The Drosophila 63F early puff contains E63-1, an ecdysone-inducible gene that encodes a novel Ca²⁺-binding protein

Steroid hormones control many aspects of the growth and development of higher organisms, including morphogenesis, tissue differentiation, reproductive functions and homeostasis. Although we know a great deal about how steroids regulate gene expression, virtually nothing is known about how these changes in gene activity direct specific developmental responses. In an effort to better understand the molecular mechanisms of steroid hormone action, we are studying the role of 20-hydroxyecdysone (hereafter referred to as ecdysone) in initiating the complete metamorphosis of the fruit fly, Drosophila melanogaster. A high-titer pulse of ecdysone at the end of larval development signals puparium formation, followed 10 hours later by a second hormone pulse that triggers head eversion and the prepupal-pupal transition (reviewed by Riddiford, 1993). The effects of ecdysone on gene expression that accompany these dramatic changes in morphology can be observed in the larval salivary gland, where they are manifested as changes in the puffing pattern of the giant polytene chromosomes.

When the polytene chromosomes are first large enough to examine, in mid-third instar larvae, only a few (<10) ‘intermolt’ puffs are present, representing the sites of glue gene transcription. These genes synthesize a glycoprotein glue that is used by the animal to affix itself to a solid surface for pupariation (reviewed by Meyerowitz et al., 1987). Many of these intermolt puffs regress following the high-titer late larval ecdysone pulse as a half dozen ‘early’ puffs are coordinately induced. The early puffs regress several hours later and a large set (>100) of ‘late’ puffs is induced coincident with the secretion of the glue into the lumen of the gland. The subsequent prepupal ecdysone pulse reinduces most of the early and late puffs as it triggers the histolysis of the larval gland and the development of the adult gland from a ring of imaginal progenitor cells (Clever, 1964; Ashburner et al., 1974; Richards, 1981a).

A series of classic studies by Ashburner and colleagues, using cultured salivary glands treated with ecdysone, led to a model for the genetic regulation of polytene chromosome puffing by ecdysone (Ashburner et al., 1974). According to this model, the ecdysone/receptor protein complex directly induces the genes contained within the early puffs and one or more of the early puff proteins is responsible for repressing the early puffs and inducing the late puffs.

Three early puff genes have been isolated and characterized in an effort to define this regulatory hierarchy at the molecular level: BR-C from the 2B5 puff (Chao and Guild, 1986; DiBello et al., 1991), E74 from the 74EF puff (Burtis et al., 1990), and E75 from the 75B puff (Segraves and Hogness, 1990). These genes are all long and complex (up to 100 kb), contain multiple ecdysone-inducible promoters, and encode site-specific DNA-binding proteins. Genetic and molecular studies of BR-C and E74 have demonstrated that they play essential roles during metamorphosis, as well as regulate the expression of late puffs and puff genes, consistent with the predictions of the Ashburner model (Belyaeva et al., 1980, 1989; Kiss et al., 1988; Guay and Guild, 1991; Fletcher et al., 1995; Fletcher and Thummel, 1995). Despite these advances in understanding the ecdysone regulatory hierarchy, the mechanisms by which the early puff genes transduce the ecdysone signal to direct the appropriate activity and morphogenesis of the salivary gland remain to be determined.

In this paper, we report the molecular characterization of a fourth early ecdysone-inducible puff locus, at 63F. This puff responds to ecdysone in a manner that is virtually indistinguishable from the 74EF and 75B early puffs, and is one of the half dozen early puffs originally described by Ashburner (1972, 1974). The 63F puff is rapidly induced by ecdysone, in the presence or absence of cycloheximide, and over a wide range of hormone concentrations. The 63F puff also has two periods of activity that correlate with the high-titer late larval and prepupal ecdysone pulses. Unlike the 2B5, 74EF and 75B early puffs, however, the 63F puff is not induced by ecdysone in the polytene chromosomes of the fat body (Richards, 1982). It was this tissue-specificity in the response of the 63F puff that motivated us to isolate the corresponding early gene(s). We hoped that a molecular analysis of the 63F puff might provide insights into the mechanisms by which the hormone directs tissue-specific developmental responses as well as a better understanding of how ecdysone directs the function and morphogenesis of the salivary gland.

We show here that 63F encompasses two ecdysoneinducible primary-response genes, E63-1 and E63-2, that parallel the puff in their transcriptional response to ecdysone. Both E63 genes are induced in the salivary glands by the late larval ecdysone pulse as well as in mass-isolated larval organs cultured with ecdysone in the presence or absence of cycloheximide. The two E63 genes also have complex spatial and temporal expression patterns, with E63-1 induction in late larvae being restricted to the salivary gland. Finally, unlike the previously studied early puff genes that encode DNA-binding proteins, E63-1 encodes a Ca²⁺-binding protein that is related to calmodulin. E63-1 provides a first link between the early events of the ecdysone response and a second-message mediated pathway in Drosophila.

A genomic phage clone that mapped to the distal end of the 63F puff, λ63F.4, was kindly provided by Ken Burtis. A chromosomal walk was initiated using this clone to screen the ISO-1 genomic cosmid library (kindly provided by John Tamkun). Eighty kb of genomic DNA was represented by four overlapping cosmids and mapped with the restriction enzymes BamHI, EcoRI, and HpaI (Fig. 1). The progress of the walk was monitored by in situ hybridization to polytene chromosomes using the Genius DNA Labeling/Detection kit (Boehringer Mannhiem) as described in Chen et al. (1992). Five to 15 kb fragments of the chromosomal walk were subcloned into Bluescript KS-vectors (Stratagene) to facilitate the synthesis of RNA probes for northern blot hybridizations.

The northern blots used for the screen presented in Fig. 3 were prepared from RNA samples isolated by Pat Hurban. The RNA was isolated from third instar larval tissues treated with either 7×10⁻⁵ M cycloheximide alone or 7×10⁻⁵ M cycloheximide and 5.5×10⁻⁶ M ecdysone (Hurban and Thummel, 1993). Equivalent amounts of RNA were fractionated on 1% agarose/formaldehyde gels, blotted onto GeneScreen nylon (NEN DuPont) and hybridized with RNA probes, essentially as described by Andres and Cherbas (1992). The blots presented in Fig. 4 were prepared as described by Karim and Thummel (1991) and hybridized with RNA probes. The blots presented in Fig. 5A were prepared from RNA isolated from staged animals described by Andres and Cherbas (1992). The RNAs used for Fig. 5B were prepared from dissected salivary glands isolated from larvae that were staged using blue food. Bromophenol blue was added to the food at a final concentration of 0.05%. Since the animal’s exposure to the late larval ecdysone pulse affects it’s feeding behavior, the larva’s exposure to the hormone can be monitored according to the amount of undigested food remaining in the gut (Maroni and Stamey, 1983). Three different late third instar larval stages can be distinguished using this technique; these animals are, on average, 18, 8, and 4 hours prior to pupariation (Andres and Thummel, 1994).

Animals were staged according to their exposure to the late larval pulse of ecdysone by using the blue food technique, allowing an approximate Puff Stage designation (Andres and Cherbas, 1992; Andres and Thummel, 1994). Frozen sections of larvae were prepared, hybridized, and autoradiographed as described in Andres and Cherbas (1992) using ³⁵S-labeled antisense RNA probes generated from an E63-1C specific cDNA.

A directionally cloned cDNA library (λHorIII) prepared from RNA isolated from larval organs cultured in the presence of cycloheximide and ecdysone was kindly provided by Pat Hurban. Genomic fragments of the p63F.20 subclone (Fig. 1), which detected the four ecdysoneinducible RNAs (Fig. 3), were used to screen for E63 cDNA clones. Full-length cDNAs were isolated from this library corresponding to each of the four E63 mRNAs.

Each cDNA clone was sequenced by generating progressive 5′ and 3′ deletions using the Erase-A-Base kit (Promega). Approximately 300 bases from the end of each successive clone was directly sequenced from the double-stranded templates using the Sequenase kit (Stratagene) and either the −40 or reverse primers (NEB), until overlapping sequence was generated on both strands. Genomic sequence corresponding to the exon/intron boundaries was determined in a similar manner from exon-specific genomic subclones. In some instances, gene-specific primers were generated and used to sequence directly across boundaries in the genomic DNA.

The start sites for each of the E63 mRNAs were identified by anchored PCR using the 5′ RACE kit (Promega), the primers described in Stone and Thummel (1993), and the following E63- specific nested primers: E63-1 P1=TTCAACGATCGTCTCGC; E63-1 P2=CGTCGCCTGCGTATTTA; E63-1 P3=TTGTCACTGG-GCACAGT; E63-2 P1=TGGATGCTCTTGTTCCC; E63-2 P2= AGAAGCCATCCCGAACTG; E63-2 P3=AGCTGATAATGAGC-TCG. The 5′ RACE products extended 15 and 9 nucleotides (nt) upstream from the longest E63-1 and E63-2 cDNAs, respectively.

Ca²⁺-binding assays were performed essentially as described by Maune et al. (1992). Labeled proteins were generated from cDNA clones of E63-1, Drosophila CaM (kindly provided by Kathy Beck-ingham), and USP, using the Promega Coupled Reticulocyte Lysate kit. Labeled proteins were prepared from 50 μl reactions and dialyzed against binding buffer (20 mM Tris-HCl, pH 7.4, 50 mM KCl, 3 mM MgCl₂, 1 mM DTT) for 8 hours at 4°C to remove any contaminating Ca2+ or Ca2+ chelators. EGTA or CaCl₂ at a final concentration of 10 mM was then added to 3 μl of labeled protein, and these samples were electrophoresed on 12.5% polyacrylamide, 0.1% SDS gels along with ¹⁴C-labeled protein standards (BRL). Gels were dried and exposed to Hyperfilm-MP (Amersham) X-ray film.

A phage insert that mapped to the distal end of the 63F puff provided the starting point for a chromosomal walk that eventually spanned 80 kb of genomic DNA (Fig. 1). In situ hybridization to polytene chromosomes using probes derived from both ends of the walk confirmed that most of the DNA from the 63F puff locus had been isolated (Fig. 2).

Because a differential screen using cDNA probes from hormone-induced and uninduced tissues failed to reveal any potential ecdysone-inducible genes within the 63F puff locus, we chose a more sensitive method to achieve this goal. Restriction fragments representing the entire walk were subcloned into plasmid vectors containing promoters for T3 and T7 RNA polymerase (Fig. 1), allowing us to synthesize high specificactivity RNA probes from both strands of the cloned genomic DNA. These probes were hybridized to northern blots containing RNA from mass-isolated third instar larval organs cultured either with cycloheximide alone or with cycloheximide and ecdysone. Using this approach, we identified six ecdysone-inducible transcripts encoded within the 63F region, as well as the non-inducible transcript for polyubiquitin (Ubi-p) – the only previously identified product from the 63F locus (Izquierdo et al., 1981; Lee et al., 1988) (Fig. 3).

A secondary screen by northern blot analysis, in which we analyzed RNA isolated from control and ecdysone-treated salivary glands as opposed to total organs, revealed that only four of the six transcripts are inducible in this tissue (data not shown). This result suggested that the 6.0 kb and the 7.0 kb transcripts shown in Fig. 3 are expressed in other tissues and thus could not be responsible for the 63F salivary gland puff. Interestingly, the four remaining ecdysone-inducible transcripts could be detected by a single genomic restriction fragment, contained within p63F.20 (Fig. 3). Three of these transcripts were detected with the T3-generated probe, while one transcript was detected with the T7-generated probe. Subsequent sequence analysis revealed that the three mRNAs detected by the T3-generated probe are derived from a single gene, designated E63-1. The three overlapping E63-1 transcripts are 2.8, 1.7 and 1.1 kb in length, and are designated E63-1A, E63-1B and E63-1C, respectively. E63-2 is encoded on the opposite strand of DNA and produces a single 1.2 kb transcript.

We next determined whether E63 transcription is induced by ecdysone as a primary- or secondary-response. Mass-isolated larval organs were cultured with either ecdysone alone, or ecdysone and cycloheximide, and RNA was isolated at 2 hour intervals and analyzed by northern blot hybridization (Fig. 4). As has been shown previously, E74A is induced by ecdysone under these conditions in either the presence or absence of protein synthesis (Karim and Thummel, 1991). E63-2 displays an identical response to the hormone, indicating that it is induced as a primary-response. The E63-1 transcripts are also induced in the presence or absence of cycloheximide, with similar kinetics to E74A. However, E63-1 induction occurs above a previously existing low level of E63-1 mRNA that is normally present in late third instar larvae (Fig. 5A).

There are six or more ecdysone pulses during Drosophila development, at least one during each stage of the life cycle (reviewed by Richards, 1981b; Andres et al., 1993; Riddiford, 1993). In order to determine how the E63 genes might respond to these hormone pulses, we analyzed their temporal profiles of transcription throughout the Drosophila life cycle. RNA samples were isolated from animals synchronized at egg deposition, hatching and puparium formation, and analyzed by northern blot hybridization (Fig. 5A). These blots reveal several interesting aspects of E63-1 gene regulation. First, only E63-1A and E63-1B transcripts are detectable, indicating that the level of E63-1C mRNA accumulation is significantly lower, possibly reflecting its transcription in a small subset of tissues. E63-1A mRNA first becomes detectable during the last half of embryogenesis, drops to low levels during most of the larval period, and accumulates again prior to pupariation, coincident with the high-titer pulse of ecdysone that occurs at this stage (De Reggi et al., 1975; Hodgetts et al., 1977; Berreur et al., 1979). The levels of E63-1A mRNA then decline following pupariation, but increase during prepupal development as the ecdysone titer rises. This is followed by an accumulation of low levels of E63-1A mRNA in early pupae and high levels of transcription in mid-pupae, coincident with the broad peak in ecdysone titer (Handler, 1982; Pak and Gilbert, 1987). E63-1B transcription closely parallels that of E63-1A, with two exceptions. E63-1B transcripts are present in early embryos, before E63-1A, and the levels of E63-1B remain high throughout larval development when E63-1A mRNA levels are very low. An explanation for this latter difference arises from the structure of the E63-1 gene (see below). The E63-1A mRNA differs from that of E63-1B in that it contains a longer 3′ untranslated region. These unique 3′ sequences could lead to E63-1A mRNA being less stable than that of E63-1B during larval development. Alternatively, the E63-1B mRNA may be transcribed in a tissue that does not recognize the E63-1A poly(A) addition site during larval development. These same mechanisms may explain why the E63-1C transcript, with its nested poly(A) addition site, is not detectable in RNA samples isolated from whole animals.

E63-2 displays a temporal pattern of transcription that is different from that of either E63-1A or E63-1B (Fig. 5A). E63-2 mRNA can be detected in early embryos, perhaps reflecting a maternal contribution, and is induced to higher levels in late embryos, coincident with E63-1 induction. E63-2 then has several brief periods of transcript accumulation – in mid-first and mid-second instar larvae, either before or during the ecdysone pulses that trigger molting of the larval cuticle, in mid-prepupae when the ecdysone titer is low, and in mid-pupae when E63-1 transcripts are abundant (De Reggi et al., 1975; Hodgetts et al., 1977; Handler, 1982; Pak and Gilbert, 1987).

At least part of this apparent complexity in E63 transcription could arise from distinct temporal responses in different tissues. Accordingly, we examined the temporal profiles of E63-1 and E63-2 transcription in staged larval salivary glands, to compare this pattern with that seen in whole animals. RNA was extracted from salivary glands dissected from staged late third instar larvae and analyzed by northern blot hybridization (Fig. 5B). As expected, E74B transcription is repressed and E74A is induced by the high-titer ecdysone pulse that occurs approximately 8 hours prior to puparium formation, confirming that the glands were accurately staged (Karim and Thummel, 1991; Huet et al., 1993). All three E63-1 mRNA isoforms and the E63-2 transcript are also induced, coincident with the induction of the 63F puff and E74A mRNA. This induction can be recapitulated in explanted salivary glands cultured with ecdysone alone, or with ecdysone and cycloheximide, indicating that all four E63 mRNAs are induced as a primary-response in this tissue (data not shown). The induction of E63-1A and E63-1B transcription in staged salivary glands parallels the responses seen in whole animals, whereas only a slight induction of E63-2 transcription can be detected in whole late third instar larvae (Fig. 5A). The increased levels of E63-2 RNA accumulation in salivary glands relative to whole animals suggests that this gene responds to ecdysone in a tissue-specific manner.

To define more precisely which tissues express E63-1, and whether expression in those tissues can be correlated with ecdysone exposure, we performed an in situ hybridization on sectioned late third instar larvae before (Puff Stage 1) and after (Puff Stage 6) the high-titer pulse of ecdysone. A common region probe was used that detects all three E63-1 transcripts (Fig. 6). The sensitivity of this technique did not allow us to detect E63-2 mRNA.

By comparing the expression patterns of E63-1 in larvae before and after the late larval ecdysone pulse, it is clear that the salivary gland is the only tissue in which we can detect E63-1 induction. Low levels of E63-1 mRNA are detected in the fat body, brain/ventral ganglion, epidermis and imaginal discs, and a moderate level of mRNA is detected in the midgut. However, E63-1 transcript levels in these tissues do not increase significantly in Puff Stage 6 animals, after the rise in ecdysone titer, as they do in salivary glands (Fig. 6 and data not shown).

Small genomic probes from the p63F.20 subclone (Fig. 1) were used to screen cDNA libraries prepared from cultured late larval organs and staged third instar larvae. Full-length cDNAs were isolated for all four E63 transcripts and their nucleotide sequences were determined. The potential transcription start site for each E63 mRNA was identified by using 5′ rapid amplification of cDNA ends (5′ RACE; Frohman, 1990). Intron/exon boundaries were identified by comparing the cDNA sequences with those of genomic subclones.

The gene structures for E63-1 and E63-2 are represented in Fig. 1. The two genes are divergently transcribed and separated by a 1.1 kb intergenic region. E63-1 spans 23 kb of DNA and consists of seven exons. All of the intron/exon boundaries conform well to the consensus splice donor and acceptor sequences for Drosophila introns (Mount, 1982; Fig. 7A). The sizes of the E63-1A, E63-1B, and E63-1C mRNAs are 2872 nt, 1694 nt and 1088 nt, respectively. These transcripts are identical at their 5′ ends and have 3′ ends defined by three nested poly(A) addition sites.

E63-2 is 1254 nt in length and contains no introns. The longest AUG-initiated open reading frame (ORF) in E63-2 begins 6 nt downstream from the 5′ end of the mRNA and encodes an 88 amino acid protein. There are also 5 other AUG-initiated ORFs in the E63-2 mRNA, between 21 and 43 codons in length. The 43 codon ORF, however, continues a long distance upstream from the AUG to a potential GUG initiation codon. This GUG-initiated ORF encodes a 111 amino acid protein with a predicted mass of 12.9×10³ (Fig. 7B). The use of CUG and GUG initiation codons has been documented for other Drosophila transcripts (Sugihara et al., 1990; Desimone and White, 1993) including the early ecdysone-inducible E74A mRNA (Boyd and Thummel, 1993) and possibly the E78B mRNA from the 78C early late puff (Stone and Thummel, 1993). A search of the protein and nucleic acid databases failed to reveal any significant similarities with E63-2 sequences, and no protein was synthesized from a full-length E63-2 mRNA in a rabbit reticulocyte lysate (data not shown). Thus it remains to be determined whether E63-2 encodes a functional protein product.

E63-1 appears more conventional in its structure. This gene contains a single long AUG-initiated ORF that is preceded by stop codons in all three reading frames. The E63-1 ORF encodes a 193 amino acid protein with a predicted mass of 22.0×10³. A protein of this size can be synthesized from a full-length E63-1C mRNA in a rabbit reticulocyte lysate (Fig. 9). A search of the databases revealed that the E63-1 protein has significant sequence identities with the family of Ca²⁺-binding proteins related to calmodulin (CaM), with approx. 35% identity to CaMs isolated from different species. As expected, maximal conservation resides within the two pairs of EF hands that form the Ca²⁺-binding domains. The standard EF hand consists of 29 amino acids arranged as two α-helices surrounding a Ca²⁺-binding loop (first described by Kretsinger, 1975). The first pair of EF hands (from residues 39-103, Fig. 7A) is most similar to a Schistosoma CaM-related protein (49% identity) and is 40% identical to Drosophila CaM. The second pair of EF hands (from residues 131-193, Fig. 7A) is most similar to Chlamydomonas CaM (57%) and is 46% identical to Drosophila CaM. This sequence similarity with CaM extends to the structure of the E63-1 gene itself. Two of the three exons within the coding region of Drosophila CaM are conserved in E63-1. The first intron, which immediately follows the CaM initiation codon, is missing from E63-1. However, the intron that interrupts the coding region for the second Ca²⁺-binding loop is precisely conserved in E63-1 while the intron in the region that encodes the fourth Ca²⁺-binding loop is displaced by one nucleotide (Smith et al., 1987).

Like CaM, the E63-1 protein contains four EF hand Ca²⁺-binding domains that are grouped together in two pairs separated by a central axis. Fig. 8 presents the structure and consensus sequence for the standard EF hand, and compares the E63-1 and Drosophila CaM EF-hand sequences. The most significant differences between the E63-1 and consensus EF hand sequences include the following: the first α-helix of each loop does not begin with a glutamic acid, serines occur at position 10 and 12 and a leucine at position 16 in loop II, and the second α-helix of loop IV is truncated.

To determine whether E63-1 encodes a Ca²⁺-binding protein of the EF hand family, we tested its ability to bind Ca²⁺ in vitro, by a gel shift assay. Labeled E63-1 protein was prepared by in vitro transcription of an E63-1C cDNA followed by in vitro translation in a rabbit reticulocyte lysate supplemented with ³⁵S-methionine. The resultant radiolabeled E63-1 protein was incubated either without Ca²⁺ (in the presence of EGTA) or with CaCl2 and electrophoresed on an SDS polyacrylamide gel. As shown in Fig. 9, both Drosophila CaM and E63-1 migrate with an increased mobility in the presence of Ca²⁺, as has been reported for other EF hand Ca²⁺-binding proteins (Geiser et al., 1991; Maune et al., 1992). USP, which is a DNA-binding protein, was included as negative control.

We describe here our initial molecular characterization of the 63F puff locus, one of the half-dozen early ecdysone-inducible puffs, described by Ashburner, in the salivary gland polytene chromosomes (Ashburner, 1972, 1974). The 63F puff contains a pair of divergently transcribed genes, E63-1 and E63-2. Both E63 genes are induced directly by ecdysone in cultured larval organs, but display complex spatial and temporal regulation during development. Surprisingly, E63-1 is unique among the early puff genes in that it encodes a Ca²⁺-binding protein similar to calmodulin, with four EF hand Ca²⁺-binding domains. E63-2 mRNA contains only two relatively short ORFs, and neither of the predicted proteins has any matches in the databases. Below we discuss the complex regulation of the E63 genes and propose possible functions for the ecdysoneinducible E63-1 Ca²⁺-binding protein.

One of the central questions of developmental biology is how a single systemic signal can program diverse tissue-specific developmental responses. Our studies of E63-1 suggest that this gene functions in a tissue-specific manner during the onset of metamorphosis. Indeed, our initial interest in the E63 genes arose from the apparent tissue-specificity of the 63F early puff. The well-studied early puffs at 74EF and 75B are induced by ecdysone in both the salivary glands and fat body of late third instar larvae. In contrast, the 63F early puff is selectively induced by ecdysone in the salivary glands, but this puff does not form in the fat body polytene chromosomes (Richards, 1982). Our molecular studies reveal that E63-1 displays a similar tissue-specificity in its transcriptional induction by ecdysone. Examination of the spatial pattern of E63-1 transcription in staged late third instar larvae reveals a low level of RNA in the brain, epidermis, imaginal discs and fat body, and moderate levels of transcript in the midgut. These levels remain relatively unchanged following the high-titer late larval ecdysone pulse, while high levels of E63-1 mRNA accumulate in the salivary gland (Fig. 6).

This remarkable tissue-specificity of E63-1 induction stands in sharp contrast to the responses of the three previously characterized early puff genes, BR-C, E74 and E75. These genes are widely expressed in late third instar larvae, in both larval and imaginal tissues (Chao and Guild, 1986; Segraves, 1988; Boyd et al., 1991; Huet et al., 1993; Emery et al., 1994). Yet, mutational analysis of the BR-C and E74 reveal that these genes perform various tissue-specific functions during metamorphosis (Kiss et al., 1988; Restifo and White, 1992; Fletcher et al., 1995). In addition, the glue genes and 71E late genes that are regulated by the BR-C and E74 (Guay and Guild, 1991; Karim et al., 1993; von Kalm et al., 1994; Fletcher and Thummel, 1995) are expressed primarily in a single tissue at a single time during development (Restifo and Guild, 1986; Barnett et al., 1990). Clearly, there must be mechanisms by which the widespread expression patterns of the early genes are refined to result in each target tissue acquiring their appropriate morphological and functional properties during metamorphosis.

One means by which these developmental pathways might be refined is through the activities of tissue-specific primaryresponse genes. A variety of these genes have been identified, none of which encode DNA-binding regulatory proteins (reviewed by Andres and Thummel, 1992). The IMP genes and the small heat shock genes are induced by ecdysone primarily in the imaginal discs (Ireland et al., 1982; Natzle et al., 1986), Fbp1 is induced in the larval fat body (Lepesant et al., 1978) and the Eip genes are induced primarily in blood cells (Andres and Cherbas, 1992).

E63-1 and E63-2 appear most similar to this latter class of primary-response genes. Yet, unlike the members of this family that have been described to date, E63-1 may function in a regulatory capacity due to its Ca²⁺-binding activity. In this sense, E63-1 could fulfil the predictions of the Ashburner model that early puff loci function as regulators that can control downstream developmental events, although this effect would not be manifested at the level of direct DNA-binding. Below we speculate on the possible functions that E63-1 may perform during the early stages of metamorphosis.

To date, more than 200 genes have been identified that encode Ca²⁺-binding proteins, including several from Drosophila. Many are tissue-specific in their expression, and the functions of only a few are known (reviewed by Heizmann and Hunziker, 1991; Beckingham, 1995). Perhaps the best understood Ca²⁺ receptor is calmodulin (CaM), a multifunctional protein that is highly conserved from yeast to humans. Using mechanisms that are quite varied it exerts a wide range of regulatory functions. When activated by Ca²⁺, CaM has been shown to interact with protein kinases and phosphodiesterases, membrane transport molecules and ion pumps, cytoskeletal elements and transcription factors (reviewed by Stoclet et al., 1987; Means et al., 1991).

The sequence similarity between E63-1 and CaM suggest that, like CaM, E63-1 has no enzymatic activity per se, but rather exerts its effects by regulating Ca²⁺-mediated signaling cascades. Furthermore, the restricted induction of E63-1 in the larval salivary gland suggests that this gene may play a role in the steroid-controlled activities of the gland, notably glue secretion and histolysis. Consistent with this proposal, both secretion and programmed cell death are controlled in other organisms by catalytic cascades regulated by Ca²⁺ ions (reviewed by Williamson and Monck, 1989).

The high-titer ecdysone pulse at the end of Drosophila larval development triggers the salivary gland cells to secrete a glycoprotein glue into the lumen of the gland (Poels et al., 1971, 1972; Zhimulev and Kolesnikov, 1975; Boyd and Ashburner, 1977). In D. virilis, the salivary gland swells immediately before the glue is secreted, most likely due to increases in water and Ca²⁺ from extracellular sources (Kress, 1974). Furthermore, Boyd and Ashburner (1977) found that the initiation of glue secretion in D. melanogaster salivary glands occurs approx. 3 hours after exposure to ecdysone and is dependent on RNA and protein synthesis. The requirements for this response led them to propose that one or more of the proteins encoded by the early puffs may play a direct role in the initiation of glue secretion into the lumen. The regulation and Ca²⁺-binding activity of E63-1 raise the interesting possibility that it could be a key player in this process.

One possible molecular mechanism by which E63-1 could exert its regulatory effects is by directly or indirectly modulating the activities of the early ecdysone-induced DNA-binding proteins. Substantial evidence indicates that Ca²⁺-binding proteins can regulate the activities of mammalian transcription factors and steroid hormone receptors (Hunter and Karin, 1992; Burns et al., 1994; Corneliussen et al., 1994; Dedhar et al., 1994; Sun et al., 1994). Similar Ca²⁺-dependent interactions and/or modifications of the early ecdysone-induced transcription factors, mediated by E63-1, could dramatically affect their regulatory functions and thus direct distinct developmental pathways in a tissue- or stage-specific manner.

These effects of Ca²⁺ on transcriptional regulation remind us of a hypothesis put forth by Kroeger and his colleagues some 30 years ago. They proposed that ecdysone exerts its effects on puffing in the polytene chromosomes by altering the intracellular cation concentrations, and that it was these changes in the cellular electrolyte metabolism that resulted in specific gene activation. Numerous detailed experiments were performed that were aimed at defining the sensitivity of different chromosomal puffs to shifts in cation concentration (primarily Na⁺, K⁺, and Mg²⁺, but also including Ca²⁺) (Kroeger, 1977, and references therein). Not surprisingly, this hypothesis generated significant debate (Ashburner and Cherbas, 1976; Kroeger, 1977), and has gathered little supporting evidence as the regulatory functions of the ecdysone receptor have been elucidated (Koelle, 1992; Yao et al., 1992). Nevertheless, the now well-known effects of Ca²⁺ on transcription factors, via specific Ca²⁺/CaM protein kinases, have clearly demonstrated that cations can have profound effects at the level of transcriptional initiation. As described above, it is possible that future studies of E63-1 function may demonstrate how changes in cation concentration may alter the pattern of transcription puffs in the polytene chromosomes and thus provide a new opportunity to reevaluate Kroeger’s data and interpret his results in the context of our current understanding of the mechanisms of transcriptional regulation.

Clearly, our understanding of E63 function requires a mutational analysis of these genes, an effort that is currently underway. Studies of salivary gland function and development in E63-1 mutant animals will allow us to determine what role, if any, this gene plays in glue secretion and histolysis. Similarly, an examination of the puffing patterns in the polytene chromosomes and the transcription of ecdysoneregulated genes in E63-1 mutants will allow us to determine whether E63-1 exerts its effects through the early DNA-binding proteins. Mutations in E63-2 may also indicate whether this gene encodes a functional protein product. These studies hold the promise of providing significant new insights into the hormonal regulation of salivary gland development, as well as the mechanisms by which Ca²⁺- and steroid-signaling pathways may converge to control stage and tissue-specific developmental pathways during metamorphosis.

We thank Ken Burtis for supplying λ63F.4, the starting point for the chromosomal walk; John Tamkun for supplying the ISO-1 genomic cosmid library; Pat Hurban for supplying RNA samples and the λHorIII cDNA library; Felix Karim for the RNA blots presented in Fig. 4, and Kathy Beckingham for supplying the Drosophila CaM clone. We also thank Pam Reid for her help with DNA sequencing and Eric Baehrecke, Mike Horner, Jenn Fletcher, Dan Cimbora, Brent Bisgrove and Kathy Beckingham for critical comments on the manuscript. A. J. A. is an Associate and C. S. T. is an Associate Investigator with the Howard Hughes Medical Institute.