The sequence of a cDNA from the giant gene of Drosophila shows that its product has a basic domain followed by a leucine zipper motif. Both features contain characteristic conserved elements of the b-ZIP family of DNA-binding proteins. Expression of the gene in bacteria or by in vitro translation yields a protein that migrates considerably faster than the protein extracted from Drosophila embryos. Treatment with phosphatase shows that this difference is due to multiple phosphorylation of the giant protein in the embryo. Ectopic expression of the protein in precellular blastoderm embryos produces abnormal phenotypes with a pattern of segment loss closely resembling that of Krüppel mutant embryos. Immunological staining shows that giant, ectopically expressed from the hsp70 promoter, represses the expression of both the Krüppel and knirps segmentation gap genes. The analysis of the interactions between Krüppel, knirps and giant reveals a network of negative regulation. We show that the apparent positive regulation of knirps by Krüppel is in fact mediated by a negative effect of Krüppel on giant and a negative effect of giant on knirps. giant protein made in bacteria or in embryos binds in vitro to the Krüppel regulatory elements CD1 and CD2 and recognizes a sequence resembling the binding sites of other b-ZIP proteins.

The segmented body pattern of Drosophila is established by a rapid cascade of gene regulation that occurs in the early stages of embryonic development (reviewed by Ingham, 1988). This process is initiated by the asymmetric distribution of maternal gene products in the unfertilized egg and culminates with the appropriate expression of the zygotic homeotic genes in the segment primordia. The maternal factors that initiate the process are distributed in gradients (Driever and Niisslein-Volhard, 1988; Tautz, 1988). In the best characterized example, the bicoid protein gradient has morphogenetic properties since it results in a concentration-dependent activation or inhibition of different zygotic genes (Driever et al., 1989; Struhl et al., 1989). In response to these maternal gradients, the segmentation gap genes are activated in spatially restricted but partially overlapping domains (Gaul and Jäckie, 1987, 1989; Tautz, 1988; Stanojevic et al., 1989; Pankratz et al., 1989; Hülskamp et al., 1990; Pignoni et al., 1990; Eldon and Pirrotta, 1991; Kraut and Levine, 1991a).

These domains, initially determined by the maternal morphogens as broad regions of expression, evolve into narrower and sharper stripes by the end of the syncytial blastoderm stage. The effects of mutations in other gap genes shows that this evolution is the result of crossregulatory interactions among the gap genes themselves (Jäckie et al., 1986; Gaul and Jäckie, 1987; Pankratz et al., 1989; Eldon and Pirrotta, 1991; Kraut and Levine, 1991a,b). For example, it has been suggested that the extent of the overlap between Krüppel (Kr) and knirps (kni) domains is due to their different sensitivity to the gradient of maternal hunchback (hb) protein (Hüls-kamp et al., 1990) but, in addition, the posterior limit of Kr is restricted by high levels of kni, since a detectable posterior shift occurs in kni mutant embryos (Jäckle et al., 1986; Gaul and Jäckle, 1987). Similarly, the kni domain expands posteriorly in giant (gt) mutants (Eldon and Pirrotta, 1991). Interactions also occur between non-adjacent gap genes. For example, although gt and Kr domains do not overlap detectably, gt expands towards the center of the embryo in the absence of Kr function (Eldon and Pirrotta, 1991; Kraut and Levine, 1991a). Multiple regulatory interactions can make it difficult to distinguish direct from indirect effects: the effect of Kr on gt might, for example, be mediated by knirps. This is ruled out by the fact that lack of kni function has only slight effects on gt.

The major gap genes have been cloned and their sequences reveal the presence of motifs characteristic of DNA-binding proteins. Kr, hb, kni and tailless (til) proteins contain zinc finger motifs (Rosenberg et al., 1986; Tautz et al., 1987; Nauber et al., 1988; Pignoni et al., 1990). Molecular experiments have shown that Kr, hb and kni interact directly with the DNA of target genes predicted by the genetic analysis (Treisman and Desplan, 1989; Pankratz et al., 1989,1990; Stanojevic et al., 1989). The most important of these target genes are the pair-rule genes. These are activated in a series of stripes with a double-segment periodicity in response to combinations of gap gene products and maternal factors. In the case of the two primary pair-rule genes, even-skipped (eve) and hairy, each stripe is controlled by a particular combination of gap and maternal gene products acting upon a distinct cri-regulatory element (Goto et al., 1989; Harding et al., 1989, Howard and Struhl, 1990; Riddihough and Ish-Horowicz, 1991; Small et al., 1991). Recent molecular evidence indicates that these elements are sensitive to the relative concentrations of gap gene products (Stanojevic et al., 1989; Pankratz et al., 1990). Therefore, the refinement of the borders of gap gene expression, achieved by their cross-regulatory interactions, assumes particular importance.

In this paper, we examine the properties of the giant gene product and its interactions with the Kr and kni genes, giant is required in the early embryo for the development of portions of the head and abdomen. At the extended germ band stage, gt amorphic mutants exhibit a fusion of the labial and prothoracic segments and the fusion of abdominal segments A5 to A7 (Petschek et al., 1987). The resulting cuticular pheno-type in the head comprises the absence of several components of the cephalopharyngeal skeleton and the failure of head involution (Gergen and Wieschaus, 1986; Mohler et al., 1989). In the abdomen, the anterior compartments of 5th to 7th and sometimes the 8th abdominal segments are deleted, resulting in an expanse of naked cuticle sometimes accompanied by appearance of ectopic filzkbrper in this region (Pets-chek and Mahowald, 1990). This is unlike the typical defects caused by gap gene mutations, in which contiguous segments are completely missing. Despite these peculiarly restricted defects, giant behaves like a typical gap gene. It is activated early and in a non-periodic pattern in response to purely maternal functions. It interacts with other gap genes in typical gap gene fashion (Eldon and Pirrotta, 1991; Kraut and Levine, 1991a,b) and it affects directly the expression of at least one early pair-rule gene (eve) (Small et al., 1991).

We show that giant encodes a DNA-binding protein that belongs to the b-ZIP family of transcription factors (Vinson et al., 1989) and is multiply phosphorylated in vivo. To identify targets of the giant product, we expressed giant in ectopic sites (see also similar experiments by Kraut and Levine, 1991b) and found that giant repressed Kr in the central domain and kni in the posterior domain, giant protein expressed in bacteria or in Drosophila embryos binds specifically to the Kr regulatory regions that direct expression in the central domain. Finally, we show that the apparent requirement of Kr for normal kni expression (Pankratz et al., 1990) is in fact the indirect consequence of the interaction between Kr and giant.

cDNA clones and expression constructs

To isolate giant cDNA clones, we used two embryonic cDNA libraries prepared with RNA extracted from 0-3 h and from 3 – 12 h embryos (a generous gift from L. Kauvar and T. Kornberg). These were screened with a hybridization probe prepared from a 1.6 kb BamHl genomic fragment containing part of the giant transcription unit. Clones ranging in length from 1.5 to 1.8 kb were obtained and sequenced after subcloning them in mpl8. The longest insert was close to the estimated length of the giant mRNA and was taken to be nearly full length. To construct the giant expression clone, the cDNA was trimmed at the 5, end with exo III and SI endonuclease, blunt ended and cloned in the EcoRI-A/wdlll sites of pUKK expression vector (Bickel and Pirrotta, 1990). The insert of the clone used for these experiments starts 6 bp upstream of the AUG translation start.

For antibody production, a 522 bp Smal-PvuH fragment that encoded amino acids 263–436 of the cDNA open reading frame with no opa sequences was inserted into the pEX-1 expression vector of Stanley and Luzio (1984), producing a lacZ-giant fusion gene. The hybrid protein was prepared and used to raise rabbit antibodies which were affinity purified as described by Eldon and Pirrotta (1991). The hs-gt transposon was constructed with a 1652 bp Mael-Nhel genomic fragment containing the entire coding region of giant but lacking the polyadenylation site. This fragment was ligated to the CaSpeR-hs vector cut with Hpal and Xbal. The CaSpeR-hs vector for germ line transformation was constructed by V. Pirrotta and C. Thummel and contains the hsp70 promoter followed by a multiple cloning site and the 3, end of the hsp70 gene, including the polyadenylation site. The transposon was injected into y >v67c23(2) embryos at a concentration of 400 ; μg/ml together with 80 μg/ml of the helper plasmid phsπ (Steller and Pirrotta, 1986). Three independent lines tested gave similar results. Krüppel genomic clones pER2 and AI3a were obtained from M. Hoch and H. Jäckie and are described in Fig. 7.

Embryonic nuclear extracts

Collections of 0– 12 h embryos with or without the hs-gt transposon were made at 25°C. The hs-gt embryos were heat shocked for 30 min at 37 °C and allowed to recover for 15 min prior to extraction. All the extraction steps were performed at 4°C and the procedure was adapted from Bickel and Pirrotta (1990). Dechorionated embryos were homogenized in 1 ml of sucrose buffer (0.35 M sucrose, 15 mM Hepes, 10 mM KC1, 0.1 mM EDTA, 0.5 mM EGTA, 1 mM DTT) containing protease inhibitors (20 μg/ml aprotinin, 1 mM benzamidine, 20 μ g/ml leupeptin, 10 μg/ml pepstatin A, 1 mM PMSF) using a 5 ml Dounce homogenizer. After filtering through glass wool, the homogenate was spun 2 min in a microfuge and the sedimented nuclei were resuspended in 100 μ l of lysis buffer (4 M urea, 40 mM Tris, 25 mM NaCl, 1 mM EDTA, 0.1% Triton X-100, 1 mM dithiothreitol) plus protease inhibitors. Care was taken to avoid resuspension of the yolk protein pellet and shearing of the genomic DNA. After 15– 30 min of rocking at 4°C, the nuclear lysate was spun at 100,000 g for 1 h to remove the DNA. The resulting supernatant was aliquoted and stored at —80°C. For the phosphatase treatment, 1 μl (5 μ g) of hs-gt nuclear extract was added to 40 μ l of l × OnePhorAll buffer (Pharmacia) containing protease inhibitors and incubated for 30 min at 3TC with or without 7 units of calf intestinal alkaline phosphatase or 5 mM sodium pyrophosphate inhibitor. The samples were then TCA precipitated prior to loading on a 10% acrylamide gel.

Bacterial extracts

Bacterial cultures containing the pUKK-gzani cDNA expression clone were grown in 100 ml L broth containing 100 μg/ml ampicillin at 37°C until OD600=0.8. Production oí giant protein was induced by addition of 1 mM IPTG and growth was continued for 3 h. The cells were harvested and resuspended in 1 ml of cold lysis buffer (40 mM Tris pH 7.6, 50 mM NaCl, 0.2 mM EDTA, 0.1% NP40,1 mM dithiothreitol, 20% glycerol) containing protease inhibitors (20 μg/ml aprotinin, 10 μg/ml leupeptin, 10 μg/ml pepstatin A, 1 mM PMSF). Lysozyme was added to a final concentration of 0.5 mg/ml and the extracts were sonicated until the DNA appeared completely sheared, then aliquoted and stored at —80°C. The total protein concentration was calculated using the Bradford assay (BioRad). The binding activity of this crude preparation appeares to be very temperature sensitive: no binding is observed at 25°C and the efficiency is greatly decreased after storage at —20°C.

In vitro translation

To produce giant protein in vitro, giant cDNA cloned in the Bluescript vector was transcribed with T7 RNA polymerase utilizing the Stratagene mRNA capping kit following the manufacturer’s instructions but without addition of capping nucleotide. The resulting RNA was translated using rabbit reticulocyte lysate (Promega).

Immunoblot analysis

Appropriate amounts of protein were adjusted to 4 M urea, icio mM Tris-HCl pH 7.6, 2% SDS, 5% μmercaptoethanol and 5% ficoll. The samples were loaded on a 10% acrylamide SDS gel and transferred to Immobilon-P membrane (Millipore) in 25 mM Tris, 192 mM glycine and 20% methanol, using a BioRad minigel apparatus. The membrane was blocked, incubated with 1 μg/ml affinity-purified anti-giant antibody and developed as described by Bickel and Pirrotta (1990).

Cuticle preparations and embryo staining

For the hs-gt experiment, y w67c23 host flies and hs-gt flies were allowed to lay eggs on grape juice plates for 30 min at 25°C. The plates were left at 25°C for additional 105 min and then floated on a waterbath at 37°C for 30 min. After aging for 24 h at 25°C, cuticles of the unhatched embryos were prepared according to Wieschaus and Ntisslein-Volhard (1986) and examined with dark-field optics. The index of mortality of the hs-gt embryos was twice that of the control (60% and 30%, respectively). About 60% of the hs-gt cuticles exhibited Kr-like defects, while the control presented abnormalities of various kinds but rarely resembling the Kr phenotype.

For antibody staining, control and hs-gt embryos were collected for 1.5 h, aged for 1.5 h and heat shocked as above. Dechorionation was started after 15 min of recovery (for giant staining) or after 30 min (for Krüppel and knirps staining). Fixation, staining and mounting were done as described by Eldon and Pirrotta (1991) using the Vectastain Elite ABC kit (Vector Laboratories). Anti-Kr antibodies were raised in rabbits by Bethyl Laboratories against full-length Kr protein expressed in bacteria from a clone kindly supplied by M. Levine. The antiserum was used without further purification at a 1:500 dilution after preabsorbtion with Canton S embryos. Anti-km, antibody (a gift from U. Naber) was used at a 1:100 dilution; anti-giant antibodies (Eldon and Pirrotta, 1991) and anti-/tz serum (a gift from H. Krause) at a 1:1,000 dilution, all without preabsorbtion. For double staining, embryos were first incubated with anti-knz antibody and reacted with the components of the Vectastain Elite ABC kit, then incubated with anti-gt and anti-Kr antibodies and the color developed with the Vectastain ABC-AP kit reagents (Vector Laboratories).

DNA binding and footprinting

The immunoprecipitation procedure was adapted from Benson and Pirrotta (1987). Each reaction was performed on ice in 25 μl of binding buffer (150 mM NaCl, 20 mM Tris-HCl pH 7.6, 0.25 mM EDTA, 1 mM dithiothreitol) containing 80 ng of digested and end-labelled DNA, 8 μg of calf thymus DNA (cold competitor), 75 μg of total bacterial extract and glycerol to a final concentration of 8%. After 1 h, 1 pl of anti-giant antiserum was added, followed 30 min later by 10 μ l Protein A-Sepharose beads (0.1 g swollen in 500 μl of binding buffer). After 30 more min, the beads were pelleted, washed and the DNA was eluted in 100 μl of footprinting stop solution (200 mM NaCl, 20 mM EDTA, 1% SDS, 40 μg/ml tRNA), extracted with phenol-chloroform, ethanol precipitated and analysed by electrophoresis on a 1.5% agarose gel or 4% acrylamide gel.

For footprinting, small binding fragments were subcloned in the Sma site of pUC18. The fragment (5 ng), labelled at one end with Klenow polymerase and 32P-dATP was immunoprecipitated as above, except that no calf thymus DNA was added. The material recovered was resuspended in 25 μl cold DNAase buffer (150 mM NaCl, 10 mM MgCl2, 5 mM CaCl2), 50 μg more extract were added and DNAase digestion was initiated by addition of 25 μl of 6 μg/ml DNAase in the same buffer. After 5 min of incubation on ice, the digestion was stopped with 50 μl of stop solution (see above) followed by phenol-chloroform extraction and ethanol precipitation. An approximately equivalent amount of DNA was digested in a parallel reaction to give the DNA ladder. The DNA samples were analysed on a 6% acrylamide, 8 M urea gel flanked by sequencing reactions to identify the protected sequences.

Sequence of the giant cDNA

We used a 1.6 kb BamHI genomic fragment that had previously been shown to hybridize to the giant mRNA (Mohler et al., 1989) to screen two cDNA libraries constructed from 0–3 h and from 3–12 h embryos (a gift from L. Kauvar and T. Kornberg). Most of the clones obtained contained inserts around 1.5 kb and only one, from the 0–3 h embryonic library, was 1.8 kb, the estimated size of the giant mRNA. Restriction analysis indicated that all inserts were related and sequencing of the 1.8 kb and of two of the shorter cDNAs confirmed that they represented the same transcript. The sequence of the long cDNA is shown in Fig. 1A together with the deduced translation product of the longest open reading frame. The first AUGs in this reading frame occur at nucleotides 113 and 119. The first of these is more likely to be the correct initiation codon since the sequence preceding it is in good agreement with the Drosophila consensus for translation start sites (Cavener, 1987) while the second, third or fourth AUG codons in this open reading frame represent very poor matches. If our interpretation is correct, the coding region of the giant gene gives rise to a protein of 448 amino acids. Comparison with the genomic sequence (not shown) reveals a TATA box and a good transcription start consensus in the sequence just preceding the beginning of the cDNA, supporting the interpretation that this clone represents a nearly full-length cDNA. However, we have not mapped the precise transcription start site. The genomic sequence reveals also the presence of a small intron of 75 nucleotides at position 164 in the sequence shown in Fig. 1.

Fig. 1.

giant cDNA and amino acid sequences. (A) Nucleotide sequence of the sense strand of the longest cDNA clone (1798 bp) and amino acid sequence of the longest ORF encoded (418 amino acids). Transcription is predicted to start a few nucleotides upstream of this sequence, based on genomic sequencing. The position of a small intron is indicated by an arrow. A polyadenylation signal precedes the poly(A) tail. In the amino acid sequence, the basic region of the predicted DNA-binding domain is boxed and the residues identifying the heptad repeats of the leucine zipper are circled. Two runs of glutamines and one of alanines are present in the first half of the protein. (B) Comparison of the C-terminal amino acid sequence of giant with the DNA-binding domains of typical b-ZIP proteins (Vinson et al., 1989). The most conserved residues in the basic region are boxed, as are the leucine positions of the leucine zipper. The nucleotide sequence will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under accession number ×61148.

Fig. 1.

giant cDNA and amino acid sequences. (A) Nucleotide sequence of the sense strand of the longest cDNA clone (1798 bp) and amino acid sequence of the longest ORF encoded (418 amino acids). Transcription is predicted to start a few nucleotides upstream of this sequence, based on genomic sequencing. The position of a small intron is indicated by an arrow. A polyadenylation signal precedes the poly(A) tail. In the amino acid sequence, the basic region of the predicted DNA-binding domain is boxed and the residues identifying the heptad repeats of the leucine zipper are circled. Two runs of glutamines and one of alanines are present in the first half of the protein. (B) Comparison of the C-terminal amino acid sequence of giant with the DNA-binding domains of typical b-ZIP proteins (Vinson et al., 1989). The most conserved residues in the basic region are boxed, as are the leucine positions of the leucine zipper. The nucleotide sequence will appear in the EMBL, GenBank and DDBJ Nucleotide Sequence Databases under accession number ×61148.

The predicted amino acid sequence of the giant protein contains runs of repeated amino acids frequently found in Drosophila coding regions. These repetitive regions include a run of alanines (12 out of 13) and two runs of glutamines (16 out of 18 residues and 17 out of 19 residues, respectively) that correspond to the OPA repeat (Wharton et al., 1985). The C-terminal region of the protein contains sequence motifs characteristic of the leucine zipper or “scissors grip” class of DNA-binding proteins (Vinson et al., 1989). These consist of a region rich in basic residues followed by a series of heptad repeats with leucines occupying preferentially the first position of the heptad. Fig. 1B shows that these motifs, which have been shown to be necessary for homo- or heterodimer formation and DNA-binding activity, are highly conserved in the giant sequence and include all the features identified by Vinson et al. (1989). These structural motifs strongly suggest that giant is a DNA-binding protein that regulates the expression of other segmentation genes. This is consistent with its nuclear localization and with the effects of giant mutations on the expression of other gap genes (Eldon and Pirrotta, 1991; Kraut and Levine, 1991a,b; Small et al., 1991).

In vivo modification of the giant protein

To raise antibodies against the giant protein, we used a 522 bp Smal-PvuYl fragment encoding amino acids 263-436 and avoiding the repetitive regions. This fragment was cloned in the pEX-1 expression vector of Stanley and Luzio (1984) to produce gzarií-β-galactosidase fusion protein for injection into rabbits. Antibodies raised against this fusion protein have been affinity purified and used to follow the pattern of expression of the giant product during embryogenesis (Eldon and Pirrotta, 1991).

To express full-length giant protein in bacteria, the cDNA fragment was trimmed at the 5, end to leave only 6 nucleotides before the initiator ATG, and was cloned in the pUKK expression vector (Bickel and Pirrotta, 1990). In this system, full-length giant protein is expressed under the control of the tac promoter. Extracts produced by the bacteria after induction were analysed on western blots using the antibodies raised against the gzant-β-galactosidase fusion protein. Fig. 2 compares the proteins detected by these antibodies in the bacterial extracts with those produced by the endogenous giant gene in early embryos or by a hsp70-giant cDNA (hs-gt) construct introduced into flies (see below). The different extracts contain immunoreactive proteins of different apparent relative molecular masses, none of them corresponding to the 49.18 × 103 predicted by the cDNA sequence. Abnormal mobility in SDS-acrylamide gels has been observed with other proteins (see for example Benson and Pirrotta, 1987) and may account for part of the discrepancy. An important factor in the apparent difference between bacterially produced and endogenous giant may be proteolysis. The giant protein extracted from bacteria or embryos is very sensitive to proteolysis and its isolation requires rapid extraction and the use of protease inhibitors, giant protein expressed in two different E. coli strains shows different degrees of breakdown, generally greater than that shown in Fig. 2. Similar results for the bacterial protein have been reported by Kraut and Levine (1991a).

Fig. 2.

Western blot of giant protein expressed in bacteria and in embryos. (A) (Lane 1) 10 μg of total protein from bacteria carrying the pUKK vector after IPTG induction. The only immunoreactive band is β-galactosidase, due to residual anti-βgal antibodies. (Lane 2) 10 μg of protein from bacteria expressing giant protein from the pUKK-gi cDNA plasmid after IPTG induction. Despite the use of protease inhibitors in the preparation, several degradation products are observed. The highest band has an apparent relative molecular mass of 60×103 (predicted giant relative molecular mass 49.2 × 103). (Lane 3) Nuclear extract of 0–1.5 h Canton S embryos (50 μg protein), giant protein is not detectable at this stage. (Lane 4) Nuclear extract of 2–4 h Canton S embryos (50 μg protein). (Lane 5) Nuclear extract of heat-shocked hs-gt embryos (5 μg protein). The hs-gt transposon produces giant protein with the same mobility as the endogenous one. The slight difference in migration is due to overloading of lane 4. (B) This figure is a composite of the autoradiogram (lanes 1 and 2) and the western blot (lanes 3–6) of preparations run on the same gel. Lanes 1 and 2 are in vitro translation products labelled with [35S]methionine in the absence (lane 1) or presence (lane 2) of gt RNA. The in vitro translated product migrates like the bacterial gt. (Lane 3) 5 μg protein from bacteria carrying the pLJKK-gt cDNA plasmid. (Lane 4) Nuclear extract of hs-gt embryos (5 μg protein) treated with calf intestinal alkaline phosphatase. The ladder of bands indicates the presence of multiple phosphorylation sites. With sufficient phosphatase, all the gt shifts to a band with the same mobility as the in vitro translated product. (Lane 5) Same as in lane 4 but in the presence of pyrophosphate to inhibit the phosphatase. (Lane 6) Untreated hs-gt extract.

Fig. 2.

Western blot of giant protein expressed in bacteria and in embryos. (A) (Lane 1) 10 μg of total protein from bacteria carrying the pUKK vector after IPTG induction. The only immunoreactive band is β-galactosidase, due to residual anti-βgal antibodies. (Lane 2) 10 μg of protein from bacteria expressing giant protein from the pUKK-gi cDNA plasmid after IPTG induction. Despite the use of protease inhibitors in the preparation, several degradation products are observed. The highest band has an apparent relative molecular mass of 60×103 (predicted giant relative molecular mass 49.2 × 103). (Lane 3) Nuclear extract of 0–1.5 h Canton S embryos (50 μg protein), giant protein is not detectable at this stage. (Lane 4) Nuclear extract of 2–4 h Canton S embryos (50 μg protein). (Lane 5) Nuclear extract of heat-shocked hs-gt embryos (5 μg protein). The hs-gt transposon produces giant protein with the same mobility as the endogenous one. The slight difference in migration is due to overloading of lane 4. (B) This figure is a composite of the autoradiogram (lanes 1 and 2) and the western blot (lanes 3–6) of preparations run on the same gel. Lanes 1 and 2 are in vitro translation products labelled with [35S]methionine in the absence (lane 1) or presence (lane 2) of gt RNA. The in vitro translated product migrates like the bacterial gt. (Lane 3) 5 μg protein from bacteria carrying the pLJKK-gt cDNA plasmid. (Lane 4) Nuclear extract of hs-gt embryos (5 μg protein) treated with calf intestinal alkaline phosphatase. The ladder of bands indicates the presence of multiple phosphorylation sites. With sufficient phosphatase, all the gt shifts to a band with the same mobility as the in vitro translated product. (Lane 5) Same as in lane 4 but in the presence of pyrophosphate to inhibit the phosphatase. (Lane 6) Untreated hs-gt extract.

However, protein synthesized in vitro with a reticulocyte lysate has the same mobility as the major band detected in bacterial extracts, suggesting that this is the correct mobility of the full-length protein and that post-translational modifications in the fly convert it to a slower migrating form. Treatment of the protein extracted from hs-gt flies with alkaline phosphatase converts it to a series of faster migrating forms converging, if sufficient phosphatase is used, to a form with the same mobility as the bacterially or in vitro synthesized form. This is not due to proteolytic activity because the mobility does not continue to increase and because no change is observed if the phosphatase is inhibited by pyrophosphate. We interpret this to mean that the giant protein in vivo is phosphorylated at multiple sites.

Effects of ectopic expression of giant in embryos

Studies of the reciprocal effects of segmentation gap genes and giant have established that giant interacts with other gap genes, some of which may be direct targets of giant control. To help identify these targets, we expressed the giant gene ectopically. We constructed a transposon containing a 1.6 kb genomic Mae\ fragment that includes the entire giant coding region and 63 bp of leader sequence, placed under the control of the hsp70 promoter (the hs-gt gene). Embryos produced by flies homozygous for this transposon were aged approximately 2 h at 25°C and then heat shocked for 30 min. Such embryos showed a high index of mortality compared to heat-shocked embryos that did not carry the transposon. At the end of embryonic development, the majority of the embryos that overexpressed giant showed a range of cuticular abnormalities that resembled the pattern deletions observed with an allelic series of Krüppel mutations (Wieschaus et al., 1984). Embryos homozygous for weak Kr alleles lack T2, T3, A2 and A4, indicating that the region most sensitive to the lack of Kr product is discontinuous. In the majority of heat-shocked hs-gt embryos, T2 and, less often, T3 are missing and abdominal segments A2 to A4 or A5 are partially or entirely deleted (Fig. 3C,D). In strong Kr alleles, the thoracic segments and abdominal denticle belts Al to A5 are absent and are replaced by an inverted denticle belt that has been interpreted as a mirror image of A6. In our hs-gt embryos, this strong phenotype was observed only occasionally. Similar experiments and similar observations have been reported by Kraut and Levine (1991b). The similarity of the Kr phenotype to that caused by ectopic gt expression suggested that gt might directly repress Kr expression. In a normal embryo, the domains of gt and Kr expression are adjacent and do not overlap detectably. However, in the absence of Kr, the gt domain expands anteriorly, implying that Kr inhibits gt. The complementary effect is not observed: in a gt mutant, Kr expression does not expand appreciably (Eldon and Pirrotta, 1991).

Fig. 3.

Ectopic giant expression at cellular blastoderm causes Kr-like cuticular defects. (A) Wild-type cuticle. (B)Kr1 homozygote exhibiting a severe Kr amorphic phenotype (Wieschaus et al., 1984). The 3 thoracic and the first five abdominal denticle belts are missing and are replaced by a mirror image duplication of a posterior one. The head fails to involute. (C-E) Cuticle preparations of embryos carrying the hs-gt construct heat shocked at approximately 2 h of embryonic development. The prothoracic denticle belt is always present while T2 and T3 are either missing or incomplete. The abdominal gap comprises segments A2-A4 (C,D) or A2-A5 (E). Fusion of denticle belts is sometimes observed (C). These phenotypes are typical of Kr hypomorphic alleles (Wieschaus et al., 1984).

Fig. 3.

Ectopic giant expression at cellular blastoderm causes Kr-like cuticular defects. (A) Wild-type cuticle. (B)Kr1 homozygote exhibiting a severe Kr amorphic phenotype (Wieschaus et al., 1984). The 3 thoracic and the first five abdominal denticle belts are missing and are replaced by a mirror image duplication of a posterior one. The head fails to involute. (C-E) Cuticle preparations of embryos carrying the hs-gt construct heat shocked at approximately 2 h of embryonic development. The prothoracic denticle belt is always present while T2 and T3 are either missing or incomplete. The abdominal gap comprises segments A2-A4 (C,D) or A2-A5 (E). Fusion of denticle belts is sometimes observed (C). These phenotypes are typical of Kr hypomorphic alleles (Wieschaus et al., 1984).

To determine whether ectopic expression of gt has a negative effect on the expression of Kr, we examined the distribution of gt, Kr and other segmentation gene products in hs-gt embryos after heat shock. At the earliest stages, the endogenous giant expression pattern (Fig. 4A), consists of a posterior domain 0-35% egg length (EL) and an anterior domain 62-80% EL (Eldon and Pirrotta, 1991; Kraut and Levine, 1991b). In the heat-shocked embryos, a uniform ectopic expression is superimposed upon the normal pattern. Embryos heat shocked during cellularization showed a nearly uniform dark staining affecting all cells except the pole cells, which are apparently resistent to heat-shock activation (Fig. 4D). Detection of the ectopic gt expression required fixing the embryos within 25 min of the heat shock treatment. With longer recovery periods, the intensity of the ectopic staining decreased rapidly, suggesting a very high turnover rate. This was more dramatic with younger embryos (fixed towards the end of nuclear cycle 14) than with embryos fixed at the beginning of gastrulation. A rapid turnover of both protein and mRNA has been observed, and is in fact necessary, for other segmentation genes (Edgar et al., 1986; Kellerman et al., 1990). Rapid turnover is consistent with the high susceptibility of the gt protein to proteolysis both in Drosophila and in bacterial extracts.

Fig. 4.

Ectopic giant causes a suppression of Kr and kni expression in the abdominal region. All embryos are oriented with the anterior end to the left and the dorsal side up. Control untransformed embryos (A,C,E,G) and embryos carrying the hs-gt transposon (B,D,F,H) were treated in parallel. (A,B) Syncytial blastoderm embryos fixed after 25 min of recovery from a 30 min heat shock and stained with anti-giant antibody. (A) The normal giant pattern extends from 62% to 80% EL in the head region and from 0% to 35% EL in the abdominal region. (B) In hs-gt embryos, a uniform level of giant is induced and is superimposed upon the endogenous pattern. The arrow points to the pole cells which do not stain. (C,D) Embryos heat shocked slightly later have reached the end of cellular blastoderm. (C) Control embryos show stripes 2, 3 and 4 of giant. (D) Induced hs-gt embryos show uniform expression of giant. (E,F) Detection of Krüppel protein in late cellular blastoderm embryos treated as above but fixed 40 min after the heat shock. (F) In the hs-gt embryo expression of Kr in the central domain is dramatically reduced compared with the posterior domain which is unaffected. (G,H) Detection of kni protein in early gastrulation embryos heat shocked and fixed as in E,F. At this stage (late cellular blastoderm) kni is expressed in an anterior ventral band and a posterior stripe (G). Ectopic giant inhibits kni expression in the posterior domain but does not affect the anterior domain (H).

Fig. 4.

Ectopic giant causes a suppression of Kr and kni expression in the abdominal region. All embryos are oriented with the anterior end to the left and the dorsal side up. Control untransformed embryos (A,C,E,G) and embryos carrying the hs-gt transposon (B,D,F,H) were treated in parallel. (A,B) Syncytial blastoderm embryos fixed after 25 min of recovery from a 30 min heat shock and stained with anti-giant antibody. (A) The normal giant pattern extends from 62% to 80% EL in the head region and from 0% to 35% EL in the abdominal region. (B) In hs-gt embryos, a uniform level of giant is induced and is superimposed upon the endogenous pattern. The arrow points to the pole cells which do not stain. (C,D) Embryos heat shocked slightly later have reached the end of cellular blastoderm. (C) Control embryos show stripes 2, 3 and 4 of giant. (D) Induced hs-gt embryos show uniform expression of giant. (E,F) Detection of Krüppel protein in late cellular blastoderm embryos treated as above but fixed 40 min after the heat shock. (F) In the hs-gt embryo expression of Kr in the central domain is dramatically reduced compared with the posterior domain which is unaffected. (G,H) Detection of kni protein in early gastrulation embryos heat shocked and fixed as in E,F. At this stage (late cellular blastoderm) kni is expressed in an anterior ventral band and a posterior stripe (G). Ectopic giant inhibits kni expression in the posterior domain but does not affect the anterior domain (H).

The heat-shocked hs-gt embryos exhibit a drastic decrease in the level of Kr protein in the central domain of Kr expression (Fig. 4F). The suppression of Kr expression is variable, ranging from very strong (Kr band virtually disappears) to undetectable (Kr band normal) in some embryos. The variability is consistent with the phenotypic effects observed in the cuticle preparations and is probably explained by the narrow time window for repression and the competing processes involved. The posterior domain of Kr expression, which appears after cellular blastoderm and is not affected by gt, provides an internal control. Embryos lacking the hs-gt transposon and heat shocked in parallel never show a decrease in the expression of Kr in the central domain relative to the posterior domain.

Previous studies indicated that gt sets a posterior limit to the posterior domain of knirps (Eldon and Pirrotta, 1991; Kraut and Levine, 1991a). It is not surprising therefore that overexpression of gt also strongly inhibits the expression of kni in its posterior domain. In control embryos, kni posterior expression is always strong before the pole cells begin to migrate and begins to decrease only after germ band extension is visibly under way. In the heat-shocked hs-gt embryos, the posterior kni stripe becomes very weak well before the first signs of pole cell migration, while the anterior kni expression is not affected (Fig. 4G,H). In contrast, the expression of the gap gene hunchback is not appreciably affected, either in its anterior or posterior domain (not shown).

The consequences of ectopic expression of gt were also monitored by staining with anti-fushi tarazu (ftz) antibodies. While normal embryos, with or without heat shock, always displayed the normal seven stripe pattern, the hs-gt embryos after heat shock showed a reduction in two or more of the stripes. Stripes 4 and 5 are the most sensitive and virtually disappear in many embryos (Fig. 5B). Stripe 2 is less sensitive but in some cases it too becomes weaker, narrower and nearly fuses with stripe 3 (Fig. 5C). This pattern approaches closely that observed in Kr mutants, where only four ftz stripes are produced, with faint residual expression occasionally detected between the first and the second stripe (Carroll and Scott, 1986). The effect of kni mutations on the stripe pattern offtz is very different: the first two stripes appear normal while stripes 3-6 are replaced by a single broad stripe. It is possible therefore that the major phenotypic effect of ectopic gt expression is due to the suppression of Kr.

Fig. 5.

The induction of ectopic giant results in an altered ftz pattern similar to that seen in Kr mutants. Embryos were treated as in Fig. 4E,H and stained with anti-ftz antibodies. (A) Kr1 homozygous embryos lack ftz stripes 2,4 and 5 with the remnants of stripe 2 sometimes still visible (Carroll and Scott, 1986). (B) After heat shock, the majority of the hs-gt embryos lose partially or totally stripes 4 and 5. This altered pattern is correlated with the loss of segments A2 through A5 found in the majority of embryos. (C) In some cases, stripe 2 fuses with stripe 3, leaving a 4-stripe pattern very similar to that of Kr mutants.

Fig. 5.

The induction of ectopic giant results in an altered ftz pattern similar to that seen in Kr mutants. Embryos were treated as in Fig. 4E,H and stained with anti-ftz antibodies. (A) Kr1 homozygous embryos lack ftz stripes 2,4 and 5 with the remnants of stripe 2 sometimes still visible (Carroll and Scott, 1986). (B) After heat shock, the majority of the hs-gt embryos lose partially or totally stripes 4 and 5. This altered pattern is correlated with the loss of segments A2 through A5 found in the majority of embryos. (C) In some cases, stripe 2 fuses with stripe 3, leaving a 4-stripe pattern very similar to that of Kr mutants.

giant and Krüppel double mutants

The simplest interpretation for the reduced kni expression in hs-gt embryos is that ectopic expression of gt suppresses kni in the posterior domain. However, Pankratz et al. (1990) have shown that the level of knirps RNA in the posterior region is greatly reduced in a Kr mutant background and suggested that Kr product is required to enhance kni expression. Therefore, the repression of kni by hs-gt might be indirect and caused by a primary inhibitory effect on Kr expression. To distinguish between these alternatives, we examined the kni pattern in Kr, gt and in Kr-gt double mutants. Fig. 6 shows four embryos produced by crossing Df(l)gf 62gl8 ; Kr double heterozygote females with males heterozygous for Kr1. Of the progeny, 3/8 should lack either gt function or Kr function and 1/16 should lack both. Since both mutations prevent synthesis of the gene product, we were able to identify the genotypes by staining with anti-gtzznr and anti-Krüppel antibodies. Fig. 6A shows a wild-type embryo with the normal pattern of giant (blue), Krüppel (blue) and knirps (brown) expression. At this stage, giant stripe 1 has not yet appeared, stripes 2 and 3 are anterior to the Krüppel centra) domain and stripe 4 is immediately posterior to knirps. In the absence of gt function (Fig. 6B), the kni stripe expands posteriorly but Kr does not appear to be significantly affected, as prevously noted by Eldon and Pirrotta (1991). In the absence of Kr (Fig. 6C), gt stripe 4 expands anteriorly, covering entirely the domain in which kni is normally expressed. The kni stripe is greatly reduced in intensity, as observed by Pankratz et al. (1990). When both gt and Kr are absent, we would expect a low level of kni, if Kr is required to enhance kni expression. Alternatively, we would expect a normal level of kni expression if the role of Kr is simply to prevent the expansion of gt. Fig. 6D shows that in embryos lacking both Kr and gt, kni posterior expression is normal in intensity, as determined by comparison with the intensity of the anterior domain, and as broad as in the gt mutant shown in Fig. 6B. We conclude that Kr is not directly required for the normal activation of kni.

Fig. 6.

knirps expression in gt; Kr double mutants. Embryos derive from a cross between females Df(l)62g18/+; Kr1/+ and males Kr1/+. Both of these mutations are deficiencies and produce no protein. All embryos were stained with anti-knirps antibody (brown) and then with anti-giant (blue) and antx-Krilppel (blue) antibodies. Absence of staining for giant (B), Krüppel (C) or both (D) indicates that the embryo is mutant for either gene or for both. Embryos are oriented as in Fig. 4 and are of similar age (early cellular blastoderm). (A) In the wild-type embryo, the posterior kni domain lies between the Kr central domain and gt stripe 4 (barely visible in this embryo) at approximately .34 – 47% EL.. Its intensity is equal to or higher than the anterior expression. (B) Embryo hemizygous for the gt deficiency. The posterior kni stripe expands posteriorly (by about 6%EL) as previously observed by Eldon and Pirrotta (1991). Its intensity is the same as in the wild type. (C) Kr1 homozygous embryo. Posterior knirps expression is abolished or strongly reduced (see also Pankratz et al., 1989). (D) gt, Kr double mutant embryo. In the absence of both gt and Kr proteins, posterior knirps is expressed at normal level and in the same domain as in the gt embryo (B). This indicates that kni expression does not require Kr, but is instead repressed by gt.

Fig. 6.

knirps expression in gt; Kr double mutants. Embryos derive from a cross between females Df(l)62g18/+; Kr1/+ and males Kr1/+. Both of these mutations are deficiencies and produce no protein. All embryos were stained with anti-knirps antibody (brown) and then with anti-giant (blue) and antx-Krilppel (blue) antibodies. Absence of staining for giant (B), Krüppel (C) or both (D) indicates that the embryo is mutant for either gene or for both. Embryos are oriented as in Fig. 4 and are of similar age (early cellular blastoderm). (A) In the wild-type embryo, the posterior kni domain lies between the Kr central domain and gt stripe 4 (barely visible in this embryo) at approximately .34 – 47% EL.. Its intensity is equal to or higher than the anterior expression. (B) Embryo hemizygous for the gt deficiency. The posterior kni stripe expands posteriorly (by about 6%EL) as previously observed by Eldon and Pirrotta (1991). Its intensity is the same as in the wild type. (C) Kr1 homozygous embryo. Posterior knirps expression is abolished or strongly reduced (see also Pankratz et al., 1989). (D) gt, Kr double mutant embryo. In the absence of both gt and Kr proteins, posterior knirps is expressed at normal level and in the same domain as in the gt embryo (B). This indicates that kni expression does not require Kr, but is instead repressed by gt.

Binding of giant to DNA

The predicted sequence of the giant protein shows that it contains a typical scissors-grip, leucine zipper DNA-binding motif. The effects of gt ectopic expression in the embryo and the evidence from double mutants imply that it acts in vivo as a repressor of Kr and kni. Therefore, Kr DNA should be a good target to study gt binding to DNA in vitro. We used immunoprecipitation to analyse Kr genomic clones including 17 kb of upstream sequences that contain several independent eft-regulatory elements (Hoch et al., 1990). Fig. 7A shows that gt protein expressed in E. coli co-immuno-precipitates with two fragments in the 5, flanking region of Kr: a 1.4 kb EcoRI-Sall fragment and a 1.1 kb EcoRI fragment containing respectively the CD2 and the CD1 control elements of Kr (Hoch et al., 1990). Each of these two elements has been shown to drive the expression of a reporter gene in the central domain of Kr expression. Fig. 7B shows binding to another subclone of the Kr locus that includes the coding and 3, flanking regions. In this experiment, gt binds, as expected, to a 3.3 kb ffindlll fragment containing the CD2 element but also to a 1.5 kb Hzndlll fragment from the 3, flanking region whose regulatory significance is unknown. The binding sequences were further localized to a 450 bp HinB fragment from the CD2 region (Fig. 7C) and to a 290 bp Hinfl-Hindlll fragment from CD1 (not shown). These results confirm that giant is a DNA-binding protein that interacts specifically with the two regulatory regions that are important for Kr expression in the domain affected by ectopic gt.

Fig. 7.

giant protein binds to Kr regulatory DNA. Each panel shows immunoprecipitation reactions using bacterial extract lacking gt or containing gt. The left lane in each case represents the input DNA. Arrowheads point to the fragments specifically bound by giant protein. (A) Immunoprecipitation of clone pER2 digested with BurnEtt, EcoRI and Sad. giant binds to a 1.4 kb EcoRl-SaZI fragment containing almost the whole CD2 element and a 1.1 kb EcoRI fragment containing part of CD1 and of CD2. (B) Immunoprecipitation of clone AI3a digested with ffindllL giant binds to a 3.3 kb fragment containing part of CD2 and a 1.45 kb fragment from the 3, region. The 4.5 kb band binding non-specifically is vector plus 500 bp of insert. (C) The 3.3 kb Hi,ndlll fragment containing the CD2-binding site was subcloned in pUC18 and digested with Xbal and Hinü. The binding site is contained in a 480 bp Hinü fragment. (D) Comparison between gt binding to Kr CD2 and to eve stripe 2 binding site and between the bacterial (75 μ g) and embryonic (13 μg) gt extracts. The input DNA contains approximately equimolar amounts of the two DNAs, mixed prior to labelling. The eve DNA contains five tandem repeats of a 54 bp fragment containing the gt binding site to the eve stripe 2 element located at -1110 (Small et al., 1991). Note the difference in recovery between the Kr and eve DNA fragments. The map below shows the 25 kb surrounding the Kr gene with the relevant restriction sites indicated by S, SalI; E, EcoRI; B, BamHI; H, HindIII. The arrow above represents the Kr transcription unit. The arrows below point to the fragments that are specifically immunoprecipitated by gt. Of the upstream regulatory elements, only CD1 and CD2 are shown (Hoch et al., 1990), with black boxes indicating the approximate positions of the gt binding sites.

Fig. 7.

giant protein binds to Kr regulatory DNA. Each panel shows immunoprecipitation reactions using bacterial extract lacking gt or containing gt. The left lane in each case represents the input DNA. Arrowheads point to the fragments specifically bound by giant protein. (A) Immunoprecipitation of clone pER2 digested with BurnEtt, EcoRI and Sad. giant binds to a 1.4 kb EcoRl-SaZI fragment containing almost the whole CD2 element and a 1.1 kb EcoRI fragment containing part of CD1 and of CD2. (B) Immunoprecipitation of clone AI3a digested with ffindllL giant binds to a 3.3 kb fragment containing part of CD2 and a 1.45 kb fragment from the 3, region. The 4.5 kb band binding non-specifically is vector plus 500 bp of insert. (C) The 3.3 kb Hi,ndlll fragment containing the CD2-binding site was subcloned in pUC18 and digested with Xbal and Hinü. The binding site is contained in a 480 bp Hinü fragment. (D) Comparison between gt binding to Kr CD2 and to eve stripe 2 binding site and between the bacterial (75 μ g) and embryonic (13 μg) gt extracts. The input DNA contains approximately equimolar amounts of the two DNAs, mixed prior to labelling. The eve DNA contains five tandem repeats of a 54 bp fragment containing the gt binding site to the eve stripe 2 element located at -1110 (Small et al., 1991). Note the difference in recovery between the Kr and eve DNA fragments. The map below shows the 25 kb surrounding the Kr gene with the relevant restriction sites indicated by S, SalI; E, EcoRI; B, BamHI; H, HindIII. The arrow above represents the Kr transcription unit. The arrows below point to the fragments that are specifically immunoprecipitated by gt. Of the upstream regulatory elements, only CD1 and CD2 are shown (Hoch et al., 1990), with black boxes indicating the approximate positions of the gt binding sites.

To obtain footprints on the CD1 or CD2 fragments, we isolated the protein-DNA complex by immuno-precipitation before digesting with DNAase I (Fig. 8). The two sequences protected by the gt protein have a general similarity but are not identical. Both contain a dyad axis of symmetry, passing between two nucleotides and centered in a core sequence ACGT common to the binding sites of many other proteins of the leucine zipper family (Fig. 9). Three other giant binding sites have been footprinted by Small et al. (1991) in the stripe 2 regulatory element of the pair-rule gene evenskipped. Their sequences are difficult to compare with ours because their footprints are extraordinarily broad, suggesting that they contain more than one binding site. By comparison with the CD1 or CD2 sites, the eve stripe 2 sites have much lower affinity for giant protein. In Fig. 7D, the binding of a fragment containing five tandem copies of eve site – 1110 is compared to the binding of CD2 DNA. In spite of the five-fold repeat, the eve site binds far less strongly, suggesting that its sequence is farther removed from an optimal consensus recognition sequence.

Fig. 8.

giant footprints in the Kr CD1 and CD2 elements. DNAase I footprinting of the Wirtfl-Wmdni CD1 fragment and Hinfl CD2 fragment (290 bp and 480 bp, respectively). The fragments were DNAase treated (left lanes)‘or immunoprecipitated with pUKK-gt cDNA extract prior to DNAase digestion. The brackets span the sequences protected, which are shown below. Asterisks indicate the matching nucleotides.

Fig. 8.

giant footprints in the Kr CD1 and CD2 elements. DNAase I footprinting of the Wirtfl-Wmdni CD1 fragment and Hinfl CD2 fragment (290 bp and 480 bp, respectively). The fragments were DNAase treated (left lanes)‘or immunoprecipitated with pUKK-gt cDNA extract prior to DNAase digestion. The brackets span the sequences protected, which are shown below. Asterisks indicate the matching nucleotides.

Fig. 9.

Comparison of binding sites of b-ZIP proteins. Nucleotide sequences recognized by b-ZIP proteins are grouped into those that resemble the CRE-binding site, the C/EBP-binding site and those that resemble the jun-AP-1 binding site. The box encloses the central core nucleotides and a conserved half site is underlined. The giant binding sites in the Kr CD1 and CD2 domains resemble most the CRE recognition sequence. Sequences of the giant binding sites in the eve stripe 2 regulatory domain were arbitrarily selected from very broad footprints (Small et al., 1991) to maximize the fit with the others. The sequences are taken from Lin and Green (1988): CREB-ATF; Lohmer et al. (1991): opaque-2-, T. Abel and T. Maniatis (personal communication): BBF2, a b-ZIP protein that binds to the enhancer of the Drosophila Adh gene; Small et al. (1991): giant binding sites in eve stripe 2 element; Johnson et al. (1987): C/EBP; Angel et al. (1987): AP-1; Hill et al. (1986): GCN4.

Fig. 9.

Comparison of binding sites of b-ZIP proteins. Nucleotide sequences recognized by b-ZIP proteins are grouped into those that resemble the CRE-binding site, the C/EBP-binding site and those that resemble the jun-AP-1 binding site. The box encloses the central core nucleotides and a conserved half site is underlined. The giant binding sites in the Kr CD1 and CD2 domains resemble most the CRE recognition sequence. Sequences of the giant binding sites in the eve stripe 2 regulatory domain were arbitrarily selected from very broad footprints (Small et al., 1991) to maximize the fit with the others. The sequences are taken from Lin and Green (1988): CREB-ATF; Lohmer et al. (1991): opaque-2-, T. Abel and T. Maniatis (personal communication): BBF2, a b-ZIP protein that binds to the enhancer of the Drosophila Adh gene; Small et al. (1991): giant binding sites in eve stripe 2 element; Johnson et al. (1987): C/EBP; Angel et al. (1987): AP-1; Hill et al. (1986): GCN4.

This experiment also suggests that the protein isolated from hs-gt embryos binds with the same specificity but possibly more efficiently than the bacterial protein. Although present at a concentration apparently tenfold lower in the fly extract, the fly protein binds nearly as much DNA as the bacterial protein under conditions far from saturation. These results suggest that the phosphorylated Drosophila protein may have a higher specific binding affinity.

b-ZIP DNA-binding proteins

The giant gene product is a DNA-binding protein belonging to the family of the scissors grip leucine zipper (b-ZIP) type, along with jun, fos, CREB and others (Vinson et al., 1989). This family of proteins shares a structural motif consisting of a basic region with a set of highly conserved residues and an a-helical leucine zipper consisting of four or five heptad repeats with leucine as the preferred residue in the first position of the heptad. This structural motif has been shown to be responsible for the formation of homo- or hetero-dimers, mediated by the pairing of the helical leucine zippers (Kouzarides and Ziff, 1989; O’Shea et al., 1989). These helical regions, in a parallel coiled-coil configuration, form the stem of a fork which trestles on the DNA double helix and brings the basic regions of the two monomers into symmetric contacts with the bases in the major groove of the DNA. Secondary structure calculations using the method of Gamier et al. (1978) strongly predict that the C-terminal region of the giant protein, from residue 385 to 447 is likely to assume an a-helical configuration. The probability of a turn rises at positions 392 and 407, consistent with a bend between the leucine zipper and basic domain that, according to the model of Vinson et al. (1989), should wrap around the DNA double helix.

The giant protein has a basic domain typical of this family of proteins. Its leucine zipper is somewhat aberrant in that two of the positions normally occupied by leucines, in the first and last heptad, are instead filled by isoleucine and phenylalanine, respectively. These are still hydrophobic residues but their size and shape differ from those of leucine and may require some distortion to be accomodated in a coiled-coil configuration. While other leucine zipper proteins include one non-leucine residue in the first position of a heptad, few have more than one. Such changes are frequently associated with decreased ability to dimerize and lower DNA-binding affinity (Hai et al., 1989; Hu et al., 1990). Departure from the canonical leucine may therefore cause giant to dimerize less well with itself but, perhaps, to form heterodimers more readily with a partner that presents complementary distortions. It is interesting therefore that the giant protein appears to be phos-phorylated in vivo. It has been reported that phosphorylation of CREB by protein kinase C increases dimerization and consequently DNA binding (Yama-moto et al., 1988). The giant protein contains several potential phosphorylation sites at serines and threonines that might modulate the ability of the protein to dimerize either with itself or with an unknown partner.

As might be expected from the high degree of conservation of the DNA-binding domain, there is a strong similarity among the nucleotide sequences recognized by members of the b-ZIP family. A survey of the available recognition sites shows that they might be divided into two classes: one exemplified by jun-APl has a center of dyad symmetry passing through a nucleotide. The other, represented by CREB, has the center of symmetry between two nucleotides. The sequences recognized by this class of proteins share a central core ACGT, with C/EBP occupying a somewhat aberrant position (central core GCGC). In addition, the left half site of these binding sites is in most cases the same as the left half site of the jun-APl recognition sequence: TGAC. Contacts made with these bases have been demonstrated for some members of the CREB family by methylation interference analysis (Hai et al., 1989). Fig. 9 shows that, on the basis of the CD1 and CD2 sequences, the giant protein fits best in the CREB class. The apparently stronger binding site, CD1, fits better than CD2 with the perfect half site TGAC but ACGC instead of the core ACGT. The sequence of three other giant binding sites have been published recently by Small et al. (1991) who determined three footprinting sites in the stripe 2 control region of the even-skipped gene. These sites appear to be considerably weaker than the CD1 and CD2 sites as well as aberrant in their sequence with respect to the motifs recognized by the majority of b-ZIP proteins. It is possible to identify in the eve footprints sequence motifs resembling the consensus illustrated in Fig. 9 but they appear less related to the CD1 and CD2 footprints than they are similar among themselves. It cannot be excluded that the eve sites are the targets not of a gt homodimer but of a heterodimer formed with a second b-ZIP protein present in the anterior region.

Interactions between gap genes

Genetic and molecular interactions among gap genes give evidence for a network strengthened by redundant cross interactions. In general, gap genes with adjacent domains of expression appear to repress one another: hb represses Kr and kni (Gaul and Jäckie, 1987; Hülskamp et al., 1990); kni represses Kr (Jäckle et al., 1986; Gaul and Jäckie, 1987); kni and gt repress one another (Eldon and Pirrotta, 1991); posterior gt is repressed by hb (Eldon and Pirrotta, 1991; Kraut and Levine, 1991b). The apparent positive interaction between Kr and kni, reported to stimulate kni expression (Pankratz et al., 1989) constituted an anomaly that accorded poorly with the generally repressive role played by Kr (Licht et al., 1990; Zuo et al., 1991). Our results show that the anomaly is only apparent: the loss of kni expression in Kr mutants is not due to a positive role of Kr but to the encroachment of gt into the kni domain. This expansion of the gt domain is caused in turn by the absence of the normal repression of gt by Kr. Pankratz et al. (1989) have found Kr binding sites in the upstream region of the kni gene and have shown that when a segment containing these sites is removed, the expression of the gene in the posterior domain is decreased. Our results indicate that this decrease is most likely due to the lack of some other transcription enhancing element. These Kr binding sites may instead mediate a modest negative effect of Kr on kni expression, similar to the repression observed between other adjacent gap genes.

The mutual repression of the adjacent gap genes serves to stabilize and limit their respective domains. However, diffusion of the gap gene products in the syncytial embryo causes an inevitable degree of overlap between the adjacent domains. This overlap is also necessary for the regulation of the downstream genes, the pair rule genes (Gaul and Jäckle, 1989; Riddihough and Ish-Horowicz, 1991). Therefore, to avoid a mutual shut off, the mutual repression by adjacent gap genes must be relatively weak so that it is effective only when the concentration of the repressing gene product is high. The interaction between gap genes with non-adjacent domains is a further device to increase the stability of the pattern. However, proteins like gt and Kr are present only at low concentrations in one another’s domain. If these are to interact, their affinity for one another must be relatively high. The strong effect of Kr on gt is essential to explain the broad pattern defects caused by Kr mutations. The overlap between the Kr and kni phenotypes is not simply due to the requirement of Kr for pair rule gene expression in the abdominal region but also to the fact that the absence of Kr entails the expansion of gt and consequent repression of kni. The Kr phenotype therefore will affect not only the parts of the abdomen that require Kr directly but also those that require high levels of kni.

Unlike the strong effect of Kr mutations on gt expression, the effect of gt mutations on Kr is not detectable. It is possible therefore that gt is not normally an important factor in setting the boundaries of the Kr central domain. Since we have no direct evidence so far that the ability of gt to repress Kr is functional in the normal embryo, we cannot yet exclude the possibility that this interaction does not normally occur and that, in our experiments, gt recognizes the targets of another gene product that normally regulates the Kr CD1 and CD2 elements. It is more likely, however, that the role of giant in regulating Kr is simply redundant and that any potential posterior expansion of Kr in a gt mutant may be checked by the intervening kni domain. However, we have not observed a stronger posterior shift of Kr in gp, kni double mutants than in kni single mutants (not shown). As has been suggested by Hülskamp et al. (1990), the central domain of Kr expression may be more determined by activation and repression exercised by both bed and hb proteins than by interactions with kni or gt.

We thank Michael Hoch, Ulrich Nauber, Henry Krause and Steve Cohen for clones, antibodies or suggestions. Rachel Kraut and Mike Levine let us use the clone containing the giant binding site in the eve stripe 2 element and shared their results before publication. M.C. is grateful to Sharon Bickel for advice in the early stages of this project and to Elaine McGuffin for the germ line transformation. We are indebted to Mitzi Kuroda, Su Qian and Juan Botas for reading and criticising this manuscript. M. C. was the recipient of a fellowship from the Ministère della Pubblica Istruzione of the Italian Government. E.D.E. was supported by an NIH postdoctoral training grant. This work was supported in part by NIH grant GM 34630 to V.P.

Angel
,
P.
,
Imagawa
,
M.
,
Chiu
,
R.
,
Stein
,
B.
,
Imbra
,
R. J.
,
Rahmsdorf
,
H. J.
,
Jonat
,
C.
,
Herrlich
,
P.
and
Karin
,
M.
(
1987
).
Phorbol ester-inducible genes contain a common cis element recognized by a TPA-modulated trans-actin factor
.
Cell
49
,
729
739
.
Benson
,
M.
and
Pirrotta
,
V.
(
1987
).
The product of the Drosophila zeste gene binds to specific DNA sequences in white and Ubx
.
EMBO J
.
6
,
1387
1392
.
Bickel
,
S.
and
Pirrotta
,
V.
(
1990
).
Self-association of the Drosophila zeste protein is responsible for trans-vection effects
.
EMBO J
.
9
,
2959
2967
.
Carroll
,
S. B.
and
Scott
,
M. P.
(
1986
).
Zygotically active genes that affect the spatial expression of the fushi tarazu segmentation gene during early Drosophila embryogenesis
.
Cell
45
,
113
126
.
Cavener
,
D. R.
(
1987
).
Comparison of the consensus sequence flanking translational start sites in Drosophila and vertebrates
.
Nucl. Acids Res
.
15
,
1353
1361
.
Driever
,
W.
and
Nüsslein-volhard
,
C.
(
1988
).
A gradient of bicoid protein in Drosophila embryos
.
Cell
54
,
83
93
.
Driever
,
W.
,
Thoma
,
G.
and
Nüsslein-volhard
,
C.
(
1989
).
Determination of spatial domains of zygotic gene expression in the Drosophila embryo by the affinity of binding sites for the bicoid morphogen
.
Nature
340
,
363
367
.
Edgar
,
B. A.
,
Weir
,
M. P.
,
Schubiger
,
G.
and
Kornberg
,
T.
(
1986
).
Repression and turnover pattern of fushi tarazu RNA in the early Drosophila embryo
.
Cell
47
,
747
754
.
Eldon
,
E. D.
and
Pirrotta
,
V.
(
1991
).
Interactions of the Drosophila gap gene giant with maternal and zygotic pattern-forming genes
.
Development
111
,
367
378
.
Garnier
,
J.
,
Osguthorpe
,
D. J.
and
Robson
,
B.
(
1978
).
Analysis of the accuracy and implications of simple methods for predicting the secondary structure of globular proteins
.
J. Molec. Biol
.
120
,
97
120
.
Gaul
,
U.
and
Jäckle
,
H.
(
1987
).
Pole region-dependent repression of the Drosophila gap gene Kriippel by maternal gene products
.
Cell
51
,
549
555
.
Gaul
,
U.
and
Jäckle
,
H.
(
1989
).
Analysis of maternal effect mutant combinations elucidates regulation and function of the overlap of hunchback and Kriippel gene expression in the Drosophila blastoderm embryo
.
Development
107
,
651
662
.
Gergen
,
J. P.
and
Wieschaus
,
E. F.
(
1986
).
Localized requirements for gene activity in segmentation of Drosophila embryos: analysis of armadillo, fused, giant and unpaired mutations in mosaic embryos
.
Roux’s Arch. Devi. Biol
.
195
,
49
62
.
Goto
,
T.
,
Macdonald
,
P.
and
Maniatis
,
T.
(
1989
).
Early and late periodic patterns of even skipped expression are controlled by distinct regulatory elements that respond to different spatial cues
.
Cell
57
,
413
422
.
Hal
,
T.
,
Liu
,
F.
,
Coukos
,
W. J.
and
Green
,
M. R.
(
1989
).
Transcription factor ATF cDNA clones: an extensive family of leucine zipper proteins able to selectively form DNA-binding heterodimers
.
Genes Dev
.
3
,
2083
2090
.
Harding
,
K.
,
Hoey
,
T.
,
Warrior
,
R.
and
Levine
,
M.
(
1989
).
Autoregulatory and gap gene response elements of the even-skipped promoter of Drosophila
.
EMBO J
.
8
,
1205
1212
.
Hill
,
D. E.
,
Hope
,
I. A.
,
Macke
,
J. P.
and
Struhl
,
K.
(
1986
).
Saturation mutagenesis of the yeast HIS3 regulatory site: requirements for transcriptional induction and for binding by GCN4 activator protein
.
Science
234
,
451
457
.
Hoch
,
M.
,
Schrttder
,
C.
,
Seifert
,
E.
and
Jäckle
,
H.
(
1990
).
Cis-acting control elements for Kriippel expression in the Drosophila embryo
.
EMBO J
.
9
,
2587
2595
.
Howard
,
K. R.
and
Struhl
,
G.
(
1990
).
Decoding positional information: regulation of the pair-rule gene hairy
.
Development
110
,
1223
1231
.
Hu
,
J. C.
,
O’shea
,
E. K.
,
Kim
,
P. S.
and
Sauer
,
R. T.
(
1990
).
Sequence requirements for coiled-coils: analysis with lambda repressor-GCN4 leucine zipper fusions
.
Science
250
,
1400
1403
.
Hfllskamp
,
H.
,
Pfeifle
,
M.
and
Tautz
,
D.
(
1990
).
A morphogenetic gradient of hunchback protein organizes the expression of the gap genes Kriippel and knirps in the early Drosophila embryo
.
Nature
346
,
577
580
.
Ingham
,
P. W.
(
1988
).
The molecular genetics of embryonic pattern formation in Drosophila
.
Nature
335
,
25
34
.
Jäckle
,
H.
,
Tautz
,
D.
,
Schuh
,
R.
,
Seifert
,
E.
and
Lehmann
,
R.
(
1986
).
Cross-regulatory interactions among the gap genes of Drosophila
.
Nature
324
,
668
670
.
Johnson
,
P. F.
,
Landschulz
,
W. H.
,
Graves
,
B. J.
and
Mcknlght
,
S. L.
(
1987
).
Identification of a rat liver nuclear protein that binds to the enhancer core element of three animal viruses
.
Genes Dev
.
1
,
133
146
.
Kellerman
,
K. A.
,
Mattson
,
D. M.
and
Duncan
,
I.
(
1990
).
Mutations affecting the stability of the fushi tarazu protein of Drosophila
.
Genes Dev
.
4
,
1936
1950
.
Kouzarides
,
T.
and
Ziff
,
E.
(
1989
).
Leucine zippers of Fos, Jun and GCN4 dictate dimerization specificity and thereby control DNA binding
.
Nature
340
,
568
571
.
Kraut
,
R.
and
Levine
,
M.
(
1991a
).
Spatial regulation of the gap gene giant during Drosophila development
.
Development
111
,
601
609
.
Kraut
,
R.
and
Levine
,
M.
(
1991b
).
Mutually repressive interactions between the gap genes giant and Kriippel define middle body regions of the Drosophila embryo
.
Development
111
,
611
621
.
Licht
,
J. D.
,
Grossel
,
M. J.
,
Figge
,
J.
and
Hansen
,
U. M.
(
1990
).
Drosophila Knlppel protein is a transcriptional repressor
.
Nature
346
,
76
79
.
Lin
,
Y. S.
and
Green
,
M. R.
(
1988
).
Interaction of a common cellular transcription factor, ATF, with regulatory elements in both E1A- and cyclic AMP-inducible promoters
.
Proc. Nam. Acad. Sci. USA
85
,
3396
3400
.
Lohmer
,
S.
,
Maddalonl
,
M.
,
Motto
,
M.
,
Di Fonzo
,
N.
,
Hartings
,
H.
,
Salaminl
,
F.
and
Thompson
,
R. D.
(
1991
).
The maize regulatory locus Opaque-2 encodes a DNA-binding protein which activates the transcription of the b-32 gene
.
EMBO J
.
10
,
617
624
.
Mohler
,
J.
,
Eldon
,
E. D.
and
Pirrotta
,
V.
(
1989
).
A novel spatial transcription pattern associated with the segmentation gene, giant, of Drosophila
.
EMBO J
.
8
,
1539
1548
.
Nauber
,
U.
,
Pankratz
,
M. J.
,
Kierdin
,
A.
,
Seifert
,
E.
,
Klemm
,
U.
and
Jäckle
,
H.
(
1988
).
Abdominal segmentation of the Drosophila embryo requires a hormone receptor-like protein encoded by the gap gene knirps
.
Nature
336
,
489
492.
O’shea
,
E. K.
,
Rutkowski
,
R.
and
Kim
,
P. S.
(
1989
).
Evidence that the leucine zipper is a coiled coil
.
Science
243
,
538
542
.
Pankratz
,
M. J.
,
Hoch
,
M.
,
Seifert
,
E.
and
Jäckle
,
H.
(
1989
).
Kruppel requirement for knirps enhancement reflects overlapping gap gene activities in the Drosophila embryo
.
Nature
341
,
337
340
.
Pankratz
,
M. J.
,
Seifert
,
E.
,
Gerwin
,
N.
,
Billl
,
B.
,
Nauber
,
U.
and
Jäckle
,
H.
(
1990
).
Gradients of Kriippel and knirps gene products direct pair-rule gene stripe patterning in the posterior region of the Drosophila embryo
.
Cell
61
,
309
317
.
Petschek
,
J.
,
Perrinton
,
N.
and
Mahowald
,
A. P.
(
1987
).
Region specific effects in 1(1) giant embryos of Drosophila
.
Dev. Biol
.
119
,
177
189
.
Petschek
,
J. P.
and
Mahowald
,
A. P.
(
1990
).
Different requirements for 1(1) giant in two embryonic domains of Drosophila melanogaster
.
Develop. Genet
.
11
,
88
96
.
Pignoni
,
F.
,
Baldarelli
,
R. M.
,
Steingrimsson
,
E.
,
Diaz
,
R. J.
,
Patapoutian
,
A.
,
Merriam
,
J. R.
and
Lengyel
,
J. A.
(
1990
).
The Drosophila gene tailless is expressed at the embryonic termini and is a member of the steroid receptor superfamily
.
Cell
62
,
151
163
.
Rlddihongh
,
G.
and
Ish-Horowicz
,
D.
(
1991
).
Individual stripe regulatory elements in the Drosophila hairy promoter respond to maternal, gap and pair-rule genes
.
Genes Dev
.
5
,
840
854
.
Rosenberg
,
U. B.
,
Schroder
,
C.
,
Preiss
,
A.
,
Kienlln
,
A.
,
Cote
,
S.
,
Riede
,
I.
and
Jäckle
,
H.
(
1986
).
Structural homology of the product of the Drosophila Kriippel gene with Xenopus transcription factor IIIA
.
Nature
319
,
336
339
.
Small
,
S.
,
Kraut
,
R.
,
Hoey
,
T.
,
Warrior
,
R.
and
Levine
,
M.
(
1991
).
Transcriptional regulation of a pair-rule stripe in Drosophila
.
Genes Dev
.
5
,
827
839
.
Stanley
,
K. K.
and
Luzio
,
J. P.
(
1984
).
Construction of a new family of high efficiency bacteria expression vectors: identification of cDNA clones coding for human liver proteins
.
EMBO J
.
3
,
1429
1434
.
Stanojevic
,
D.
,
Hoey
,
T.
and
Levine
,
M.
(
1989
).
Sequence-specific DNA-binding activities of the gap proteins encoded by hunchback and Kriippel in Drosophila
.
Nature
341
,
331
335
.
Steller
,
H.
and
Pirrotta
,
V.
(
1986
).
P transposons controlled by the heat shock promoter
.
Molec. cell. Biol
.
6
,
1640
1649
.
Struhl
,
G.
,
Struhl
,
K.
and
Macdonald
,
P. M.
(
1989
).
The gradient morphogen bicoid is a concentration-dependent transcriptional activator
.
Cell
57
,
1259
1273
.
Tautz
,
D.
(
1988
).
Regulation of the Drosophila segmentation gene hunchback by two maternal morphogenetic centres
.
Nature
332
,
281
284
.
Tautz
,
D.
,
Lehmann
,
R.
,
Schnurch
,
H.
,
Schuh
,
R.
,
Seifert
,
E.
,
Kienlin
,
A.
,
Jones
,
K.
and
Jäckle
,
H.
(
1987
).
Finger protein of novel structure encoded by hunchback, a second member of the gap class of Drosophila segmentation genes
.
Nature
327
,
383
389
.
Trelsman
,
J.
and
Desplan
,
C.
(
1989
).
The products of the Drosophila gap genes hunchback and Kriippel bind to the hunchback promoters
.
Nature
341
,
335
337
.
Vinson
,
C. R.
,
Sigler
,
P. B.
and
Mcknight
,
S. L.
(
1989
).
Scissors-grip model for DNA recognition by a family of leucine zipper proteins
.
Science
246
,
911
916
.
Wharton
,
K. A.
,
Yedvobnick
,
B.
,
Finnerty
,
V. G.
and
Artavanis-Tsakonas
,
S.
(
1985
).
Opa: a novel family of transcribed repeats shared by the Notch locus and other developmentally regulated loci in D. melanogaster
.
Cell
40
,
55
62
.
Wieschaus
,
E.
and
Nüsslein-volhard
,
C.
(
1986
).
Looking at embryos. In Drosophila
.
A Practical Approach
, (ed.
D. B.
Roberts
), (
Washington, D.C
.:
IRL Press
), pp.
199
227
.
Wieschaus
,
E.
,
Nüsslein-volhard
,
C.
and
Kluding
,
H.
(
1984
).
Kriippel, a gene whose activity is required early in the zygotic genome for normal embryonic segmentation
.
Dev. Biol
.
104
,
172
186
.
Yamamoto
,
K. K.
,
Gonzalez
,
G. A.
,
Biggs
,
W. H. HI
and
Montmlny
,
M. R.
(
1988
).
Phosphorylation-induced binding and transcriptional efficacy of nuclear factor CREB
.
Nature
334
,
494
498
.
Zuo
,
P.
,
Stanojevic
,
D.
,
Colgan
,
J.
,
Han
,
K.
,
Levine
,
M.
and
Manley
,
J. L.
(
1991
).
Activation and repression of transcription by the gap proteins hunchback and Krüppel in cultured Drosophila cells
.
Genes Dev
.
5
,
254
264
.