-R. B.
and
Engel
,
J. D.
(
1988
).
Developmental regulation of p-globin gene switching
.
Cell
55
,
17
26
.
Cullen
,
B. R.
and
Greene
,
W. C.
(
1989
).
Regulatory pathways governing HIV-1 replication
.
Cell
58
,
423
426
.
Dasgupta
,
P.
,
Saikumar
,
P.
,
Reddy
,
C. D.
and
Reddy
,
E. P.
(
1990
).
Myb protein binds to human immunodeficiency virus 1 long terminal repeat sequences and trans-activates LTR-mediated transcription
.
Proc. Nat. Acad. Sci. USA
87
.
8090
8094
.
Dolan
,
M.
,
Sugarman
,
B. J.
,
Dodgson
,
J. B.
and
Engel
,
J. D.
(
1981
).
Chromosomal arrangement of the chicken p-type globin genes
.
Cell
24
,
669
677
.
Engel
,
J. D.
,
Choi
,
O.-R.
,
Endean
,
D. J.
,
Foley
,
K. P.
,
Gallarda
,
J. L.
and
Yang
,
Z.
(
1989
).
Genetics and biochemistry of the embryonic to adult switch in the chicken e- and β-globin genes
.
In Hemoglobin Switching part A: Transcriptional Regulation
(ed.
G.
Stamatoyannopoulos
and
A.
Neinhuis
), pp
89
103
.
New York
:
Alan R. Liss
.
Evans
,
T.
,
Reitman
,
M.
and
Felsenfeld
,
G.
(
1988
).
An erythrocytespecific DNA-binding factor recognizes a regulatory sequence common to all chicken globin genes
.
Proc. Nat. Acad. Sci. USA
85
,
5976
5980
.
Foley
,
K. P.
and
Engel
,
J. D.
(
1992
).
Individual stage selector element mutations lead to reciprocal changes in P-vs. ε-globin gene transcription: genetic confirmation of promoter competition during globin gene switching
.
Genes Dev
.
6
,
730
744
.
Gallarda
,
J. L.
,
Foley
,
K. P.
,
Yang
,
Z.
and
Engel
,
J. D.
(
1989a
).
The P-globin stage selector element factor is erythroid-specific promoter/enhancer binding protein NF-E4
.
Genes Dev
.
3
,
1845
1859
.
Gallarda
,
J. L.
,
Yang
,
Z.
,
Endean
,
D. J.
,
Foley
,
K. P.
and
Engel
,
J. D.
(
1989b
).
cis and trans determinants of chicken p-globin transcription
.
In Tissue Specific Gene Expression
(ed.
R.
Renkawitz
), pp.
103
121
.
Weinheim
:
VCH Publishers
.
Gorman
,
C. M.
,
Moffat
,
L. F.
and
Howard
,
B. H.
(
1982
).
Recombinant genomes which express chloramphenicol acetyltransferase in mammalian cells
.
Mol. Cell. Biol
.
2
,
1044
1051
.
Graham
,
F. L.
and
van der Eb
,
A. J.
(
1973
).
A new technique for the assay of infectivity of human adenovirus 5 DNA
.
Virol
.
52
,
456
467
.
Hannon
,
R.
,
Evans
,
T.
,
Felsenfeld
,
G.
and
Gould
,
H.
(
1991
).
Structure and promoter activity of the gene for the erythroid transcription factor GATA-1
.
Proc. Nat. Acad. Sci. USA
88
,
3004
3008
.
Harrich
,
D.
,
Garcia
,
J.
,
Wu
,
F.
,
Mitsuyasu
,
R.
,
Gonzalez
,
J.
and
Gaynor
,
R.
(
1989
).
Role of SPl-binding domains in in vivo transcriptional regulation of the human immunodeficiency virus type 1 long terminal repeat
.
J. Virology
63
,
2585
2591
.
Hauber
,
J.
and
Cullen
,
B. R.
(
1988
).
Mutational analysis of the transactivation-responsive region of the human immunodeficiency virus type 1 long terminal repeat
.
J. Virology
62
.
673
679
.
Jakobovits
,
A.
,
Smith
,
D. H.
,
Jakobovits
,
E. B.
and
Capon
,
D. J.
(
1988
).
A discrete element 3’ of human immunodeficiency virus 1 (HIV-1) and HIV-2 mRNA initiation sites mediates transcriptional activation by an HIV trans activator
.
Mol. Cell. Biol
.
8
,
2555
2561
.
Jones
,
K. A.
,
Kadonaga
,
J. T.
,
Luciw
,
P. A.
and
Tjian
,
R.
(
1986
).
Activation of the AIDS retrovirus promoter by the cellular transcription factor, Spl
.
Science
232
,
755
759
.
Kato
,
H.
,
Horikoshi
,
M.
and
Roeder
,
R. G.
(
1991
).
Repression of HIV-1 transcription by a cellular protein
.
Science
251
,
1476
1479
.
Knezetic
,
J. A.
and
Felsenfeld
,
G.
(
1989
).
Identification and characterization of a chicken a-globin enhancer
.
Mol. Cell. Biol
.
9
,
893901
.
Ko
,
L. J.
,
Yamamoto
,
M.
,
Leonard
,
M. W.
,
George
,
K. M.
,
Ting
,
P.
and
Engel
,
J. D.
(
1991
).
Murine and human T lymphocyte GATA-3 factors mediate transcription through a c/s-regulatory element within the human T cell receptor 8 gene enhancer
.
Mol. Cell. Biol
.
11
,
2778
2784
.
Kumar
,
V.
,
Green
,
S.
,
Staub
,
A.
and
Chambón
,
P.
(
1986
).
Localisation of the oestradiol-binding and putative DNA-binding domains of the human oestrogen receptor
.
EMBO J
.
5
,
2231
2236
.
LaVail
,
J. H.
and
Cowan
,
W. M.
(
1971a
).
The development of the chick optic tectum. I. Normal morphology and cytoarchitectonie development
.
Brain Res
.
28
,
391
419
.
LaVail
,
J. H.
and
Cowan
,
W. M.
(
1971b
).
The development of the chick optic tectum. II. Autoradiographic studies
.
Brain Res
.
28
,
421
441
.
Lees
,
J. A.
,
Fawell
,
S. E.
and
Parker
,
M. G.
(
1989
).
Identification of two transactivation domains in the mouse oestrogen receptor
.
Nucl. Acids Res
.
17
,
5477
5488
.
Martin
,
D. I. K.
,
Tsai
,
S.-F.
and
Orkin
,
S. H.
(
1989
).
Increased y-globin expression in a nondeletion HPFH mediated by an erythroid-specific DNA-binding factor
.
Nature
338
,
435
438
.
Martin
,
D. I. K.
,
Zon
,
L. I.
,
Mutter
,
G.
and
Orkin
,
S. H.
(
1990
).
Expression of an erythroid transcription factor in megakaryocytic and mast cell lineages
.
Nature
344
,
444
447
.
McCune
,
J. M.
(
1991
).
HIV-1: The infective process in vivo
.
Cell
64
,
351363
.
Nabel
,
G.
and
Baltimore
,
D.
(
1987
).
An inducible transcription factor activates expression of human immunodeficiency virus in T cells
.
Nature
326
,
711
713
.
Perkins
,
N. D.
,
Nicolas
,
R. H.
,
Plumb
,
M. A.
and
Goodwin
,
G. H.
(
1989
).
The purification of an erythroid protein which binds to enhancer and promoter elements of haemoglobin genes
.
Nucl. Acids Res
.
17
,
12991314
.
Reitman
,
M.
and
Felsenfeld
,
G.
(
1988
).
Mutational analysis of the chicken β-globin enhancer reveals two positive-acting domains
.
Proc. Nat. Acad. Sci. USA
85
,
6267
6271
.
Romeo
,
P.-H.
,
Prandini
,
M.-H.
,
Joulin
,
V.
,
Mignotte
,
V.
,
Prenant
,
M.
,
Vainchenker
,
W.
,
Marguerie
,
G.
and
Uzan
,
G.
(
1990
).
Megakaryocytic and erythrocytic lineages share specific transcription factors
.
Nature
344
,
447
449
.
Rosen
,
C. A.
,
Sodroski
,
J. G.
and
Haseltine
,
W. A.
(
1985
).
The location of cL-acting regulatory sequences in the human T cell lymphotropic virus type III (HTLV-III/LAV) long terminal repeat
.
Cell
41
,
247
253
.
Shaw
,
J.-P.
,
Utz
,
P. J.
,
Durand
,
D. B.
,
Toole
,
J. J.
,
Emmel
,
E. A.
and
Crabtree
,
G. R.
(
1988
).
Identification of a putative regulator of early T cell activation genes
.
Science
241
,
202
205
.
Smale
,
S. T.
and
Baltimore
,
D.
(
1989
).
The ‘initiator’ as a transcriptional control element
.
Cell
57
,
1259
1273
.
Smith
,
D. B.
and
Johnson
,
K. S.
(
1988
).
Single-step purification of polypeptides expressed in Escherichia coli as fusions with glutathionine S-transferase
.
Gene
67
,
31
40
.
Smith
,
M. R.
and
Greene
,
W. C.
(
1989
).
The same 50-kDa cellular protein binds to the negative regulatory elements of the interleukin 2 receptor a-chain gene and the human immunodeficiency virus type 1 long terminal repeat
.
Proc. Nat. Acad. Sci. USA
86
,
8526
8530
.
Trainor
,
C. D.
,
Evans
,
T.
,
Felsenfeld
,
G.
and
Boguski
,
M. S.
(
1990
).
Structure and evolution of a human erythroid transcription factor
.
Nature
343
,
92
96
.
Tsai
,
S.-F.
,
Martin
,
D. I. K.
,
Zon
,
L. I.
,
D’Andrea
,
A. D.
,
Wong
,
G. G.
and
Orkin
,
S. H.
(
1989
).
Cloning of cDNA for the major DNA-binding protein of the erythroid lineage through expression in mammalian cells
.
Nature
339
,
446
451
.
Tsai
,
S.-F.
,
Strauss
,
E.
and
Orkin
,
S. H.
(
1991
).
Functional analysis and in vivo footprinting implicate the erythroid transcription factor GATA-1 as a positive regulator of its own promoter
.
Genes Dev
.
5
,
919
931
.
Vaishnav
,
Y. N.
and
Wong-Staal
,
F.
(
1991
).
The biochemistry of AIDS
.
Ann. Rev. Biochem
.
60
,
577
630
.
Wall
,
L.
,
deBoer
,
E.
and
Grosveld
,
F.
(
1988
).
The human p-globin gene 3’ enhancer contains multiple binding sites for an erythroid-specific protein
.
Genes Dev
.
2
,
1089
1100
.
Webster
,
N. J. G.
,
Green
,
S.
,
Jin
,
J. R.
and
Chambón
,
P.
(
1988
).
The hormone-binding domains of the estrogen and glucocorticoid receptors contain an inducible transcription activation function
.
Cell
54
,
199207
.
Yamamoto
,
M.
,
Ko
,
L. J.
,
Leonard
,
M. W.
,
Beug
,
H.
,
Orkin
,
S. H.
and
Engel
,
J. D.
(
1990
).
Activity and tissue-specific expression of the transcription factor NF-E1 multigene family
.
Genes Dev
.
4
,
1650
1662
.
Zon
,
L. I.
,
Mather
,
C.
,
Burgess
,
S.
,
Bolee
,
M. E.
,
Harland
,
R. M.
and
Orkin
,
S. H.
(
1991
).
Expression of GATA-binding proteins during embryonic development in Xenopus laevis
.
Proc. Nat. Acad. Sci. USA
88
,
10642
10646
.