ABSTRACT
The 2B5 early puff locus corresponds to the Broad-Complex (BR-C) and encodes a family of transcription factors whose members are induced by the molting hormone ecdysone. Mutations in the br subcomplementation group substantially reduce the levels of Dopa decarboxylase (DDC) in the epidermis of mature third instar larvae but not in mature second instar organisms. Enzyme levels are normal in the central nervous system of the two mutants examined. The specificity of these effects suggests that a product of the BRC locus mediates the rapid appearance of DDC in mature third instar larvae experiencing an elevated titer of ecdysone. The likely identity of this protein has been confirmed by pursuing the observation that the br28 allele is caused by the insertion of a P element into the Z2 DNAbinding domain. Both the transcript and a protein carrying this domain are present in the epidermis and a BR-C recombinant protein carrying the Z2 finger binds to the first intron of the Ddc gene. Five binding sites have been identified within the intron by DNAase I footprinting and a core consensus sequence has been derived which shares some identity with the consensus binding site of the Z2 protein to the Sgs-4 regulatory region. Our demonstration that Ddc is a target of BR-C in the epidermis is the first direct evidence of a role for this early gene in a tissue other than the salivary glands. The data reinforce the idea that BR-C, which clearly mediates a salivary gland-specific response to ecdysone, may play a widespread role in the hormone’s activation of gene cascades in other target tissues.
INTRODUCTION
Studies on the insect moulting hormone began 60 years ago with the classic work of Fraenkel (1935) which demonstrated the release of the metamorphic hormone near the end of the larval stage in the blow-fly Calliphora. Subsequent work by Clever, Karlson, Sekeris and their colleagues (for review, see Sekeris, 1991) led to the identification of 20-OH-ecdysone (herein referred to as ecdysone) as the molting hormone and established the ecdysone system as a useful model for understanding the mechanisms of vertebrate steroid hormone action. The enzyme Dopa decarboxylase (DDC) was the first welldefined gene product whose appearance during development appeared to be regulated by ecdysone (Karlson and Sekeris, 1962). DDC catalyses the conversion of Dopa to dopamine that is required to tan and harden the newly moulted cuticle of diptera. A conceptual framework describing how Ddc gene regulation might occur was provided by Ashburner and Richards from their studies on the induction of chromosome puffing patterns in explanted salivary glands cultured under a variety of ecdysone regimes (Ashburner et al., 1974). Approximately six ‘early genes’, visualized as the corresponding set of early puffs, is directly induced by an ecdysone-receptor complex at the end of the third instar period just prior to puparium formation. These genes act as autoregulators, eventually shutting off their own transcription and that of the intermolt genes some of which encode the salivary gland glue proteins. They also induce the large set of ‘late genes’, which comprise the tissue-specific cascade seen in the puffing studies. Recent studies in Drosophila have confirmed the general features of the model. The gene that encodes the ecdysone receptor (EcR) has been cloned and the majority of the EcRbinding sites on the salivary gland chromosomes correspond to early and late puff loci (Koelle et al., 1991). The three early genes that have been characterized to date all encode families of transcription factors (Burtis et al., 1990; Segraves and Hogness, 1990; DiBello et al., 1991). In two cases, E74A and E75A, the protein products have been shown to bind to both the early and late target puff loci (Urness and Thummel, 1990; Hill et al., 1993). A central role in the regulatory hierarchy controlled by ecdysone is played by the Broad-Complex (BR-C), the early gene which resides at puff locus 2B5 on the X-chromosome. Mutations in BR-C affect not only intermolt and late gene expression (Guay and Guild, 1991) but also E74A, E75A and BR-C early gene induction, which has prompted a revision of the original model (Karim et al., 1993).
In fact, the Ashburner model fails to convey adequately the complexity of responses to the variations in hormone titer that occur in late larval stages and through the prepupal period. The late gene class has been differentiated into early-late and latelate genes on the basis of hormone withdrawal experiments (Ashburner and Richards, 1976). The early-late genes share with the classic early genes a requirement for the presence of ecdysone for their continued expression. They are similar to the late genes in that the appearance of puffs at early-late loci is dependent upon continued protein synthesis. The E78 earlylate gene has now been cloned and shown to encode two nested transcription units E78A and E78B (Stone and Thummel, 1993). Both transcripts appear during the pupal stage but the E78B transcript profile in vivo shows maximal expression at pupariation. However, unlike its puff at 78C, the E78B transcript is induced by ecdysone in the presence of cycloheximide and the locus is therefore more properly viewed as a variant of the early gene class (Stone and Thummel, 1993; Huet et al., 1995).
The hormonal control of epidermal Ddc appears to be very similar to that of E78B in several respects. The Ddc transcript reaches maximum levels at pupariation (Andres et al., 1993) and Ddc mRNA is rapidly induced in hormonally naive epidermis following exposure to exogenous ecdysone (Kraminsky et al., 1980). Also, although inhibitors of protein synthesis reduce they do not eliminate Ddc mRNA accumulation (Clark et al., 1986). Since the inhibition of protein synthesis does reduce Ddc levels substantially from control levels, the full induction of Ddc must involve a protein product synthesized after the addition of exogenous ecdysone. The extensive collection of BR-C mutants facilitated the genetic test of a possible interaction between this early gene and Ddc. We demonstrate the existence of such an interaction and have pursued the analysis to the point of showing that a recombinant protein encoded by BR-C binds to the first intron of the Ddc gene. The sequences to which binding occurs are included within a cis-acting region that Shen and Hirsh (1994) have shown controls the tissue specificity of the splicing of the primary Ddc transcript. This raises the interesting possibility that the BR-C might participate in regulating the Ddc splicing pathway in the epidermis.
MATERIALS AND METHODS
Drosophila stocks
BR-C mutations were maintained over Binsn, an X-chromosome balancer carrying the markers Bar and singed (Lindsley and Zimm, 1992). The mutations were carried on a chromosome marked with yellow and were isolated by Kiss et al. (1988), Belyaeva et al. (1980) and C. Bayer (personal communication). All are recessive lethals with the exception of brAub and, amongst the lethals, all but npr17 pupariate but die shortly thereafter. The molecular basis of the mutants with the exception of br28 (see later) is unknown. All organisms were maintained on standard fly food (Nash and Bell, 1968) at 22°C.
Enzyme assays
Dopa decarboxylase activity was measured (Walter et al., 1991) in extracts prepared from frozen organisms at concentrations of 1 organism/20 μl homogenization buffer (newly molted third instar larvae), 1 organism/50 μl buffer (white prepupae) or 1 brain/5 μl. Only male organisms were selected and those carrying yellow (experimental group) separated from wild type (controls) based on the pigmentation of the denticle belts and mouth parts.
Purification of BR-C recombinant proteins
The expression plasmids from which we obtained the BR-C proteins were constructed by cloning cDNA fragments into the vector pDSMCS (Schindler et al., 1992) as described elsewhere (von Kalm et al., 1994). The fragments contain the core region and one of the four possible zinc finger motifs encoded at the 3′ end of the gene (DiBello et al., 1991; Bayer and Fristrom, personal communication). The Z1 isoform is designated as BRcore-Q1-Z1 since it contains a glutaminerich (Q) region and the Z3 isoform as BRcore-NS-Z3 since it contains an asparagine and serine-rich (NS) domain (DiBello et al., 1991). The recombinant proteins were purified by Ni2-NTA column chromatography (Hochuli, 1990) as in von Kalm et al. (1994). Fractions containing the recombinant protein were identified by SDS acrylamide gel electrophoresis and pooled. The denatured protein was dialysed successively for 2 hour periods at 4°C against a buffer containing 10 mM Hepes (pH 7.9), 80 mM KCl, 1 mM DDT, 0.1% Triton X-100, 10 μM ZnCl2, 5% glycerol and (a) 1 M urea, (b) 0.1 M urea and (c) no urea and 20% glycerol. In the final dialysis step for Z2 isoform purification, a white flocculent precipitate developed but it disapppeared upon further dialysis into 8 M urea, 0.1 M NaH2PO4, 0.1 M Tris-HCl (pH 8.0) and did not reappear when the protein was again renatured by carrying out the dialysis steps above. Purified Z2 and Z3 proteins were stored in 10 mM Hepes (pH 7.9), 80 mM KCl, 1 mM DDT, 0.1% Triton X-100, 10 μM ZnCl2, 20% glycerol. Z1 and Z4 were stored in the same buffer containing 1 M urea, since they aggregated and precipitated at lower urea concentrations. The recombinant BR-C proteins all have a POZ domain at their N terminus; this domain (also known as the BTB domain; Zollman et al., 1994) facilitates protein-protein interactions in both homomeric and heteromeric molecules depending on the particular protein involved (Bardwell and Treisman, 1994). This might explain the propensity of the BR-C recombinant proteins to aggregate during renaturation.
Filter binding assay
Binding of the purified Z2 recombinant protein to a 7.6 kb restriction fragment that encodes the entire Ddc locus (Scholnick et al., 1983; Chen and Hodgetts, 1987) was carried out using nitrocellulose filters according to Papoulas (1988). An appropriate restriction digest of the fragment was end labelled using the Klenow fragment of E. coli DNA polymerase I (Sambrook et al., 1989) and binding to the BR-C recombinant protein carried out in a 50 μl reaction containing: 2×105 cts/minute labelled fragments, 1-5 μl protein (50 μg /ml), 1 μg BSA, 50 mM KCl, 10% glycerol, 0.1 mM DTT, 0.1 mM PMSF and 20 mM Tris-HCl (pH 7.5). The bound fragments were eluted, precipitated and fractionated on a 2% agarose gel on which an equal quantity of the input DNA was loaded so that preferential retention of the bound DNA could be visualized on the subsequent autoradiograph of the dried gel.
Gel retardation assay
Mobility shift assays with recombinant BR-C protein were performed on Ddc DNA fragments as described by Chodash (1988), with minor modifications. A plasmid containing the EcoRI-SalI Ddc region of interest (Fig. 1) was linearized and end-labelled with γ-32P-ATP and polynucleotide kinase. The end-labelled fragment was released from the vector by restriction, subjected to agarose gel electrophoresis and purified from the gel using Geneclean II (Bio101, Inc). Binding was carried out for 60 minutes at 4°C with 4-6×104 cts/minute of the labelled fragment, 4-8 μg poly(dI-dC).(dI-dC) and up to 8 μg BR-C protein in a 20 μl reaction volume containing 20 mM Tris-HCl (pH 7.5), 10% glycerol and 50 mM KCl. The binding reaction was fractionated on a 4% acrylamide gel which was dried and subjected to autoradiography. The conditions used to demonstrate binding to the 217 bp DraI-TaqI Ddc DNA fragment (Fig. 1) differed somewhat from those just described. The fragment that was subcloned into pUC19 was released from the vector by EcoRI-HindIII digestion, labelled by end-filling the restriction sites using Klenow and purified from an acrylamide gel. The binding reaction contained 0.5 μg of poly(dI-dC).
Localization of Z2 binding to the Ddc genomic region using the filter binding assay. (A) The neuraland epidermal-specific splicing patterns of theDdc transcription unit are shown in alignment with a restriction map of the 7.6 kb genomic fragment which is sufficient for normal Ddc developmental expression. The sites shown are EcoRI (R1), DraI (D), HindIII (H), TaqI (T), BamHI (B), SalI (S), PstI (P) and SstI (Ss). An enlargement of the 736 bp EcoRI-SalI fragment within which binding was eventually localized is shown beneath the restriction map. (B) The filter binding assay (Materials and Methods) was carried out following endlabelling of the products of a HindIII-SstI digest of a plasmid carrying the 7.6 kb fragment. The three shortest molecules of the nine produced (see A) were not visualized on the gel shown. Lane 1, an aliquot of the unbound digest; lanes 2 and 3, fragments eluted from the filter following binding to 0.125μg of Z2 and 0.25 μg of Z2 respectively. Each lane was loaded with the same number of counts. The arrows indicate the fragments that were preferentially retained on the filter by Z2: the 2.7 kb molecule comprises the entire vector into which the genomic DNA was cloned and the 1.3 kb fragment is the HindIII fragment that includes the two small Ddc exons (see A). (C) The digest shown in B was restricted with SalI which cleaves the 1.3 kb HindIII fragment into 720 and 624 bp molecules; all ten products in the double digest are visible on the autoradiograph of the gel (lane 1). Lanes 1-3 were as described in B and binding (arrow) was confined to the 720 bpHindIII-SalI portion of the 1.3 kb fragment. (D) The 720 bp HindIII-SalI fragment is included within a slightly larger EcoRI-SalI fragment, a DraI digest of which yields products of 487 bp and 249 bp (see A). These were end-labelled and subjected to filter binding. Lane 1, unbound fragments; lane2, fragments eluted from the filter after binding to 0.5μg of Z2. Binding (arrow) occurred predominately to the 487 bp fragment.
Localization of Z2 binding to the Ddc genomic region using the filter binding assay. (A) The neuraland epidermal-specific splicing patterns of theDdc transcription unit are shown in alignment with a restriction map of the 7.6 kb genomic fragment which is sufficient for normal Ddc developmental expression. The sites shown are EcoRI (R1), DraI (D), HindIII (H), TaqI (T), BamHI (B), SalI (S), PstI (P) and SstI (Ss). An enlargement of the 736 bp EcoRI-SalI fragment within which binding was eventually localized is shown beneath the restriction map. (B) The filter binding assay (Materials and Methods) was carried out following endlabelling of the products of a HindIII-SstI digest of a plasmid carrying the 7.6 kb fragment. The three shortest molecules of the nine produced (see A) were not visualized on the gel shown. Lane 1, an aliquot of the unbound digest; lanes 2 and 3, fragments eluted from the filter following binding to 0.125μg of Z2 and 0.25 μg of Z2 respectively. Each lane was loaded with the same number of counts. The arrows indicate the fragments that were preferentially retained on the filter by Z2: the 2.7 kb molecule comprises the entire vector into which the genomic DNA was cloned and the 1.3 kb fragment is the HindIII fragment that includes the two small Ddc exons (see A). (C) The digest shown in B was restricted with SalI which cleaves the 1.3 kb HindIII fragment into 720 and 624 bp molecules; all ten products in the double digest are visible on the autoradiograph of the gel (lane 1). Lanes 1-3 were as described in B and binding (arrow) was confined to the 720 bpHindIII-SalI portion of the 1.3 kb fragment. (D) The 720 bp HindIII-SalI fragment is included within a slightly larger EcoRI-SalI fragment, a DraI digest of which yields products of 487 bp and 249 bp (see A). These were end-labelled and subjected to filter binding. Lane 1, unbound fragments; lane2, fragments eluted from the filter after binding to 0.5μg of Z2. Binding (arrow) occurred predominately to the 487 bp fragment.
DNAase I footprinting analysis
The 217 bp DraI-TaqI Ddc DNA fragment (Fig. 1) was end-labelled by polynucleotide kinase at the HindIII (transcribed strand) or EcoRI (non-transcribed strand) sites flanking the insert and then released from the vector by appropriate restriction. DNAase I (Worthington) footprinting was carried out as we have described elsewhere (von Kalm et al., 1994).
RNA extraction and RT-PCR
The detection of BR-C transcript levels in the epidermis was carried out using RT-PCR, as described elsewhere (O’Keefe et al., 1995). The epidermis of wandering third instar larvae was dissected free of all other tissue in 0.75% saline and total RNA obtained (O’Keefe et al., 1995). RNA was also obtained for amplification from whole organisms, collected at the same time. The RNA (2 μg) was primed with oligo(dT) and transcribed by reverse transcriptase (Superscript II, Gibco BRL). The subsequent DNA amplification was carried out on 1/5 of the cDNA in a PCR mix containing 10% glycerol and primer pairs (0.2 μM) chosen to amplify each of the zinc-finger-binding domains as shown in Fig. 5. The reactions were carried out as follows: cycle 1: 93°C for 4 minutes, 60°C for 1 minute and 73°C for 2 minutes followed by 29 repetitions of cycle 1 except that the denaturation step was 91°C for 1 minute. All four of the PCR reactions were anchored with a common primer (C) located at the 3′ end of the core region of BR-C. The other primers were located just within or immediately 3′ to the respective zinc finger motifs. The sequences of the five primers were obtained from DiBello et al., (1991) with the exception of Z4, which was designed from sequence data kindly provided by Cindy Bayer. (C): 5′-ACAAGATGTTCCATGCAGCC-3′; Z1: 5′-TGCTGGTGCTGCTGGTGATA-3′; Z2: 5′-TCATCTCCATTTCGCCGGGA-3′; Z3: 5′-GATGGCGGTCGTCTTAAGCA-3′; Z4: 5′GTGGTTGTTCAGCGAGTTCA-3′.
Western blot analysis
Whole animals and hand-dissected epidermis were processed for western blotting as described in Emery et al. (1994). Following electrophoresis on 7% acrylamide gels, the proteins were transfered to Hybond-ECL membranes and visualized using the ECL detection kit (Amersham) according to the manufacturer’s instructions. The monoclonal antibody 25E9, specific for the BRcore domain (Emery et al., 1994), was used to detect all BR-C isoforms. Protein loading was determined by the visualization of Coomassie-stained residual protein bands that remained on the gel following transfer.
RESULTS
DDC enzyme levels are under control of the BroadComplex
Organisms were collected within a 1 hour period following the second molt and again as white prepupae with the exception of npr17 which fails to pupariate. In this case, wandering third instar larvae were collected from food that had been supplemented with 0.05% bromophenol blue (Andres and Thummel, 1994). Those whose guts were blue were returned to the food to allow clearing of the dye which was monitored 1.5 hours later. Mutant and wild-type organisms whose guts had cleared of dye within this period were returned to the food for 2 hours and then harvested for analysis. The time of clearing of the dye in wild-type animals was determined to occur 2.5 hours before pupariation (data not shown). In presenting the results, DDC activity of mutant males is expressed as a percentage of the activity of the wild-type sibs segregating in the stock. Enzyme levels in whole organisms reflect predominately epidermal levels since the contribution of neural DDC to the total is less than 10% (Scholnick et al., 1983). Two and sometimes three independent collections of organisms were assayed at the two stages and the results are shown in Table 1. Organisms assayed at the second moult showed no significant difference among mutant strains. However, an analysis of variance (not shown) showed significant (P<0.01) relative differences in DDC levels among the mutant strains at pupariation. Duncan’s multiple range test revealed three different groups of mutant strains. The first group contained strains whose enzyme levels were equivalent to the wild type: 2Bc1, 2Bc2, rbp4, brAub, 2Bab1, 2Bab3 and 2Bab6. The second group contained strain br8 with about half the normal enzyme level and the third group was comprised of br5, br28, npr17 and 2Bab5 all of which exhibited DDC levels less than one third normal. There is a strong correlation between mutations in the br subcomplementation group and reduced DDC activity. In fact the only other mutants that affected DDC activity were npr17 and 2Bab5. The former fails to complement any of the BR-C mutations while the latter belongs to the 2Bab subcomplementation group, whose lethal alleles also fail to complement the br group (Belyaeva et al., 1980; Kiss et al., 1988). The observation that only one of the four 2Bab alleles tested reduced DDC activity is somewhat surprising although the 2Bab5 allele is the strongest of the four that we examined (C. Bayer, personal communication).
Ddc transcript levels in a subset of these mutants have been determined by quantitative RT-PCR (O’Keefe et al., 1995). In every case where there were substantial reductions in DDC activity levels, there were comparable reductions in the steady state levels of the Ddc transcript pool. In the rbp4 mutant where the enzyme levels were approximately normal, transcript levels were also normal.
In order to test the effect of BR-C mutations on neural DDC activity, enzyme activity was assayed in the dissected brains of mature larvae from two of the mutants strains showing the largest reductions in total DDC levels. The neural complex comprising the optic lobes and the ventral ganglion from hemizygous npr17 wandering male larvae was collected and the DDC determinations expressed as a percentage of the values obtained in the brains of wild-type organisms segregating in the same cross. The mean value of the three extracts was 120% in contrast to the value of 22% in whole organisms (Table 1), which reflects the epidermal level predominately. One assay on brains collected from br5 mutant males yielded 111% of the activity of brains from wild-type males. These data make it clear that, although allele-specific reductions in epidermal levels of DDC occur in mature larvae, the enzyme activity in the brains of these organisms is not reduced by the BR-C mutations.
BR-C proteins can bind to the Ddc genomic region
The data in Table 1 strongly suggest that mutations in the br subcomplementation group affect Ddc gene activation at pupariation. The fact that one of these mutations, br28, is caused by a P element insertion into a genomic region encoding the BR-C Z2 isoform (DiBello et al., 1991; Emery et al., 1994), led us to ask directly whether the Z2 isoform can bind to Ddc genomic DNA. Distinct tissuespecific Ddc transcripts, localized to the central nervous system and the epidermis respectively, are derived from this gene (Eveleth et al., 1986; Morgan et al., 1986). They are illustrated in Fig. 1A along with a restriction map of the gene. The recombinant Z2 isoform was prepared as described in the Materials and Methods and the filter binding assay was carried out on different restriction digests of the 7.6 kb PstI genomic DNA fragment (Fig. 1A) known to include all regulatory elements required for proper temporal and spatial Ddc expression (Scholnick et al., 1983; Marsh et al., 1985; Chen and Hodgetts, 1987). A HindIII-SstI digest of a plasmid carrying this fragment produces 9 products (Fig. 1A), two of which were preferentially retained on the filter by the Z2 protein (Fig. 1B). The bound 2.7 kb DNA molecule is derived exclusively from the vector. The retained 1.3 kb HindIII fragment includes much of the first intron of the epidermal transcript (Fig. 1A). Further digestion with SalI cleaved the 1.3 kb fragment into two fragments of 720 and 624 bp. Using the triple digest in the filter binding assay (Fig. 1C) clearly shows Z2 binding to only the 720 bp HindIII-SalI region which lies almost entirely within the first intron of the epidermal transcript (Fig. 1A). The final localization of the binding region is shown in Fig. 1D. The 736 bp EcoR1-SalI fragment was isolated, end-labelled and restricted with DraI which produces two fragments of 249 and 487 bp respectively (Fig. 1A). The Z2 isoform binds almost exclusively to the larger fragment, which is located predominately within the first intron of the epidermal Ddc gene but includes the second exon of the neural transcript.
The 736 kb EcoRI-SalI fragment and subfragments within it (see Fig. 1A) were subjected to mobility shift experiments with the recombinant Z2 protein as described in the Materials and Methods. The full-length EcoRI-SalI fragment shows extensive retardation by Z2 and the binding proceeds through one or more intermediates at non-saturating levels of protein (Fig. 2A). The non-discreet nature of these intermediates may result from the rather large size of this fragment. The binding to the EcoRI-SalI piece is eliminated in the presence of a 75fold excess of unlabelled fragment (Fig. 2B). Additional binding studies (data not shown) revealed that little binding occurred to either the 249 bp EcoRI-DraI or the 272 bp TaqISalI intervals on the 736 bp EcoRI-SalI fragment. These results suggested that all of the Z2 binding was localized to the central 217 bp DraI-TaqI fragment (see Fig. 1A). When this fragment was isolated, labelled and subjected to mobility shift analysis, extensive retardation was seen (Fig. 2C).
Mobility shift experiments using Z2. The 736 bp EcoRI-SalI fragment or the 217 bp DraI-TaqI subfragment within it (Fig. 1A) were end-labelled and subjected to mobility shift analyses as described in the Materials and Methods. (A) Binding to the EcoRI-SalI fragment was carried out in the presence of 0, 0.5, 1, 2, 4 and 8 μg of Z2 protein, respectively. (B) Binding of Z2 to the EcoRI-SalI fragment was carried out with 0 (lane 1) and 6 μg of Z2 (lanes 2 and 3). In lane 3 the binding was carried out in the presence of a 75-fold excess of unlabelled fragment. Some of the retarded DNA did not enter the gel in lane 2 and forms an apparent band at the origin. (C) Binding of Z2 to the DraI-TaqI fragment in the presence of 0, 0.5, 1 and 4 μg respectively of protein.
Mobility shift experiments using Z2. The 736 bp EcoRI-SalI fragment or the 217 bp DraI-TaqI subfragment within it (Fig. 1A) were end-labelled and subjected to mobility shift analyses as described in the Materials and Methods. (A) Binding to the EcoRI-SalI fragment was carried out in the presence of 0, 0.5, 1, 2, 4 and 8 μg of Z2 protein, respectively. (B) Binding of Z2 to the EcoRI-SalI fragment was carried out with 0 (lane 1) and 6 μg of Z2 (lanes 2 and 3). In lane 3 the binding was carried out in the presence of a 75-fold excess of unlabelled fragment. Some of the retarded DNA did not enter the gel in lane 2 and forms an apparent band at the origin. (C) Binding of Z2 to the DraI-TaqI fragment in the presence of 0, 0.5, 1 and 4 μg respectively of protein.
The DraI-TaqI fragment was subjected to DNAase I foot-printing using the recombinant Z2 protein (Fig. 3). Four obvious footprints, each associated with a DNAase I hypersensitive site, were resolved on the transcribed strand (Fig. 3A). These footprints were numbered beginning at the promoter proximal site. The extent of each footprint is shown beside the autoradiograph. Analysis of the non-transcribed strand (Fig. 3B) confirmed the existence of these four footprints and revealed the existence of a fifth which is situated within a gap in the control DNAase I ladder on the transcribed strand, immediately upstream of site 5. Fig. 3 also shows the footprinting analyses that were carried out with Z1, Z3 and Z4 recombinant proteins, the other BR-C isoforms. One or more of these proteins protects part of every one of the five Z2-binding sites. In fact, there appears to be just one unique site among all the binding sites visualized and that is Z45. The complex pattern of overlaps among the binding sites is summarized schematically in Fig. 4A. The nucleotide sequence of the DraI-TaqI fragment is shown in Fig. 4B. When the sequences on both DNA strands within the 17 binding sites identified in Fig. 3 were analysed and aligned, the consensus binding sites shown in Fig. 4B were derived. Five AT-rich Z2 binding sites all of which share a common core sequence, CTAT, were identified. This core sequence is also present in the consensus binding site of Z2 to the Sgs-4 regulatory domain (von Kalm et al., 1994) as the comparison of Z2 footprints at the two target loci in Fig. 4B shows. Not surprisingly, given the coincidence of the Z1and Z4-binding sites (Fig. 4A), their consensus sequences are rather similar (Fig. 4B). However, the Z1 consensus contains a perfectly conserved ATTA core at all four binding sites whereas the Z4 sites all share a TAA triplet. In the case of Z1, the conserved core at Ddc is included within the consensus binding site at Sgs-4, although the absolutely conserved sequence at Sgs-4 is CAA. Likewise, although the Ddc and Sgs-4 binding sites for Z4 are similar (Fig. 4B), the conserved triplets are TAA and AAA, respectively.
DNAase I footprinting analyses using the BR-C recombinant proteins. Binding of the 217 bp DraI-TaqI fragment (Fig. 1A) to the four recombinant proteins was followed by DNAase I digestion and electrophoresis on a 6% denaturing polyacrylamide gel. A 3 day autoradiograph of the dried gel is shown and the extent of the footprints on each of the two DNA strands is indicated in A and B, respectively. The sites are numbered from upstream to downstream relative to the direction of transcription of the Ddc gene. Lanes marked A+G contain sizing ladders of the templates used for footprinting and those marked with (−) are controls with no protein added. Z1: 10 μg BRcore-Q1-Z1 protein; Z2: 4 μg BRcore-Z2 protein; Z3: 2.5 μg BRcore-NS-Z3 protein; Z4: 10 μg BRcore-Z4 protein. All reactions contained a 2000to 2500-fold excess (0.5 μg) of poly(dI/dC) as a non-specific DNA competitor.
DNAase I footprinting analyses using the BR-C recombinant proteins. Binding of the 217 bp DraI-TaqI fragment (Fig. 1A) to the four recombinant proteins was followed by DNAase I digestion and electrophoresis on a 6% denaturing polyacrylamide gel. A 3 day autoradiograph of the dried gel is shown and the extent of the footprints on each of the two DNA strands is indicated in A and B, respectively. The sites are numbered from upstream to downstream relative to the direction of transcription of the Ddc gene. Lanes marked A+G contain sizing ladders of the templates used for footprinting and those marked with (−) are controls with no protein added. Z1: 10 μg BRcore-Q1-Z1 protein; Z2: 4 μg BRcore-Z2 protein; Z3: 2.5 μg BRcore-NS-Z3 protein; Z4: 10 μg BRcore-Z4 protein. All reactions contained a 2000to 2500-fold excess (0.5 μg) of poly(dI/dC) as a non-specific DNA competitor.
(A) The six BR-C-binding sites revealed by the footprinting data are shown below the line and the putative ecdysone receptor-binding site (EcR) above the line that represents the 217 bp DraI-TaqI fragment of the first intron (Fig. 1A). The various isoforms that bind at or near these sites are shown below each site. (B) The DNA sequences of the 17 BR-C isoform-binding sites identified by the footprinting are summarized below the sequence of the non-transcribed strand of the DraI-TaqI fragment that was taken from Eveleth et al. (1986). Since this sequence differed in one place from that published in Morgan et al. (1986), we confirmed our data in Eveleth et al. (1986) using Sanger’s method and note further that the A+G ladders used for the footprinting (Fig. 3) are consistent with the sequence shown. The location of the sequence which is very similar to an edysone receptor-binding site (Cherbas et al., 1991) is identified by the underlined sequence. The footprint sequences are all shown on the non-transcribed strand with the exception of Z44. Absolutely conserved core regions in the binding sites are underlined. The Sgs-4 consensus sequences are taken from von Kalm et al. (1994). Y=either pyrimidine; N=any nucleotide; W=A or T; M=A or C.
(A) The six BR-C-binding sites revealed by the footprinting data are shown below the line and the putative ecdysone receptor-binding site (EcR) above the line that represents the 217 bp DraI-TaqI fragment of the first intron (Fig. 1A). The various isoforms that bind at or near these sites are shown below each site. (B) The DNA sequences of the 17 BR-C isoform-binding sites identified by the footprinting are summarized below the sequence of the non-transcribed strand of the DraI-TaqI fragment that was taken from Eveleth et al. (1986). Since this sequence differed in one place from that published in Morgan et al. (1986), we confirmed our data in Eveleth et al. (1986) using Sanger’s method and note further that the A+G ladders used for the footprinting (Fig. 3) are consistent with the sequence shown. The location of the sequence which is very similar to an edysone receptor-binding site (Cherbas et al., 1991) is identified by the underlined sequence. The footprint sequences are all shown on the non-transcribed strand with the exception of Z44. Absolutely conserved core regions in the binding sites are underlined. The Sgs-4 consensus sequences are taken from von Kalm et al. (1994). Y=either pyrimidine; N=any nucleotide; W=A or T; M=A or C.
The BR-C is expressed in the epidermis prior to maximal Ddc expression
A reverse transcriptase-polymerase chain reaction (RT-PCR) approach was used to test whether BR-C transcripts are present in epidermal preparations that express DDC enzyme activity. Total RNA was isolated from hand-dissected epidermal preparations of wandering larvae, amplified and the products visualized on an agarose gel (Fig. 5). The products detected in whole organisms of the same stage are included for comparison. All four of the zinc finger classes could be detected and the sizes of the amplified bands (0.78 kb (Z1), 0.32 kb (Z2), 0.78 kb (Z3) and 1.1 kb (Z4)) were those expected for the BRcore-Q1-Z1, BRcore-Z2, BRcore-Z3 and BRcore-Z4 isoforms, respectively (DiBello et al., 1991; C. Bayer, personal communication). The two isoforms of Z1, Q1 and Q2, differ by only 51 nucleotides (DiBello et al., 1991) and may not have been resolved on our gels. The ‘TNT’ variants of the Q1-Z1 isoform (DiBello et al., 1991) could have been resolved but were never observed. In every case, the epidermal products that we did detect were indistinguishable from those detected in whole organisms.
Identification of epidermal BR-C proteins by western blotting
Protein extracts from epidermal preparations of wild-type and br28 mutant organisms selected as white prepupae were subjected to analysis. Extracts from similarly staged whole organisms were included for comparison. Proteins were visualized on blots probed with a monoclonal antibody able to detect all BR-C proteins. Isoforms of Z1, Z2 and Z3 were detected in both whole organisms and in the epidermis (Fig. 6A). The two Z2 isoforms, p64-Z2 and p57-Z2, represent the predominant BR-C proteins in the epidermis. The p57 protein co-migrates with the recombinant BRcore-Z2 molecule produced in E. coli and the p64 protein may represent a previously unidentified isoform of Z2 (Emery et al., 1994). Neither of these isoforms is present in br28 larvae (Emery et al., 1994; Fig. 6B) nor in epidermal extracts of the mutant (Fig. 6C). Instead, a novel, apparently truncated BR-C protein appears in br28 (Fig. 6B,C). Additional developmental western experiments show that BR-C proteins are present at low levels in late third instar epidermis, increase to maximal levels by 3 hours after puparium formation and decline to background levels by 9 hours after puparium formation (data not shown). The Z2 isoforms are the predominant BR-C proteins at all time points.
BR-C transcript analysis in the epidermis of mature third instar larvae. An exon map of BR-C, taken from DiBello et al. (1991) is shown at the top of the figure. The upstream half of BR-C, including the first exon, is not shown in the figure as indicated by the co-ordinates on top of the map. Translation of the various isoforms, Z1-Z4, begins in the common core exon. The predicted sizes of the amplification products are given beneath map. The location of the exon in which Z4 resides was kindly provided by Cindy Bayer. Total RNA was primed with oligo(dT) and reverse transcribed. The cDNA products were then amplified using the primer pairs shown below the exon map and visualized on a 2% agarose gel stained with ethidium bromide. M1, 1 kb ladder (Gibco BRL) and M2, ladder VII (Boehringer Mannheim). E, epidermal RNA; W, whole organism RNA.
BR-C transcript analysis in the epidermis of mature third instar larvae. An exon map of BR-C, taken from DiBello et al. (1991) is shown at the top of the figure. The upstream half of BR-C, including the first exon, is not shown in the figure as indicated by the co-ordinates on top of the map. Translation of the various isoforms, Z1-Z4, begins in the common core exon. The predicted sizes of the amplification products are given beneath map. The location of the exon in which Z4 resides was kindly provided by Cindy Bayer. Total RNA was primed with oligo(dT) and reverse transcribed. The cDNA products were then amplified using the primer pairs shown below the exon map and visualized on a 2% agarose gel stained with ethidium bromide. M1, 1 kb ladder (Gibco BRL) and M2, ladder VII (Boehringer Mannheim). E, epidermal RNA; W, whole organism RNA.
BR-C protein expression in the epidermis at puparium formation. (A) Extracts were made from wild-type whole organisms (W) or dissected epidermis (E). (B) Extracts were made from wildtype (br+) or mutant (br28) whole organisms. (C) Extracts were made from dissected wild-type (br+) or mutant (br28) epidermal preparations. A monoclonal antibody directed against the BRcore domain was used to probe western blots of the protein extracts. BRC isoform mobilities and identities are indicated on the left. The Z1 and Z3 proteins have been defined using isoform-specific monoclonal antibodies, while Z2-containing isoforms were defined by their absence in br28 animals (Emery et al., 1994). The p118 protein may represent a Z4 isoform. The position of a novel, apparently truncated, BR-C protein that accumulates in br28 animals (B and C) is indicated by an arrow. A protein of molecular mass 98×103, visualized on the Coomassie-stained gel after transfer, was used to show protein loading in the samples in B and C.
BR-C protein expression in the epidermis at puparium formation. (A) Extracts were made from wild-type whole organisms (W) or dissected epidermis (E). (B) Extracts were made from wildtype (br+) or mutant (br28) whole organisms. (C) Extracts were made from dissected wild-type (br+) or mutant (br28) epidermal preparations. A monoclonal antibody directed against the BRcore domain was used to probe western blots of the protein extracts. BRC isoform mobilities and identities are indicated on the left. The Z1 and Z3 proteins have been defined using isoform-specific monoclonal antibodies, while Z2-containing isoforms were defined by their absence in br28 animals (Emery et al., 1994). The p118 protein may represent a Z4 isoform. The position of a novel, apparently truncated, BR-C protein that accumulates in br28 animals (B and C) is indicated by an arrow. A protein of molecular mass 98×103, visualized on the Coomassie-stained gel after transfer, was used to show protein loading in the samples in B and C.
DISCUSSION
The first comprehensive intragenic complementation map of the BR-C (Belyaeva et al., 1980) indicated that rbp, br, 2Bc and 2Bd belonged to independent subgroups. Kiss et al. (1988) subsequently suggested that br and rbp be grouped together with 2Bd. However, both Guay and Guild (1991) and Karim et al. (1993) showed that whereas rbp+ function was essential for the transcription of the glue genes Sgs-3, Sgs-4 and Sgs-5, mutants in the br group had no effect on glue gene expression suggesting that the rbp and br functions do not overlap. This conclusion is confirmed by our data which show, conversely, that mutations in the br group strongly reduce DDC enzyme and mRNA levels (O’Keefe et al., 1995), whereas an rbp mutant is without effect.
The pronounced reduction in Ddc mRNA levels in npr17 is expected on the basis of its inability to complement any other BR-C mutant (Belyaeva et al., 1980). The situation with the 2Bab mutants is more complex. These mutants fail to complement only the br and rbp mutants according to Belyaeva et al. (1980) although Kiss et al. (1988) were inclined to group these mutants with the npr1 non-complementing class. Our data suggest that the 2Bab mutants are a heterogeneous group since only one of the four alleles that we examined had an effect on DDC activity (Table 1).
There is a strong correlation between the br+ function and the Z2 isoform because the br28 mutation is known to disruptthe Z2 zinc finger pair (DiBello et al., 1991; Schouls, 1993; Emery et al., 1994; C. Bayer, pers. comm.). The strongest Z2 protein-binding sites on the entire 7.6 kb genomic region known to be sufficient for correct developmental expression of Ddc (Scholnick et al., 1983; Chen and Hodgetts, 1987) are confined to a 217 bp region in the first intron (Fig. 3). The five AT-rich binding sites all share a common core sequence, CTAT, which is also present in the consensus binding site of Z2 to the Sgs-4 regulatory domain (von Kalm et al., 1994; Fig. 4B). The data of von Kalm et al. (1994) and Emery et al. (1994) suggest that rbp+ function acts through the Z1 domain to control Sgs-4 expression in the salivary glands and our data suggest that the br+ function acts through the Z2 domain to control Ddc expression in the epidermis. Since it is clear that rbp+ function (and by inference the Z1 isoform) controls additional genes expressed exclusively in the late larval and prepupal salivary gland (Guay and Guild, 1991; Karim et al., 1993), it will be interesting to determine whether the Z2 isoform also controls additional target genes in the epidermis. We hypothesise that the binding of a BR-C Z2 isoform is required for full activation of Ddc transcription and failure to produce wild-type levels of this isoform in br mutants explains the genetic interaction between the BR-C and Ddc loci reported here. The presence of the Z2 isoform transcript and the predominance of its protein in the epidermis (Fig. 5, 6) is consistent with this regulatory interaction. We have recently obtained in vivo evidence that confirms the significance of the interaction revealed by the binding studies. Epidermal Ddc transcript levels are about 5-fold lower in br5 mutant larvae than in wildtype organisms (O’Keefe et al., 1995). However, when high levels of Z2 protein are synthesized from a transgene under the control of a heat-shock promoter in a br5 mutant background, Ddc transcripts accumulate to normal levels. In addition, heatshock expression of Z2, but no other BR-C protein, rescues tanning in br5 prepupae (C. Bayer, LvK and J. Fristrom, unpublished). Since several mutants in the Ddc structural gene exhibit reduced levels of tanning of the puparium (Wright, 1987), this suggests the br5 mutant’s inability to tan is a specific effect of a failure to activate the Ddc gene. Despite the requirement for a Z2 product, we do not believe that induction of the BR-C is a sufficient condition for maximum Ddc transcript accumulation at pupariation. In the first place, none of the mutants eliminates epidermal Ddc expression entirely (Table 1). Secondly, we have shown that Ddc mRNA levels do increase substantially in the absence of protein synthesis at this stage (Clark et al., 1986). This suggests a role for a protein product, present before the rise in the late third instar ecdysone titer, in Ddc expression. It is possible that the BR-C Z2 isoform works in concert with other regulators to control Ddc expression. We note the presence of two potential ecdysone receptor response elements (EcREs, Cherbas et al., 1991) within the first intron of Ddc. One lies about 300 bp upstream of the first Z2binding site and the second lies between sites Z23,4 (Fig. 4A) which suggests that an interaction between the Z2 isoform and the ecdysone receptor complex could promote full induction of Ddc at pupariation. In the absence of the br+ function, the receptor complex alone might elicit limited Ddc induction. This could explain the lower levels of Ddc induction seen in naive epidermis following the addition of ecdysone and inhibitors of protein synthesis (Clark et al., 1986).
The existence of binding sites on the DraI-TaqI fragment for all four of the BR-C recombinant proteins is still consistent with our suggestion that the Z2 isoform mediates full Ddc induction at pupariation. The Z2-binding sites fall into three classes that can be distinguished by their overlap with the other BR-C-binding sites (Figs 3, 4A). The Z22,4 sites5 overlap with Z1-, Z3and Z4binding sites; sites Z23,5 overlap with Z1and Z4binding sites and Z21 overlaps with a Z3-binding site. Binding of the other BR-C isoforms members in close proximity to the Z2 sites could indicate that BR-C plays a role in repressing Ddc expression in tissues other than the epidermis. The occupancy state of the Z2 sites could be determined by the stochiometric ratios of the four BRC proteins and, in turn, regulate Ddc expression. Since a Z2 isoform is the most abundant species in the epidermis, it might occupy the Z2-binding sites causing the exclusion of some or all of the other BR-C proteins from these sites. In the salivary glands, an analogous situation might prevail at Sgs-4. Although all four BR-C isoforms can footprint on the critical element III of the regulatory region (von Kalm et al., 1994), only the Z1 isoform accumulates to substantial levels in this tissue (Emery et al., 1994 and unpublished data) suggesting that this isoform provides the rbp+ function by acting through element III to direct late larval expression of Sgs-4 (von Kalm et al., 1994).
The lack of any mutant effects on DDC activity at the molt into the third instar period indicates that BR-C does not play any direct role in the appearance of the DDC peak that occurs at this time (Kraminsky et al., 1980). This is not surprising given that the maximum ecdysone titer occurs nearly 12 hours before the molt (Kraminsky et al., 1980) and transcript levels of BR-C are very low at the second to third instar molt (Andres et al., 1993). In fact, Hiruma and Riddiford (1990) have shown that, in cultured epidermis from fourth instar larvae of Manduca sexta, Ddc expression required the addition of hormone and then its removal, a result similar to that reported for the Drosophila Ddc in cultured imaginal discs (Clark et al., 1986). We have suggested that Ddc induction in the larval epidermis involves at least two very different stage-specific mechanisms (Hodgetts et al., 1986). At pupariation, ecdysone induction of the metamorphic-specific BR-C leads to Ddc activation in the larval epidermis whereas at the end of embryogenesis, during the first two larval instars and following disc evagination, BR-C-independent mechanisms may lead to Ddc expression. A protein(s), induced by ecdysone, has recently been identified in Manduca sexta, whose presence appears to be responsible for repressing Ddc during non-metamorphic stages when ecdysone levels are high (Hiruma et al., 1995). Decay of this protein following the ecdysone peaks could lead to the delayed appearance of DDC that we have noted in Drosophila at stages other than pupariation (Kraminsky et al., 1980; Clark et al., 1986).
The lack of br mutant effects on DDC levels in the CNS of mature larvae, suggests an ecdysone-independent regulation of the complex spatial distribution of neural DDC (Beall and Hirsh, 1987; Konrad and Marsh, 1987). Recent evidence suggests that the CNS-specific splicing pattern (Fig. 1A) is a default pathway and that cis-acting sequences located within the first intron interact with an epidermal repressor to block the 3′-acceptor site of the neural-specific exon B (Shen and Hirsh, 1994). Although it is tempting to postulate that the BR-C encodes this repressor, the prediction that this would lead to an enrichment of the CNS splice form in the epidermis of br mutants is not substantiated by our analysis of Ddc transcripts using RT-PCR (O’Keefe et al., 1995). However, no systematic attempt was made in that study to assess the relative levels of the different splice forms and a confirmation of a role for the BR-C in Ddc splicing awaits further investigation.
Ddc induction clearly shares the major features of early-late gene induction in the salivary glands: induction occurs rapidly following a rise in ecdysone titer (Kraminsky et al., 1980) and full activation is dependent on BR-C function (Karim et al., 1993) and continued protein synthesis (Clark et al., 1986). Cascades of ecdysone-induced gene expression can be composed of two gene types. Genes encoding transcripton factors are used to propagate the regulatory cascade and genes encoding the effector proteins are used to carry out the developmental function of the cascade. A hormone signal could result in the ordered production of both regulatory and effector proteins at each step in the cascade (e.g., see Stone and Thummel, 1993; Natzle, 1993). Thus the late larval pulse of ecdysone activates a regulatory gene cascade of transcription factors composed of primary response early genes like the BRC followed by early-late genes like E78. The same hormone pulse can drive the sequential expression of effector genes. Unlike the transcription factor genes, effector gene transcription seems to exhibit some degree of tissue specificity. For example, both the IMP-E1 (Natzle, 1993) and the Ddc genes respond as early-late primary response genes in imaginal discs and in epidermis, respectively. Late genes are temporally even more distant from the hormone signal. In salivary glands, the L71 genes are completely dependent on primary response early genes for their activation (Guay and Guild, 1991; Karim et al., 1993). Our results and those of Karim et al. (1993) suggest a central role is played by the BR-C in mediating the tissuespecific gene cascades induced by ecdysone during metamorphosis.
ACKNOWLEDGEMENTS
This work was supported by a grant from the Natural Sciences and Engineering Research Council of Canada (to R. B. H.) and from the American Cancer Society (to G. M. G.). LvK was supported by a grant from the National Institutes of Health (US) to Dr James Fristrom. We are indebted to Dr Ron Cai Yang for conducting the statistical analyses and to Dr Cindy Bayer for communicating unpublished results.