In vertebrates, transcriptional regulators of the GATA family appear to have a conserved function in differentiation and organ development. GATA-1, -2 and -3 are required for different aspects of hematopoiesis, while GATA-4, -5 and -6 are expressed in various organs of endo-dermal origin, such as intestine and liver, and are impli-cated in endodermal differentiation. Here we report that the Drosophila gene serpent (srp) encodes the previously described GATA factor ABF. The multiple functions of srp in Drosophila suggest that it is an ortholog of the entire ver-tebrate Gata family. srp is required for the differentiation and morphogenesis of the endodermal gut. Here we show that it is also essential for Drosophila hematopoiesis and for the formation of the fat body, the insect organ analogous to the liver. These findings imply that some aspects of the molecular mechanisms underlying blood cell development as well as endodermal differentiation are early acquisitions of metazoan evolution and may be common to most higher animals.

The alimentary canal of Drosophila is an excellent model system for studying molecular mechanisms which underlie organogenesis. It is composed of four major components: the foregut, the midgut, the Malpighian tubules and the hindgut. The midgut is the endodermal part of the gut. It is formed from two primordia which are spatially separated by the mesoderm primordium ventrally and by the primordium of the ectoderm in the dorsolateral region of the blastoderm embryo. The anterior aspect of the endoderm (anterior midgut) is derived from the ventral side of the anterior pole. In addition, cells that originate from the anterior tip of the ventral furrow, also called anterior midgut invagination, may contribute to the anterior midgut. The posterior part of the endoderm (the posterior midgut) derives from the posterior pole of the embryo and becomes internalized during the amnioproctodeal invagination. Both parts of the midgut later lose their epithelial properties, become mesenchymal and migrate towards each other. When the two parts have met on both sides of the yolk, they have adopted an epithelial organization again. Subsequently, the cells migrate ventrally and dorsally to surround the yolk and form the tube of the midgut (for a review see Skaer, 1993).

The major components of the digestive tract apart from the midgut are ectodermal in origin. The hindgut and the Malpighian tubules develop from the proctodeum which originates from the region anterior to the primordium of the posterior midgut in the blastoderm. Both of these primordia are internalized during the amnioproctodeal invagination. They become morphologically distinct when the pouches of the developing Malpighian tubules begin to evaginate from the proctodeum and when the posterior midgut cells slightly later lose their epithelial character and become mesenchymal. The foregut develops from the stomodeum which invaginates in the region anterior to the ventral furrow about two hours after the proctodeum and the posterior midgut have invaginated.

The terminal gap gene huckebein (hkb) establishes the primordia of both anterior and posterior midgut (Weigel et al., 1990; Brönner et al., 1994; Reuter and Leptin, 1994). It sets the border between endoderm and mesoderm, and it is essential for the invagination of the posterior midgut primordium from the posterior pole. However, its action is not entirely specific to the endoderm, since in hkb embryos the stomodeum also fails to invaginate. Hence, most of the foregut does not form (Reuter and Leptin, 1994). hkb encodes a transcription factor with a zinc finger binding domain of the Sp1/egr type (Brönner et al., 1994). We have proposed previously that the gene serpent (srp) is one of the genes which are regulated by hkb, and acts as a selector gene in midgut development (Reuter, 1994). srp is essential for organ-specific morphogenesis and differentiation, and it is required to prevent the midgut primordia from adopting the fate of ectodermal foregut or hindgut, respectively. srp does not participate in other functions of hkb, since it is not implicated in setting the borders of the midgut primordia and is not required for the gastrula-tion movements of these cells or in the invagination of the stomodeum. However, it should be noted that in addition to its function in midgut development, srp is also essential for differ-entiation of the yolk cell and of the extraembryonic amnioserosa (Reuter, 1994). We were interested to elucidate the molecular nature of the srp gene product because of its fun-damental role in endodermal development. In this paper we present evidence that srp encodes a protein which belongs to the GATA family of transcription factors, previously described as ABF or dGATAb (Abel et al., 1993). In vertebrates tran-scriptional regulators of the GATA family apparently have a conserved function in differentiation and organ development. GATA-1 is required for primitive and definitive erythropoiesis (Pevny et al., 1991), GATA-2 for early hematopoiesis (Tsai et al., 1994), while GATA-3 is implicated in the differentiation of T-lymphocytes (Ko et al., 1991). Other members of the family, GATA-4, -5 and -6, are expressed during development in various organs of endodermal origin (Arceci et al., 1993; Laverriere et al., 1994) and are implicated in endodermal differentiation (Tamura et al., 1993; Soudais et al., 1995). Since srp is essential for endodermal development and, as we also show, for hematopoiesis in Drosophila, we consider srp as a gene which is functionally homologous to several, if not all members of the vertebrate Gata family.

Fly stocks

st srp6Ge and ru h th st cu srp9Lsr e ca: amorphic alleles and probably strong hypomorphic allele (Jürgens et al., 1984). Both alleles form RNA, in homozygous srp6G embryos no SRP protein is detectable (data not shown). st e and ru h th st cu sr e ca: parental chromosomes of srp6G and srp9L, respectively (obtained from the Tübingen stock col-lection). srpAS (known as Ins(3R)neo45) and srpPZ (known as Ins(3R)PZ1549): non-complementing P element insertions in 89B (Cooley et al., 1988). The new srp alleles srpHO1 to srpHO3 were generated by mobilization of the P element of srpAS (which is not marked by an eye color gene) with the transposase source Δ2-3 on a TM3 balancer (Laski et al., 1986; Reuter et al., 1993b). They were isolated by screening about 700 candidate lines for the visible embryonic srp phenotype. The other srp alleles were isolated as eye color revertants of srpPZ after mobilization of the P element with the stable transposase source and later classified as either full revertants, partial revertants (srpPZRV1-24) or new amorphic alleles (srpPZHO1-20). hkb2: hypomorphic hkb allele; Df(3R)hkbA: commonly used as hkb deficiency (Weigel et al., 1990).

In situ detection of RNA and protein

mRNA was detected in situ as described by Tautz and Pfeifle (1989) and protein as described previously (Reuter et al., 1993a). The anti-PEROXIDASIN antibody has been described by Nelson and coworkers (1994).

Microscopy

Pictures were taken on a Zeiss Axiophot microscope using Kodak Ektachrome 64T or Agfachrome 100RS slide film and were digitized by transfer to Kodak PhotoCDs. The figures were assembled in Adobe Photoshop on a Macintosh PowerPC and were printed on a Kodak dye sublimation printer.

Isolation of the srp locus

Genomic DNA flanking the P element insertion of srpAS was isolated by plasmid rescue (Cooley et al., 1988). Using this DNA as probe a cosmid library was screened (Tamkun et al., 1992). One of the iden-tified genomic clones (#9) contained the srp transcription unit. The P element insertion of srpAS and the deletions associated with srpHO1 to srpHO3 were mapped by Southern blotting. The genomic organization of srp, the sequence of the srp6G and srp9L alleles, the site of the P element insertion of srpPZ and the molecular nature of some of the new srp alleles derived from srpPZ were determined by sequencing of genomic DNA using appropriate primers. DNA from Oregon R flies, from the cosmid #9 and the cDNAs served as wild-type DNA. Mutant DNA was prepared from homozygous srp embryos which were iden-tified on the apple juice agar plates by their visible embryonic phenotype. In the genomic regions where srp9L and srp6G deviated from wild-type also the corresponding parental chromosomes st e and ru h th st cu sr e ca were sequenced. The 5′ end of the abf cDNA is not present in the srp locus and is probably an artifact. The bases 3-84 of the abf cDNA (Abel et al., 1993) are a direct repeat of the bases 398-479. The accession number of the srp cDNA at the EMBL Nucleotide Sequence Database is Y07662.

Phylogenetic analysis

The evolutionary tree displayed in Fig. 2C was generated by the program ‘protpars’ written by J. Felsenstein as part of the Phylip package (Felsenstein, 1988). The tree is not the only possible tree, but it is the most frequent one in the output of 50 independent runs of the program. 98% of the trees showed the same principal arrangement of the vertebrate, the invertebrate and the fungal GATA factors.

Molecular genetics of the srp locus

We cloned and characterized the srp locus using the P-element-induced allele srpAS (Fig. 1). This allele fails to complement the lethality of the two independently isolated, EMS-induced alleles srp9L and srp6G, but it does not affect gut development or the other known srp functions (data not shown and Fig. 5H). About 8 kb to the 5′ side of the P-element of srpAS we identi-fied a transcription unit which we considered as a candidate gene for srp, since it is expressed in the expected embryonic regions, i.e. in the midgut primordia, the amnioserosa and the yolk nuclei (see Fig. 3). Several srp alleles specifically affect this transcription unit. The alleles srpHO1, srpHO2 and srpHO3 were isolated after mobilization of the P element of srpAS. srpHO1 and srpHO3 are associated with small deletions (Fig. 1) and abolish all known srp functions. The srp candidate gene is not transcribed from these alleles, but no other transcription units to the 3′ side of the P element are affected (data not shown). srpHO2, associated with a smaller deletion (Fig. 1), is an allele of intermediate strength and partially reduces srp function in the midgut and amnioserosa. It is associated with a significant reduction of expression in the respective embryonic regions (data not shown). Another P-element-induced srp allele is srpPZ which also results in the lack of srp transcription. The P element of srpPZ is inserted about 30 bp 5′ of the putative srp transcription start site. The insertion itself affects the srp function in srpPZ, since mobilization of the P element led in 28 out of 72 cases to a complete reversion of the phenotype, and restored viability. In 20 cases, amorphic srp alleles were isolated that had been generated by imprecise excision of the P element. The remaining 24 cases were novel hypomorphic srp alleles that are partial revertants of srpPZ with respect to the midgut or germ band retraction phenotype, but which do not complement the other srp alleles (with the exception of srpAS) to adult viability.

Fig. 1.

Structure of the srp locus. The P element insertion Ins(3R)neo45 (srpAS) (Cooley et al., 1988) was used to isolate the srp locus and specifically affects the function of srp in hemocyte development (see Fig. 5). The srp alleles srpHO1 to srpHO3 were isolated after mobilization of the P element of srpAS. srpHO2 is an allele of intermediate strength which partially reduces srp function in the midgut and amnioserosa, and completely abolishes srp function in the blood primordium. The alleles srpHO1 and srpHO3 abolish all known srp functions; the srp gene is not transcribed in these alleles. The P element insertion Ins(3R)PZ1549 (srpPZ), which also results in the lack of srp transcription, acts as an enhancer trap and is inserted about 30 bp 5′ of the putative transcription start site. Within the srp transcription unit the zinc finger motif is indicated in black, the open reading frame is shaded grey and the untranslated regions are open.

Fig. 1.

Structure of the srp locus. The P element insertion Ins(3R)neo45 (srpAS) (Cooley et al., 1988) was used to isolate the srp locus and specifically affects the function of srp in hemocyte development (see Fig. 5). The srp alleles srpHO1 to srpHO3 were isolated after mobilization of the P element of srpAS. srpHO2 is an allele of intermediate strength which partially reduces srp function in the midgut and amnioserosa, and completely abolishes srp function in the blood primordium. The alleles srpHO1 and srpHO3 abolish all known srp functions; the srp gene is not transcribed in these alleles. The P element insertion Ins(3R)PZ1549 (srpPZ), which also results in the lack of srp transcription, acts as an enhancer trap and is inserted about 30 bp 5′ of the putative transcription start site. Within the srp transcription unit the zinc finger motif is indicated in black, the open reading frame is shaded grey and the untranslated regions are open.

Fig. 2.

(A) Predicted sequence of the SRP protein. Our data suggests that the SRP protein is 949 amino acids in size. The C-terminal 680 amino acids (including the zinc finger domain; underlined) are identical with the reported sequence of ABF (Abel et al., 1993). An asterisk indicates the position of the stop codon found in the srp6G allele, triangles indicate the position of the introns. (B) Comparison of the zinc finger domain of SRP with the (if applicable C-terminal) zinc finger of other GATA factors. In the srp9L allele the conserved asparagine N29 is replaced by a lysine. The amino acids diagnostic for GATA-1/2/3 (Evans and Felsenfeld, 1989; Zon et al., 1990, 1991; Yamamoto et al., 1990) and GATA-4/5/6 (Tamura et al., 1993; Kelley et al., 1993; Laverriere et al., 1994) and shared by SRP are boxed. (C) Phylogenetic tree of the GATA factors based on the domain of 66 amino acids shown in B. The unrooted tree displays the most parsimonious way in which the protein sequences could have evolved. The sub-families GATA-1/2/3, GATA-4/5/6 and the fungal GATA factors regulating nitrogen or siderophore metabolism (AREA: Kudla et al., 1990; Haas et al., 1995; NIT-2: Fu and Marzluf, 1990; GLN3: Minehart and Magasanik, 1991; URBS1: Voisard et al., 1993) each form a branch. There is no significance to the length or to the angle of the lines. Every node represents an inferred intermediate in the evolution of the sequences.

Fig. 2.

(A) Predicted sequence of the SRP protein. Our data suggests that the SRP protein is 949 amino acids in size. The C-terminal 680 amino acids (including the zinc finger domain; underlined) are identical with the reported sequence of ABF (Abel et al., 1993). An asterisk indicates the position of the stop codon found in the srp6G allele, triangles indicate the position of the introns. (B) Comparison of the zinc finger domain of SRP with the (if applicable C-terminal) zinc finger of other GATA factors. In the srp9L allele the conserved asparagine N29 is replaced by a lysine. The amino acids diagnostic for GATA-1/2/3 (Evans and Felsenfeld, 1989; Zon et al., 1990, 1991; Yamamoto et al., 1990) and GATA-4/5/6 (Tamura et al., 1993; Kelley et al., 1993; Laverriere et al., 1994) and shared by SRP are boxed. (C) Phylogenetic tree of the GATA factors based on the domain of 66 amino acids shown in B. The unrooted tree displays the most parsimonious way in which the protein sequences could have evolved. The sub-families GATA-1/2/3, GATA-4/5/6 and the fungal GATA factors regulating nitrogen or siderophore metabolism (AREA: Kudla et al., 1990; Haas et al., 1995; NIT-2: Fu and Marzluf, 1990; GLN3: Minehart and Magasanik, 1991; URBS1: Voisard et al., 1993) each form a branch. There is no significance to the length or to the angle of the lines. Every node represents an inferred intermediate in the evolution of the sequences.

Fig. 3.

Expression of srp mRNA at (A) blastoderm stage and (B) after gastrulation. The arrowheads indicate srp expression in the putative hemocyte primordium within the anterior mesoderm primordium (am, anterior midgut primordium; pm, posterior midgut primordium; as, amnioserosa primordium; yk, yolk cell). (C) At stage 10 srp expression disappears from the posterior midgut primordium, but is still strongly present in the anterior midgut primordium which invaginates with the stomodeum. This expression masks part of the mesodermal cells expressing srp in the head when viewed from the side. Secondary srp expression in the mesoderm has become distinguishable in segmentally repeated clusters of cells (asterisks), the putative fat body precursors (as, amnioserosa). (D) At stage 11 the prohemocytes (arrowheads) migrate into the head. The secondary expression in the putative fat body primordium becomes stronger. (E) Same embryo as in D in a horizontal view. (F) Later during stage 11, the fat body precursor cells rearrange and begin to form a sheet on each side of the germ band (arrows). In addition to the fat body also precursors of the somatic mesoderm transiently express srp (open arrowheads). (G) After germ band retraction srp expression has become very weak in the hemocytes which have already dispersed over the interstitial space. srp is strongly expressed in the differentiating fat body (fb) and now also in the developing lymph gland (arrowhead) and the ring gland (arrow). (H) Same embryo as in G seen in a medial horizontal section.

Fig. 3.

Expression of srp mRNA at (A) blastoderm stage and (B) after gastrulation. The arrowheads indicate srp expression in the putative hemocyte primordium within the anterior mesoderm primordium (am, anterior midgut primordium; pm, posterior midgut primordium; as, amnioserosa primordium; yk, yolk cell). (C) At stage 10 srp expression disappears from the posterior midgut primordium, but is still strongly present in the anterior midgut primordium which invaginates with the stomodeum. This expression masks part of the mesodermal cells expressing srp in the head when viewed from the side. Secondary srp expression in the mesoderm has become distinguishable in segmentally repeated clusters of cells (asterisks), the putative fat body precursors (as, amnioserosa). (D) At stage 11 the prohemocytes (arrowheads) migrate into the head. The secondary expression in the putative fat body primordium becomes stronger. (E) Same embryo as in D in a horizontal view. (F) Later during stage 11, the fat body precursor cells rearrange and begin to form a sheet on each side of the germ band (arrows). In addition to the fat body also precursors of the somatic mesoderm transiently express srp (open arrowheads). (G) After germ band retraction srp expression has become very weak in the hemocytes which have already dispersed over the interstitial space. srp is strongly expressed in the differentiating fat body (fb) and now also in the developing lymph gland (arrowhead) and the ring gland (arrow). (H) Same embryo as in G seen in a medial horizontal section.

Sequence analysis of the transcription unit showed that the srp candidate gene encodes the previously identified Drosophila GATA factor ABF, also known as dGATAb (Abel et al., 1993). However, in genomic and cDNA sequences we noted a few deviations from the published sequence. Our results indicate that the transcribed mRNA is about 4 kb in size and encodes a protein of 949 amino acids, rather than 779 as published previously (Abel et al., 1993). The C-terminal 680 amino acids (including the zinc finger domain) are identical with the reported sequence of ABF (Abel et al., 1993). The putative translation start site of the srp mRNA is in good agreement with the consensus translation start site for that of Drosophila (Cavener, 1987). This mRNA, which we consider as full-length, is efficiently translated in the embryo, in contrast to the truncated mRNA directed by the abf cDNA (data not shown).

To confirm that the srp gene corresponds to abf we have sequenced both EMS-induced srp alleles. The srp6G allele is associated with a nonsense mutation that would lead to the translation of a truncated protein lacking the zinc finger domain (Fig. 2A). This finding is consistent with the amorphic phenotype of homozygous srp6G embryos. In the srp9L allele an asparagine (N29 of the consensus sequence in Fig. 2B) is replaced by a lysine within the second cysteine pair of the zinc finger motif. This mutation probably causes the strong loss-of-function phenotype of homozygous srp9L embryos, since asparagine N29 is apparently essential for the DNA-binding capacity of the GATA factors; this asparagine is present in all members of the family described to date and has been proposed to contact two central DNA residues of the GATA consensus binding site (Omichinski et al., 1993). No polymorphisms were observed which would alter the protein sequence apart from a poly-glutamine stretch 3 amino acids shorter between amino acids 118 and 120 on the st e and the st srp6Ge chromosomes.

Expression of the srp gene

During the blastoderm stage and gastrulation srp is expressed in five regions of the Drosophila embryo (Abel et al., 1993). At the anterior pole srp is found in a region (Fig. 3A,B) which later invaginates through the stomodeum (Fig. 3C) before expression of srp is down-regulated (Fig. 3D). This region is the primordium of the anterior midgut (see Discussion). At the posterior pole, the srp-expressing cells invaginate as the posterior midgut primordium (Fig. 3A,B) and likewise cease expression some time after the invagination (Fig. 3C). The expression of srp in the anterior and the posterior midgut primordium is not observed in hkb embryos (Fig. 4B), while expression in the other domains is initiated normally. Thus, as predicted, srp is downstream of hkb within the genetic hierarchy that directs midgut development.

Fig. 4.

Expression of srp in the primordia of anterior and posterior midgut depends on hkb. srp mRNA was detected in (A) wild-type and (B) hkb embryos at early gastrulation. In the hkb embryo srp expression at the poles, i.e. in the anterior midgut primordium and the posterior midgut primordium, is not initiated, while other aspects are not affected. The weak staining in the yolk nuclei of the embryos in both panels reflects the variability in the detection of this aspect of srp expression.

Fig. 4.

Expression of srp in the primordia of anterior and posterior midgut depends on hkb. srp mRNA was detected in (A) wild-type and (B) hkb embryos at early gastrulation. In the hkb embryo srp expression at the poles, i.e. in the anterior midgut primordium and the posterior midgut primordium, is not initiated, while other aspects are not affected. The weak staining in the yolk nuclei of the embryos in both panels reflects the variability in the detection of this aspect of srp expression.

Fig. 5.

srp is essential for Drosophila hematopoiesis. (A-D) The hemocyte precursors do not proliferate in srp embryos. The hemocyte precursor cells are visualized by their srp expression in (A,C) wild-type embryos and in (B,D) embryos homozygous for the srp6G allele at stage 11. (A,B) Optical sagittal section; (C,D) tangential horizontal section, srp9L gives the same result (E,F) Hemocytes, visualized by an anti-PEROXIDASIN antibody (Nelson et al., 1994), populate the interstitial space in wild-type embryos. (E) Stage 12, lateral view; (F)stage 15, horizontal view. (G) They are not detectable in srp embryos carrying strong or amorphic alleles or (H) in embryos trans-heterozygous for the P-element-induced allele srpAS and the strong srp allele srp9L. (I,J) The srp expression in the putative hemocyte primordium within the mesoderm (arrowhead) is specifically decreased in (J) hemizygous srpAS embryos as compared to (I) wild type.

Fig. 5.

srp is essential for Drosophila hematopoiesis. (A-D) The hemocyte precursors do not proliferate in srp embryos. The hemocyte precursor cells are visualized by their srp expression in (A,C) wild-type embryos and in (B,D) embryos homozygous for the srp6G allele at stage 11. (A,B) Optical sagittal section; (C,D) tangential horizontal section, srp9L gives the same result (E,F) Hemocytes, visualized by an anti-PEROXIDASIN antibody (Nelson et al., 1994), populate the interstitial space in wild-type embryos. (E) Stage 12, lateral view; (F)stage 15, horizontal view. (G) They are not detectable in srp embryos carrying strong or amorphic alleles or (H) in embryos trans-heterozygous for the P-element-induced allele srpAS and the strong srp allele srp9L. (I,J) The srp expression in the putative hemocyte primordium within the mesoderm (arrowhead) is specifically decreased in (J) hemizygous srpAS embryos as compared to (I) wild type.

On the dorsal side of the embryo srp gene product is seen in the primordium of the amnioserosa (Fig. 3A,B) and later in the amnioserosa itself while it connects the dorsal edges of the germ band (Fig. 3C-F). In the center of the embryo srp is expressed by the yolk nuclei (Fig. 3A), most of which subse-quently migrate to the periphery of the yolk (Fig. 3B). Finally, srp is expressed in a patch of cells within the mesoderm pri-mordium, which was not expected from the previously described phenotype of srp. These cells invaginate with the ventral furrow anterior to the cephalic furrow (Fig. 3A,B) and become located laterally to the stomodeum which in part also expresses srp (Fig. 3C). Slightly later they differentiate into prohemocytes which migrate into the head (Fig. 3D-F). Sub-sequently, these cells become distributed throughout the body and differentiate as mature hemocytes, at which point srp expression is downregulated (Fig. 3G). We propose that the mesodermal patch of srp expression constitutes the hemocyte primordium at blastoderm stage.

srp function in hematopoiesis

The expression of srp in the putative hemocyte primordium raised the question of whether srp has a function in hematopoiesis. The Drosophila blood cells (hemocytes) are part of the insect immune system and mainly develop into macrophages. They are known to derive from the anterior part of the mesoderm, phagocytose apoptotic cells in the embryo and express a number of specific gene products, such as PER-OXIDASIN (Tepass et al., 1994; Nelson et al., 1994). In srp embryos the primordium of the hemocytes invaginates with the ventral furrow. Expression of mutant srp RNA is maintained during germ band extension (Fig. 5B,D), however, the cells fail to proliferate or to migrate (Fig. 5B,D) and subsequently die (data not shown). As a consequence srp embryos are devoid of any mature hemocytes (Fig. 5G) which by stage 12 would normally be found throughout the embryo (Fig. 5E). Thus, srp is required for the development of hemocytes.

Interestingly, srpAS, the P-element-induced allele which does not affect the function of srp in gut, yolk or amnioserosa, specifically disturbs hemocyte differentiation. In homozygous srpAS embryos the number of hemocytes is severely reduced (data not shown), and no hemocytes are formed in embryos trans-heterozygous for srpAS and either of the strong EMS-induced srp alleles: srp6G and srp9L (Fig. 5H). Consistent with the failure of hemocyte development a drastic decrease of srp expression is observed in the putative hemocyte primordium of hemizygous srpAS embryos (Fig. 5J). Other aspects of srp expression are not altered. Thus, the P element insertion of srpAS specifically affects the srp function required for devel-opment of the hemocytes, and it is likely that the P-element is inserted into a regulatory region of the gene which is essential for the anterior mesodermal srp expression.

srp function in fat body development

During germ band extension expression of srp is initiated in the developing fat body, one of the subdivisions of the mesoderm (Abel et al., 1993; see also Fig. 3C). It is expressed in a series of nine clusters of cells located bilaterally within the mesoderm of thoracic segment t2 to the abdominal segment a7. These clusters begin to form a sheet of cells on each side of the embryo during late stage 11 (Fig. 3F) from which the fat body develops (Fig. 3G,H). srp continues to be expressed in the fat body until the end of embryogenesis. The GATA transcription factor encoded by srp has been described to be sufficient to activate the fat body-specific expression of Alcohol dehydrogenase (Adh) (Abel et al., 1993). In fact, srp is required for that expression, since Adh transcription in the mesoderm is not detected in srp mutant embryos (data not shown). However, the lack of Adh expression in these embryos does not necessarily support the notion of a direct activation of Adh transcription by srp in vivo, since fat body development is impaired at a very early stage in the mutants.

The fat body precursors, which can be visualized by their srp RNA expression, are present in embryos homozygous for the srp9L or the srp6G allele (Fig. 6B). However, the cells do not proliferate and do not rearrange to form the continuous sheet of cells observed in wild-type embryos at late stage 11 (Fig. 6A). Furthermore, the early events of fat body differentiation do not take place in srp embryos. seven-up (svp) has been described as a gene with an important role in fat body development (Hoshizaki et al., 1994). Expression of svp commences in the fat body precursors at about stage 11 and is main-tained in the maturing fat body until stage 15 (Hoshizaki et al., 1994). In srp embryos svp expression is not initiated in the mesoderm at a position which would correspond to the fat body or its precursors (Fig. 6D,F), however the expression of svp within the ventral nervous system is not affected. Thus, srp is essential for the earliest known steps in the morphogenesis and the differentiation of the fat body.

Fig. 6.

Fat body development is affected in srp embryos. (A,B) The fat body precursor cells (arrows) do not proliferate and do not form the typical continuous sheet of cells in srp embryos. srp expression was detected in (A) wild-type and (B) srp9L embryos at late stage 11 (the same result is seen with srp6G). The region from the second thoracic to the third abdominal segment is shown; the embryonic midline corresponds to the bottom of the panel. The srp-expressing cells between the fat body and the midline are probably precursors of the somatic mesoderm (open triangles), as judged by their position (Borkowski et al., 1995). The magnification is about 2× higher than in C-H. (C-H) svp expression in the fat body depends on srp function. svp expression is shown (C,E,G) in wild-type and (D,F, H) in srp embryos at about stage 11 (C-F) and at about stage 14 (G,H). (C,E,G) While svp is strongly expressed in the developing fat body of wild-type embryos (arrowheads), (D,F,H) no such svp expression is detectable in the srp embryos. svp-expressing cells are visible in the ventral nervous system (D,F,H) and also in the periphery. The embryos are viewed in (C,D) parasagittal, (E,F) tangential horizontal and (G,H) medial horizontal sections. C and E as well as D and F show the same embryo, respectively. The srp embryo shown in H is grossly distorted due to the failure of germ band retraction.

Fig. 6.

Fat body development is affected in srp embryos. (A,B) The fat body precursor cells (arrows) do not proliferate and do not form the typical continuous sheet of cells in srp embryos. srp expression was detected in (A) wild-type and (B) srp9L embryos at late stage 11 (the same result is seen with srp6G). The region from the second thoracic to the third abdominal segment is shown; the embryonic midline corresponds to the bottom of the panel. The srp-expressing cells between the fat body and the midline are probably precursors of the somatic mesoderm (open triangles), as judged by their position (Borkowski et al., 1995). The magnification is about 2× higher than in C-H. (C-H) svp expression in the fat body depends on srp function. svp expression is shown (C,E,G) in wild-type and (D,F, H) in srp embryos at about stage 11 (C-F) and at about stage 14 (G,H). (C,E,G) While svp is strongly expressed in the developing fat body of wild-type embryos (arrowheads), (D,F,H) no such svp expression is detectable in the srp embryos. svp-expressing cells are visible in the ventral nervous system (D,F,H) and also in the periphery. The embryos are viewed in (C,D) parasagittal, (E,F) tangential horizontal and (G,H) medial horizontal sections. C and E as well as D and F show the same embryo, respectively. The srp embryo shown in H is grossly distorted due to the failure of germ band retraction.

We have identified the previously described GATA factor ABF (Abel et al., 1993) as the product of the gene srp. srp is expressed in five principal regions of the embryo: the midgut primordium, the yolk, the putative hemocyte primordium, the amnioserosa and the fat body. In all of these regions srp is required for proper development (Reuter, 1994 and Figs 5, 6). Three srp alleles are associated with small deletions which specifically affect the srp transcription unit (Fig. 1). One P element insertion is associated with a decrease in srp tran-scription in the putative hemocyte primordium and with a specific failure in hemocyte development (Fig. 5). This insertion probably affects a regulatory region within the srp gene that is essential for this aspect of the expression pattern. One other P element insertion abolishes all srp functions: a reversion of the insertion alleviates the phenotype. Finally, the two EMS-induced srp alleles are associated with mutations in the coding sequence which in the case of srp6G eliminate, and in the case of srp9L most likely eliminate the ability of SRP to act as a transcriptional regulator. The nonsense mutation in srp6G results in a truncated protein lacking the zinc finger domain. The missense mutation of srp9L replaces an asparagine (N29) within the DNA binding domain which is conserved in all known members of the GATA family (Fig. 2B). Exchanges of amino acids conserved to the same degree in the zinc finger domain of the fungal GATA factor AREA are also associated with a loss of function (Kudla et al., 1990). Furthermore, the asparagine N29 is one of the residues of the GATA zinc finger domain which contacts the core of the consensus DNA binding motif (Omichinski et al., 1993), a function which cannot be fulfilled by the lysine found at this position in srp9L. Taken together, our findings establish that the transcription unit encoding the GATA factor ABF is identical to srp.

srp function in the Drosophila embryo

srp is required for proper differentiation and morphogenesis of the midgut and has been suggested to act as a selector gene in gut development acting downstream of hkb (Reuter, 1994). The pattern of srp expression and its dependence on hkb in the midgut primordia are consistent with such a function (Fig. 4). However, a role in terminal midgut differentiation can probably be excluded. The SRP protein dis-appears from the tissue before germ band retraction is completed (data not shown). Of particular interest is the domain of srp expression at the anterior pole. This domain is located at the ventral side of the embryo anterior to the ventral furrow. It corresponds to the stomodeal anterior midgut primordium which has been shown by lineage analysis to contribute to the anterior part of the anterior midgut (Technau and Campos- ortega, 1985). In hkb embryos no anterior midgut is formed, and the only srp expression domain missing in the anterior half of the embryo is the anteroventral domain (Fig. 4). We therefore propose that the anterior midgut is exclusively derived from this single primordium at the ventral side of the anterior pole. That would imply that most cells of the anterior midgut are internalized by the stomodeal invagination (Fig. 3C). Only few cells invaginate as the front edge of the ventral furrow (visible for instance in the Figs 3B or 5I).

SRP (ABF) is capable of activating tran-scription of fat-body-specific genes (Abel et al., 1993; Lossky and Wensink, 1995). Here we have shown that it is required for the early steps in fat body morphogenesis and differentiation (Fig. 6). Based on the early onset of expression, srp might even determine the fat body precursors and integrate the positional information that specifies them in the mesoderm. Since a mature fat body is not formed in srp embryos, it has to date not been possible to rigorously test whether srp is indeed also required for the terminal differen-tiation in the fat body, e.g. for Adh transcription. SRP, however, is the only known GATA factor expressed at this time in the fat body, and therefore it is very likely that it is responsible for tissue-specific gene activation. srp would then have a biphasic function in the physiology of the fat body, for the early mor-phogenesis and differentiation and for the maintenance of organ-specific gene expression.

A striking finding is the function of srp in the hematopoiesis of Drosophila. No mature hemocytes are formed in srp embryos, and the earlier differentiation of prohemocytes is also impaired in the mutants (Fig. 5). Furthermore, expression pattern and phenotype of srp suggest that a particular region within the anterior mesoderm is the primordium of the hemocytes. The cells of the expression domain located ventrally in front of the cephalic furrow can be traced until they become morphologically distinguishable as hemocytes. Moreover, the srpAS allele, which specifically affects hemocyte development, reduces srp expression in this domain of the mesoderm (Fig. 5H,J). However, lineage analysis is required to confirm that the proposed primordium exclusively gives rise to hemocytes, as the srp expression pattern and srp phenotype suggest. srp expression in the putative hemocyte primordium commences at blastoderm stage (Fig. 3A) and is main-tained as protein expression when the mature hemocytes already migrate through the embryo (data not shown). It is therefore conceivable that srp has a biphasic function in hemocyte develop-ment as possibly it does in the fat body. srp might be required for the specification of the blood cells within the mesoderm at an early stage and subsequently later for gene expression during their differen-tiation and maturation.

srp is a Drosophila ortholog of the vertebrate Gata gene family

Since srp is required for hematopoiesis in Drosophila, as Gata-1, -2, and probably -3 are in mice, and for endodermal development, as Gata-4, -5 and -6 probably are, srp can be considered as a Drosophila ortholog of the entire vertebrate GATA family. We do not know yet to what degree the processes of hematopoiesis and endodermal development are homologous at the molecular level in insects and in vertebrates. For instance, it is unclear to what extent the sets of genes that are regulated by these GATA factors overlap in different phyla. Nevertheless, we suspect that these processes are far more similar than previously expected, supporting the notion that the principal molecular mechanisms underlying organogenesis are early phylogenetic achievements.

The Gata-4/5/6 genes are also implicated in the differentiation of the vertebrate heart. srp does not have a corresponding role in Drosophila, and it is open to discussion whether a GATA factor exists in the fly which functions in this respect. One other GATA factor, PANNIER (PNR) or dGATAa, has been found in Drosophila which is reportedly expressed in the amnioserosa and the dorsal epidermis adjacent to the cardiac progenitor cells (Ramain et al., 1993; Winick et al., 1993). However, so far there is no indication that pnr fulfills a function homologous to the one of the vertebrate Gata genes. The failure of pnr mutant embryos to form a proper heart (data not shown) is probably due to the general defects in differentiation and morphogenesis on the dorsal side of the body.

The hematopoietic system of Drosophila is far simpler than that of a vertebrate. Thus, it is plausible that in vertebrates three structurally distinct GATA factors are required in overlapping compartments of the hematopoietic system, while only one orthologous GATA factor, SRP, functions in this capacity in the fly. The two other GATA factors described in Drosophila, PNR (Ramain et al., 1993; Winick et al., 1993) and dGATAc (Lin et al., 1995), are not expressed in the hemocyte primordium. Similarly, the derivatives of the endoderm are more complex in vertebrates than in Drosophila, and thus only one GATA factor might be required in the fly. However, it is highly probable that the function of SRP only corresponds to earlier aspects of the function of GATA-4/5/6, since srp expression disappears from the midgut primordium before terminal differ-entiation is initiated. This later aspect might be carried out by dGATAc which is expressed in these cells after the decline in the level of srp transcripts (Lin et al., 1995).

It is striking that SRP combines functions of both the ‘blood-specific’ (GATA-1/2/3) and the ‘endoderm-specific’ (GATA-4/5/6) vertebrate GATAs. This is particularly astonishing if one considers the presence of only one zinc finger domain in SRP as opposed to the two zinc finger domains in the vertebrate GATAs. Future studies will show whether this property is a new acquisition of advanced insects or whether it is an original feature of more primitive organogenesis. The ‘combined’ function of SRP may in part derive from the chimaeric nature of its DNA-binding domain (Fig. 2B,C). Like three of the other four invertebrate GATA factors, ELT-1 (Spieth et al., 1991), ELT-2 (Hawkins and McGhee, 1995) and PNR, the zinc finger domain of SRP has a structure that lies between the two vertebrate GATA sub-families (Fig. 2C). Therefore, these invertebrate GATA factors might be structurally closer to the ancestors of the vertebrate factors. However, the specific activity of SRP in the Drosophila embryo is certainly deter-mined by the expression pattern of the gene. The specificity must depend on the direct or indirect interaction with other gene products which are present in the same primordium or tissue, and it will be interesting to investigate the extent to which these interactions are conserved between Drosophila and vertebrates.

We thank Drs M. Brennan, N. Brown, L. Fessler, J. Fessler M. Mlodzik and J. Tamkun and the Bloomington and Tübingen stock centers for reagents and Drs E. Knust, M. Leptin and particularly R. Wilson for comments on the manuscript. We are indebted to Dr José Casal for advice and help during the initial phase of the project. This work was supported by the Deutsche Forschungsgemeinschaft as part of the SFB 243.

Abel
,
T.
,
Michelson
,
A. M.
and
Maniatis
,
T.
(
1993
).
A Drosophila GATA family member that binds to Adh regulatory sequences is expressed in the developing fat body
.
Development
119
,
623
633
.
Arceci
,
R. J.
,
King
,
A. A. J.
,
Simon
,
M. C.
,
Orkin
,
S. H.
and
Wilson
,
D. B.
(
1993
).
Mouse GATA-4: a retinoic acid-inducible GATA-binding transcription factor expressed in endodermally derived tissues and heart
.
Mol. Cell. Biol.
13
,
2235
2246
.
Borkowski
,
O. M. D.
,
Brown
,
N. H.
and
Bate
,
M.
(
1995
).
Anterior-posterior subdivision and the diversification of the mesoderm in Drosophila
.
Development
121
,
4183
4193
.
Brönner
,
G.
,
Chu-la Graff
,
Q.
,
Doe
,
C. Q.
,
Cohen
,
B.
,
Weigel
,
D.
,
Taubert
,
H.
and
Jäckle
,
H.
(
1994
).
Sp1/egr-like zinc-finger protein required for endoderm specification and germ-layer formation in Drosophila
.
Nature
369
,
664
668
.
Cavener
,
D. R.
(
1987
).
Comparison of the consensus flanking translational start sites in Drosophila and vertebrates
.
Nucl. Acids Res.
15
,
1353
1361
.
Cooley
,
L.
,
Kelley
,
R.
and
Spradling
,
A.
(
1988
).
Insertional mutagenesis of the Drosophila genome with single P elements
.
Science
329
,
1121
1128
.
Evans
,
T.
and
Felsenfeld
,
G.
(
1989
).
The erythroid-specific transcription factor Eryf1: a new finger protein
.
Cell
58
,
877
885
.
Felsenstein
,
J.
(
1988
).
Phylogenies from molecular sequences: inference and reliability
.
Ann. Rev. Genet.
22
,
521
565
.
Fu
,
Y.-H.
and
Marzluf
,
G. A.
(
1990
).
nit-2, the major nitrogen regulatory gene of Neurospora crassa, encodes a protein with putative zinc finger DNA- binding domain
.
Mol. Cell. Biol.
10
,
1056
1065
.
Haas
,
H.
,
Bauer
,
B.
,
Redl
,
B.
,
Stöffler
,
G.
and
Marzluf
,
G. A.
(
1995
).
Molecular cloning and analysis of nre, the major nitrogen regulatory gene of Penicillium chrysogenum
.
Curr. Genet.
27
,
150
158
.
Hawkins
,
M. G.
and
McGhee
,
J. D.
(
1995
).
elt-2, a second GATA factor from the nematode Caenorhabditis elegans
.
J. Biol. Chem.
270
,
14666
14671
.
Hoshizaki
,
D. K.
,
Blackburn
,
T.
,
Price
,
C.
,
Ghosh
,
M.
,
Miles
,
K.
,
Ragucci
,
M.
and
Sweis
,
R.
(
1994
).
Embryonic fat-cell lineage in Drosophila melanogaster
.
Development
120
,
2489
2499
.
Jürgens
,
G.
,
Wieschaus
,
E.
,
Nüsslein-Volhard
,
C.
and
Kluding
,
H.
(
1984
).
Mutations affecting the pattern of the larval cuticle in Drosophila melanogaster. II. Zygotic loci on the third chromosome
.
Roux’s Arch. Dev. Biol.
193
,
283
295
.
Kelley
,
C.
,
Blumberg
,
H.
,
Zon
,
L. I.
and
Evans
,
T.
(
1993
).
GATA-4 is a novel transcription factor expressed in endocardium of the developing heart
.
Development
118
,
817
827
.
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 cis-regulatory element within the human T- cell receptor δ gene enhancer
.
Mol. Cell. Biol.
11
,
2778
2784
.
Kudla
,
B.
,
Caddick
,
M. X.
,
Langdon
,
T.
,
Martinez-Rossi
,
N. M.
,
Bennett
,
C. F.
,
Sibley
,
S.
,
Davies
,
R. W.
and
Arst
,
H. N.
(
1990
).
The regulatory gene areA mediating nitrogen metabolite repression in Aspergillus nidulans. Mutations affecting specificity of gene activation alter a loop residue of a putative zinc finger
.
EMBO J.
9
,
1355
1364
.
Laski
,
F. A.
,
Rio
,
D. C.
and
Rubin
,
G. M.
(
1986
).
Tissue specificity of Drosophila P element transposition is regulated at the level of mRNA splicing
.
Cell
44
,
7
19
.
Laverriere
,
A. C.
,
MacNeill
,
C.
,
Mueller
,
C.
,
Poelmann
,
R. E.
,
Burch
,
J. B. E.
and
Evans
,
T.
(
1994
).
GATA-4/5/6, a subfamily of three transcription factors transcribed in developing heart and gut
.
J. Biol. Chem.
269
,
23177
23184
.
Lin
,
W.-H.
,
Huang
,
L.-H.
,
Yeh
,
J.-Y.
,
Hoheisel
,
J.
,
Lehrach
,
H.
,
Sun
,
Y. H.
and
Tsai
,
S.-F.
(
1995
).
Expression of a Drosophila GATA transcription factor in multiple tissues in the developing embryo
.
J. Biol. Chem.
270
,
25150
25158
.
Lossky
,
M.
and
Wensink
,
P. C.
(
1995
).
Regulation of Drosophila yolk protein genes by an ovary-specific GATA factor
.
Mol. Cell. Biol.
15
,
6943
6952
.
Minehart
,
P. L.
and
Magasanik
,
B.
(
1991
).
Sequence and expression of GLN3, a positive nitrogen regulatory gene of Saccharomyces cerevisiae encoding a protein with a putative zinc finger DNA-binding domain
.
Mol. Cell Biol.
11
,
6216
6228
.
Nelson
,
R. E.
,
Fessler
,
L. I.
,
Takagi
,
Y.
,
Blumberg
,
B.
,
Keene
,
D. R.
,
Olson
,
P. F.
,
Parker
,
C. G.
and
Fessler
,
J. H.
(
1994
).
Peroxidasin: a novel enzyme- matrix protein of Drosophila development
.
EMBO J.
13
,
3438
3447
.
Omichinski
,
J. G.
,
Clore
,
G. M.
,
Schaad
,
O.
,
Felsenfeld
,
G.
,
Trainor
,
C.
,
Appella
,
E.
,
Stahl
,
S. J.
and
Gronenborn
,
A. M.
(
1993
).
NMR structure of a specific DNA complex of Zn-containing DNA binding domain of GATA-1
.
Science
261
,
438
446
.
Pevny
,
L.
,
Simon
,
M. C.
,
Robertson
,
E.
,
Klein
,
W. H.
,
Tsai
,
S.-F.
,
D’Agati
,
V.
,
Orkin
,
S. H.
and
Costantini
,
F.
(
1991
).
Erythroid differentiation in chimaeric mice blocked by a targeted mutation in the gene for transcription factor GATA-1
.
Nature
349
,
257
260
.
Ramain
,
P.
,
Heitzler
,
P.
,
Haenlin
,
M.
and
Simpson
,
P.
(
1993
).
pannier, a negative regulator of achaete and scute in Drosophila, encodes a zinc finger protein with homology to the vertebrate transcription factor GATA-1
.
Development
119
,
1277
1291
.
Reuter
,
R.
(
1994
).
The gene serpent has homeotic properties and specifies endoderm versus ectoderm within the Drosophila gut
.
Development
120
,
1123
1135
.
Reuter
,
R.
,
Grunewald
,
B.
and
Leptin
,
M.
(
1993a
).
A role for the mesoderm in endodermal migration and morphogenesis in Drosophila
.
Development
119
,
1135
1145
.
Reuter
,
G.
,
Hoffmann
,
G.
,
Dorn
,
R.
,
Gausz
,
J.
and
Saumweber
,
H.
(
1993b
).
Construction and characterization of a TM3 balancer carrying P[(ry[+])Δ2-3] as a stable transposase source
.
Dros. Inf. Serv.
72
,
78
.
Reuter
,
R.
and
Leptin
,
M.
(
1994
).
Interacting functions of snail, twist and huckebein during the early development of germ layers in Drosophila
.
Development
120
,
1137
1150
.
Skaer
,
H.
(
1993
).
The alimentary canal
. In
The Development of Drosophila melanogaster
. (ed.
Bate
,
M.
Martinez-Arias
,
A.
), pp
941
1012
.
Plainview, NY
:
Cold Spring Harbor Laboratoy Press
.
Soudais
,
C.
,
Bielinska
,
M.
,
Keikinheimo
,
M.
,
MacArthur
,
C. A.
,
Narita
,
N.
,
Saffitz
,
J. E.
,
Simon
,
M. C.
,
Leiden
,
J. M.
and
Wilson
,
D. B.
(
1995
).
Targeted mutagenesis of the transcription factor GATA-4 gene in mouse embryonic stem cells disrupts visceral endoderm differentiation in vitro
.
Development
121
,
3877
3888
.
Spieth
,
J.
,
Shim
,
Y. H.
,
Lea
,
K.
,
Conrad
,
R.
and
Blumenthal
,
T.
(
1991
).
elt- 1, an embryologically expressed Caenorhabditis elegans gene homologous to the GATA transcription factor family
.
Mol. Cell. Biol.
11
,
4651
4659
.
Tamkun
,
J. W.
,
Deuring
,
R.
,
Scott
,
M. P.
,
Kissinger
,
M.
,
Pattatucci
,
A. M.
,
Kaufman
,
T. C.
and
Kennison
,
J. A.
(
1992
).
brahma: A regulator of Drosophila homeotic genes structurally related to the yeast transcriptional activator SNF2/SWI2. Cell
68
,
561
572
.
Tamura
,
S.
,
Wang
,
X.-H.
,
Maeda
,
M.
and
Futai
,
M.
(
1993
).
Gastric DNA- binding proteins recognize upstream sequence motifs of parietal cell-specific genes
.
Proc. Natl. Acad. Sci. USA
90
,
10876
10880
.
Tautz
,
D.
and
Pfeifle
,
C.
(
1989
).
A non-radioactive in situ hybridization method for the localization of specific RNAs in Drosophila embryos reveals translational control of the segmentation gene hunchback
.
Chromosoma
98
,
81
85
.
Technau
,
G. M.
and
Campos-Ortega
,
J. A.
(
1985
).
Fate-mapping in wild- type Drosophila melanogaster. II. Injections of horseradish peroxidase in cells of the early gastrula stage
.
Roux’s Arch. Dev. Biol.
194
,
196
212
.
Tepass
,
U.
,
Fessler
,
L. I.
,
Aziz
,
A.
and
Hartenstein
,
V.
(
1994
).
Embryonic origin of hemocytes and their relationship to cell death in Drosophila
.
Development
120
,
1829
1837
.
Tsai
,
F. Y.
,
Keller
,
G.
,
Kuo
,
F. C.
,
Weiss
,
M.
,
Chen
,
J. Z.
,
Rosenblatt
,
M.
,
Alt
,
F. W.
and
Orkin
,
S. H.
(
1994
).
An early haematopoietic defect in mice lacking the transcription factor GATA-2
.
Nature
371
,
221
226
.
Voisard
,
C.
,
Weang
,
J.
,
McEvoy
,
J. L.
,
Xu
,
P.
and
Leong
,
S. A.
(
1993
).
urbS1, a gene regulating siderophore biosynthesis in Ustilago maydis, encodes a protein similar to the erythroid transcription factor GATA-1
.
Mol. Cell Biol.
13
,
7091
7100
.
Weigel
,
D.
,
Jürgens
,
G.
,
Klingler
,
M.
and
Jäckle
,
H.
(
1990
).
Two gap genes mediate maternal terminal pattern information in Drosophila
.
Science
248
,
495
498
.
Winick
,
J.
,
Abel
,
T.
,
Leonard
,
M. W.
,
Michelson
,
A. M.
,
Chardon-Loriaux
,
I.
,
Holmgren
,
R. A.
,
Maniatis
,
T.
and
Engel
,
J. D.
(
1993
).
A GATA family transcription factor is expressed along the embryonic dorsoventral axis in Drosophila melanogaster
.
Development
119
,
1055
1065
.
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.
,
Tsai
,
S. F.
,
Burgess
,
S.
,
Matsudaira
,
P.
,
Bruns
,
G. A.
and
Orkin
,
S. H.
(
1990
).
The major human erythroid DNA-binding protein (GF-1): primary sequence and localization of the gene to the X chromosome
.
Proc. Natl. Acad. Sci. USA
87
,
668
672
.
Zon
,
L. I.
,
Mather
,
C.
,
Burgess
,
S.
,
Bolce
,
M. E.
,
Harland
,
R. M.
and
Orkin
,
S. H.
(
1991
).
Expression of GATA-binding proteins during embryonic development in Xenopus laevis. Proc. Natl. Acad. Sci. USA
88
,
10642
10646
.