Genes Expressed in the Ring Gland, the Major Endocrine Organ of Drosophila melanogaster
Peter D. Harvie, Maria Filippova, Peter J. Bryant


We have used an enhancer-trap approach to begin characterizing the function of the Drosophila endocrine system during larval development. Five hundred and ten different lethal PZ element insertions were screened to identify those in which a reporter gene within the P element showed strong expression in part or all of the ring gland, the major site of production and release of developmental hormones, and which had a mutant phenotype consistent with an endocrine defect. Nine strong candidate genes were identified in this screen, and eight of these are expressed in the lateral cells of the ring gland that produce ecdysteroid molting hormone (EC). We have confirmed that the genes detected by these enhancer traps are expressed in patterns similar to those detected by the reporter gene. Two of the genes encode proteins, protein kinase A and calmodulin, that have previously been implicated in the signaling pathway leading to EC synthesis and release in other insects. A third gene product, the translational elongation factor EF-1α F1, could play a role in the translational regulation of EC production. The screen also identified the genes couch potato and tramtrack, previously known from their roles in peripheral nervous system development, as being expressed in the ring gland. One enhancer trap revealed expression of the gene encoding the C subunit of vacuolar ATPase (V-ATPase) in the medial cells of the ring gland, which produce the juvenile hormone that controls progression through developmental stages. This could reveal a function of V-ATPase in the response of this part of the ring gland to adenotropic neuropeptides. However, the gene identified by this enhancer trap is ubiquitously expressed, suggesting that the enhancer trap is detecting only a subset of its control elements. The results show that the enhancer trap approach can be a productive way of exploring tissue-specific genetic functions in Drosophila.

IN insects, as in all animals, many aspects of development are under hormonal control. The most important insect hormones are the ecdysteroid molting hormone (EC), which is secreted from the prothoracic glands and the sesquiterpenoid juvenile hormone (JH), which is secreted from the corpus allatum (CA). The generally accepted view is that ecdysteroid peaks determine the time of molting from one instar to the next, whereas JH levels determine whether the animal molts to a larval, pupal, or adult form.

The details of the neuroendocrine control of insect development have been best characterized in lepidopterans. Many, but not all, of the biosynthetic steps and intermediates leading from dietary cholesterol to the biologically active EC 20-hydroxyecdysone (20-HE) have been identified (Grieneisen 1994). EC receptors and numerous EC-responsive genes have been identified, and progress is being made in understanding the molecular nature of the EC response (Cherbas and Cherbas 1996). One EC peak early in the last larval instar (the “commitment peak”) apparently causes the epidermis to become committed to producing either larval or pupal cuticle at the next molt, whereas a later, sharper EC peak (the “prepupal” or “molting peak”) is responsible for initiating the molt itself (Riddiford 1978). Five JHs have been identified in lepidopterans, and shown to be sesquiterpenoids that are synthesized from acetate and/or propionate in the CA (Gilbertet al. 1996). The main form in Drosophila is methyl 6,7;10,11-bisepoxy-3,7,11-trimethyl-(2E)-dodecenoate (JHB3; Richardet al. 1989).

The levels of EC and JH are regulated by adenotropic neuropeptides that are produced in the developing brain and delivered to the endocrine glands via the axons of neurosecretory cells (Gilbertet al. 1996). Some of the neuropeptides (large and small prothoracicotropic hormone or PTTH) stimulate EC production by the prothoracic glands, whereas others either stimulate [allatotropic hormone or allatotropin (ATH)] or inhibit [allatostatic hormone or allatostatin (ASH)] JH production by the CA. Immunohistochemical studies in both lepidopterans (Mizoguchiet al. 1987) and Drosophila (Zitnanet al. 1993) show that each of these peptides is produced by a small number of neurosecretory cells located in defined positions of the developing brain.

The signaling pathways leading from the adenotropic neuropeptides to the synthesis and release of EC and JH have been investigated most extensively in the tobacco hornworm Manduca sexta. For the commitment peak of EC occurring during mid-fifth larval instar, PTTH appears to act via a Ca2+/calmodulin-dependent cAMP pathway leading to the phosphorylation of a specific set of proteins including ribosomal protein S6 and β-tubulin (Smith 1995; Gilbertet al. 1996). The mechanisms involved in the stimulation of molting and metamorphosis by the later, larger peak of EC are not yet clear. In studies of the control of JH production by the CA, Reagan et al. (1992) showed that ATH induces phosphoinositide hydrolysis and that inhibition of Ca2+-ATPase, protein kinases A and C, and ATP-dependent Ca2+ sequestration inhibited production of the hormone. These results suggest that the inositol 1,4,5-triphosphate pathway may be involved in the response to ATH and possibly other neuropeptides.

The difficulty of genetic analysis in lepidopterans has restricted the possibilities for functional analyses of the neuroendocrine pathways controlling the development of these insects. However, similar neuroendocrine mechanisms appear to operate in dipterans including Drosophila, an insect for which excellent molecular and genetic techniques have been developed. In higher dipterans, the larval prothoracic glands, CA, and corpus cardiacum are fused into a single compound structure, the ring gland (Kinget al. 1966) which has been shown experimentally to produce both EC (Robertset al. 1984) and JH (Richardet al. 1989). Based mainly on morphological homologies between the ring gland and the endocrine glands of larger insects, it is thought that EC is produced by the large lateral cells (LC) homologous to the prothoracic glands of other insects and JH is produced by the smaller medial cells (MC) homologous to the CA. Ventral ganglion/brain complexes of wandering third-instar Drosophila larvae show PTTH activity in vitro (Henrichet al. 1987) and two forms of PTTH separable by molecular weight have been isolated from late third-instar brains (Paket al. 1992). A 66-kD glycoprotein with PTTH activity on ring glands in vitro has been purified from Drosophila larvae but it has no significant homology to known PTTH peptides (Kimet al. 1997). Furthermore, Drosophila late-larval brain extract stimulates EC production from M. sexta prothoracic glands, and partially purified large and small M. sexta PTTH stimulate EC production from Drosophila ring glands (Henrich 1995). Antibodies made against B. mori large and small PTTH stain Drosophila neurosecretory cells in a spatiotemporal manner similar to that seen in lepidopterans (Zitnanet al. 1993). Antibodies to M. sexta ASH and ATH also cross-react in specific neurosecretory cells of larval fly brains but the correlation to lepidopteran patterns is less clear. An ASH activity also has been described from the Drosophila brain (Richardet al. 1989; Richardet al. 1990). Taken together, these results strongly suggest a molecular conservation of basic neuroendocrine processes between moths and flies, with some order-specific modifications.

In the present study, we have used the “enhancertrap” technique (O'Kane and Gehring 1987; Grossniklauset al. 1989; Wilsonet al. 1989; Bellenet al. 1989) to identify Drosophila genes that are strongly expressed in all or part of the ring gland during development. With this method, a reporter gene encoding β-galactosidase under the control of a weak promoter, namely the P transposase promoter, is carried to various genomic sites by a transposable element. When the reporter gene shows expression in the ring gland, it is probably inserted into or near a gene with a similar expression pattern (Wilsonet al. 1989; Bieret al. 1989). We have restricted our attention to lines in which the insertion is lethal so that analysis of the terminal phenotype might provide clues to the disrupted gene's function. Similar techniques have been used to detect genes expressed in other tissues and/or during specific developmental stages (e.g., Mlodzik and Hiromi 1992), but the ring gland has not been studied in previous screens. By focusing on the ring gland and screening for phenotypes suggestive of neuroendocrine defects, we hope to identify genes involved in the signaling or biosynthetic pathways downstream of the adenotropic hormones as well as those required for the development of the ring gland.


Drosophila stocks: Lethal PZ insertional mutations (Mlodzik and Hiromi 1992) were generated in the laboratory of Dr. Allan Spradling (Karpen and Spradling 1992) and provided to us by Dr. John Tower at the University of Southern California (Los Angeles). Fly cultures were maintained on standard cornmeal-molasses-agar medium at 25° unless otherwise stated. Mutant chromosomes were balanced over CyO or In(2LR)Gla, Gla Bc Elp (second-chromosome lines) or In(3LR) TM3 y+ ri p psep bx34e es (third-chromosome lines).

5-bromo-4-chloro-3-indoxyl-β-d-galactopyranoside (X-Gal) staining: Tissue was dissected in Ringer solution or phosphate-buffered saline (PBS), fixed in 1% glutaraldehyde for 15–30 min at room temperature, and incubated in 2% X-Gal (Gold Biotechnology, St. Louis, MO) dissolved in N,N′-dimethylformamide diluted 1:10 in staining buffer (10 mm sodium phosphate, 3.1 mm ferropotassium cyanide, 3.1 mm ferripotassium cyanide, 150 mm sodium chloride, 1 mm magnesium chloride) for up to 12 hr at room temperature.

Antibody staining: Tissue was dissected in Ringer solution or PBS, fixed in 4% paraformaldehyde for 20 min at room temperature and incubated in a monoclonal antibody to β-galactosidase (Promega, Madison, WI) diluted to 1 μg/ml in PBS plus 0.1% Triton X-100 (Fisher Scientific, Pittsburgh, PA) for 1 hr at room temperature. Tissues were incubated in a biotinylated anti-mouse IgG secondary antibody and immunoreactivity was visualized using Vectastain Elite and Vector VIP kits (Vector Laboratories, Burlingame, CA).

Plasmid rescue: Adult flies were ground in lysis buffer (0.32 mM sucrose, 10 mm TRIS pH 8.0, 5 mm MgCl2, 1% Triton-X in TBS), filtered through Nytex cloth, and centrifuged at 1100 g at 4° for 12 min. The pellet was resuspended in 75 mm NaCl, 24 mm EDTA and digested with 4 μg/fly equivalent Proteinase K (BRL, Hercules, CA) plus 1% Triton-X at 37° for 2–18 hr. Following phenol/chloroform extraction, the DNA was precipitated in KCl/ethanol. The pellet was resuspended in TE and 3–4 μg of DNA was digested with XbaI or XbaI and NotI (Promega), phenol/chloroform extracted, NaOAc/ethanol precipitated, and resuspended in dH2O. Ligation of 1–2 μg of digested DNA was performed in 200 μl volume using 1–3 units T4 DNA ligase (Promega) at 14° for 4–5 hr. Following phenol/chloroform extraction and NaOAc/ethanol precipitation, the ligated DNA was transfected into DH5α Max Efficiency competent cells (BRL) which were then plated on 50 μg/ml kanamycin NZCYM plates for plasmid rescue (Pirrotta 1986). DNA isolated from individual colonies was checked for PZ elements by restriction mapping of miniprep DNA with NotI, XbaI, and NotI/XbaI (Sambrooket al. 1989; Mlodzik and Hiromi 1992).

Polymerase chain reaction (PCR): All PCR reactions were performed in a thermocycler (MJ Research, Inc., Watertown, MA) according to standard protocols (Inniset al. 1990) using Taq DNA polymerase (Promega), dNTPs (Epicentre Technologies Corp., Madison, WI), and oligonucleotide primers (Operon Technologies, Inc., Alameda, CA). When the gene identity was known from published data, primers were based on sequences at the end of the PZ insert and on published sequence from the affected gene. The latter included the genomic region adjacent to the PZ insertion in l(2)1275 (available from the Berkeley Drosophila Genome Project [Berkeley, CA, 1998 at], and for the EF-1α F1 gene [Hovemannet al. 1988; Flybase 1998 at]). DNA was amplified, subcloned into the TA vector (Invitrogen, San Diego), and sequenced.

Library construction and screening: For l(2)6072, DNA adjacent to the PZ insertion was cloned from a genomic DNA library constructed from adult flies. Following partial digestion with Sau3A and A/G nucleotide fill-in, DNA was ligated into the Lambda Fix II vector and packaged with the Gigapack III Gold packaging extract (Stratagene, La Jolla, CA). From approximately 1.5 × 105 pfu of unamplified library plated with NM538 bacteria on NZCYM plates, a single clone was plaque-purified (Sambrooket al. 1989) using a random-primed radiolabeled probe (Prime-It; Stratagene) complementary to a 400 bp PCR product containing genomic DNA adjacent to the PZ insertion in l(2)6072 (see above).

The transcription units corresponding to l(2)4524 and l(2)6072 were cloned from cDNA libraries. Two third-instar central nervous system (CNS)/ring-gland cDNA libraries, one constructed in the EMBL4 vector and one in the Lambda ZAP vector (Stratagene), were screened using standard protocols (Sambrooket al. 1989). In each case, plaque purification began with an initial hybridization to 8 × 106 pfu plated with NM538 bacteria on NZCYM plates. Random-primed radiolabeled probes (Prime-It; Stratagene) were made using 3.0-kb and 1.4-kb fragments adjacent to the PZ insertions of l(2)4524 and l(2)6072, respectively.

Subcloning and sequencing: For sequencing, four HindIII fragments representing the 3.0 kb adjacent to the PZ insertion in l(2)4524 were subcloned into the Bluescript SK+ plasmid (Stratagene), and the PCR products generated for l(2)1275 were subcloned into the TA vector (Invitrogen). All other sequencing of genomic DNA was performed directly from the rescued PZ plasmids. Sequencing of cDNAs was carried out either directly from purified phage DNA (Sambrooket al. 1989) or, in the case of the Lambda ZAP clones, from excised phagemid DNA (Stratagene). All sequencing was carried out as described for the dsDNA Cycle Sequencing System (BRL) using radiolabeled dCTP (New England Nuclear, Boston).

Lethal phase analysis: The lethal phase(s) for 15 second-chromosome PZ insertion lines balanced over In(2LR)Gla, Gla Bc Elp were determined as follows: Larvae from 1.0–1.5 hr egg collections were allowed to develop at 25° until mid-third instar and transferred to 18° for 16 hr to enhance the Black Cell (Bc) phenotype. Larvae were scored as heterozygous or PZ-homozygous by the presence or absence of black cells, respectively. Since Bc/Bc homozygotes die as first instars, a 2:1 ratio of third-instar heterozygotes to homozygotes was taken as evidence that neither significant lethality nor developmental delay had occurred in insertion homozygotes prior to third instar. Homozygotes were placed on fresh food and allowed to develop at 25°. Larvae were checked twice daily and the number of live and dead animals and their developmental stage scored. Examining the animals for up to 14 days allowed the detection of abnormal developmental timing as well as determination of lethal phase.

Whole-mount RNA in situ hybridization: CNS/ring gland complexes were dissected into PBS, fixed in 4% paraformaldehyde, PBS, 0.1% Tween 20 for 20–30 min, and rinsed in PBS, 0.1% Tween three times for 20 min each. Tissue was pretreated with Proteinase K (Promega) and hybridized as described by Tautz and Pfeifle (1989) except that hybridization was carried out for 1 day at 65° and washes lasted for 2 days at 65°. RNA probes were generated using the T3 and T7 promoters and the DIG Labeling Mixture (Boehringer Mannheim, Indianapolis), and were boiled for 30 min prior to use. Hybridization was detected with a monoclonal antibody against digoxygenin coupled to alkaline phosphatase (Boehringer Mannheim).


Enhancer-trap screen: We screened 510 lethal PZ enhancer-trap lines, in which the recessive lethality had previously been shown to be caused by the enhancertrap insertion (Mlodzik and Hiromi 1992; A. Spradling, personal communication). We selected those showing strong β-galactosidase reporter gene activity as detected by X-Gal staining in the ring gland of wandering third-instar larvae. We have not excluded lines showing expression outside the ring gland, because in other insects there is evidence for production of EC in tissues other than the prothoracic glands (Rees 1985). Since the β-galactosidase protein is fairly stable in vivo, assaying wandering third-instar larvae allowed us to identify enhancer-trap insertions into or near to genes expressed in the ring gland at other times during the third instar including presumably the times of the commitment and prepupal peaks of EC.

X-Gal staining in the ring gland was detected in 76/510 = 15% of lines examined (31/270 = 11% of second-chromosome and 45/240 = 19% of third-chromosome lines). In 25 of the 76 lines, the ring gland was the predominant or most darkly staining tissue (Table 1), but in only three lines was X-Gal staining completely restricted to the ring gland at this stage. Significant staining occurred in the CNS in 70 lines and in one or more pairs of imaginal discs in 59 lines.

Expression profiles: Based on intensity and specificity of ring-gland staining, 18 second-chromosome and 13 third-chromosome lines were selected for a more detailed analysis (Table 2). The lacZ expression patterns in the ring gland, CNS, imaginal discs, lymph gland, fat body, and salivary glands were determined enzymatically by incubating with X-Gal and/or immunohistochemically by staining with a monoclonal antibody to β-galactosidase. Expression was assayed in first-instar, second-instar, second- to third-instar molting, and early, middle, and late third-instar larvae.

View this table:

Tissue staining patterns in wandering third-instar larvae of lines expressing lacZ in the ring gland

View this table:

Staining of different parts of the ring gland in enhancer-trap lines

In 23 lines, β-galactosidase activity was detected in the ring gland at all stages analyzed (Table 2). Among these lines, staining was observed throughout the ring gland, i.e., in the LC, MC, and the corpus cardiacumhomologous cells (CCC) located at the two horns of the ring gland, in two cases, was restricted to the LC and MC in seven cases and to the LC in 14 cases. There was significant staining in one or more pairs of imaginal discs in all 23 lines. In line l(2)1857, the ring gland stained darkly at all larval stages (Figure 1) while the only other significant staining was in the eye discs of late third-instar larvae. In embryos of this line, X-Gal staining marked the apparent ring-gland precursor cells as early as 5 hr after egg laying (AEL; Figure 1B).

In eight lines, β-galactosidase activity was detected in the ring gland of only some of the larval stages analyzed (Table 2). In l(2)0248, staining was detected in the ring glands of second-instar larvae prior to but not during molting, and again in mid- and late third-instar larvae, the times that immediately precede EC peaks. In the remainder of the lines, staining was observed in late larval stages but not earlier ones. All ring gland cells stained in one line, and staining was restricted to the LC and MC in four lines and to the LC in three lines (Figure 2A). In two lines, l(2)2535 and l(2)6072, lacZ expression within the ring gland was clearly restricted to the JH-producing MC (Figure 2, D and E). All but one line, l(2)7447, also showed staining in imaginal discs.

Figure 1.

Developmental profile of l(2)1857 staining. (A–E) X-Gal staining; (F) Immunohistochemical staining with monoclonal antibody (mAb) to beta-galactosidase. (A) 1–2-hr AEL; anterior pole stained. (B) 5-hr AEL; apparent ring gland precursor cells stained. (C) Newly hatched homozygous first-instar larva; ring gland darkly stained. (D) CNS/ring gland complex from heterozygous late first-instar larva; ring gland darkly stained. (E) Third-instar ring gland; LC stain darkly but reaction product has apparently diffused into MC as well. (F) Anti-β-galactosidase staining; only nuclei of LC stain darkly. Bars, 100 μm.

Figure 2.

(A–C) l(2)4524 third-instar ring glands. (A) Only the LC stain with X-Gal, while both the LC and the MC stain with a mAb to β-galactosidase (B) and in RNA in situ hybridization using a lacZ probe (C). (D–G) l(2)6072 third-instar ring glands. Only the MC stain with X-Gal (D), a mAb to β-galactosidase (E), and in RNA in situ hybridization using a lacZ probe (F). However, RNA in situ hybridization using a V-ATPase C subunit probe shows that a nearby endogenous gene is expressed throughout the entire ring gland (G) as well as in other tissues. All ring glands were of similar size when dissected from larvae. Bar, 50 μm.

Isolation and characterization of genomic DNA adjacent to the PZ insertion: The genomic DNA adjacent to the PZ insertion was isolated from five second-chromosome and five third-chromosome lines, l(2)3909, l(2)4012, l(2)4524, l(2)6353, l(2)7447, l(3)3540, l(3)3544, l(3)5822, l(3)6015, and l(3)6286, by plasmid rescue, from line l(2)1275 by PCR utilizing sequence data available from the Berkeley Drosophila Genome Project and from line l(2)6072 by screening a genomic library made from that line. l(2)6072 shows expression in the MC, whereas the remaining lines show expression in the LC or both LC and MC. That the genomic DNA was in fact immediately adjacent to the PZ insertion was confirmed by sequencing across the insertion/genomic DNA boundary using an internal enhancer-trap primer. We also sequenced the initial 200–700 bp of genomic DNA from each isolate. These sequences have been deposited in GenBank. BLAST searches (Altschulet al. 1990) indicated that for six of the sixteen lines analyzed, the PZ insertion is in or near a previously cloned Drosophila gene or element (Table 3). The insertion is located within 60 bases 5′ of the transcription start site in three of these, 71 bases into the 5′ untranslated region in one, and into the intron in another.

View this table:

PZ enhancer-trap insertions into or near to Drosophila genes or elements

l(2)3909: The extended third-instar phenotype of l(2)3909, in which the PZ insertion is 60 bp 5′ to the transcriptional start site of the calmodulin (Cam) gene, is similar to that reported for EMS-induced Cam alleles (K. Beckingham, personal communication). Homozygous Cam null animals die as first instar larvae (Heimanet al. 1996) but in the heteroallellic condition with either of two EMS-induced alleles they produce animals that survive to pupae. The extended third larval instar seen in these heteroallelic combinations (K. Beckingham, personal communication) is suggestive of an endocrine defect. The l(2)3909 lacZ and wild-type Cam expression patterns are very similar or identical for all stages of development from embryo to late third instar (Kovalick and Beckingham 1992; K. Beckingham, personal communication). This pattern is dynamic during the third instar. In early third-instar larvae, expression is limited to and very robust in the ring gland but by the prepupal stage it has expanded to include most tissues (Figure 3, E–G). We were unable to definitively determine by RNA in situ hybridization whether Cam expression within the ring gland is limited to the LC or whether it also includes the MC, but immunohistochemical localization of β-galactosidase indicated that the reporter gene is expressed in both the LC and the MC.

l(2)6353: The enhancer trap of l(2)6353 is inserted into the 5′ untranslated region of the gene encoding the DC0 catalytic subunit of cAMP-dependent protein kinase (cAMP-PKA or PKA) (Lane and Kalderon 1993). Homozygotes of l(2)6353 display slow development, molting to third instar one day later than heterozygous siblings, and most die as third-instar larvae. Many of the larvae surviving to third instar remain as such for up to eight days and initiate wandering behavior but never pupariate. Approximately 15% of the homozygotes, however, do form small pupae within 48 hr of their heterozygous siblings, and about two-thirds of these survive to eclose as miniature adults with unexpanded wings. A simple explanation for this is that PKA is involved in the signal transduction pathway leading to the large prepupal EC peak, and that in the nonpupariating animals the signal is not sufficient to produce the peak. Since l(2)6353 does not appear to be a null mutation for cAMP-PK, it is possible that some animals can generate sufficient signal to produce the EC required to pupariate, accounting for the escapers produced in this line.

Figure 3.

A comparison of lacZ expression patterns as detected by X-Gal staining and by mAb staining: (A) l(2)2278 X-Gal and (B) mAb; (C) l(2)1275 X-Gal and (D) mAb. The dynamic pattern of expression in l(2)3909 is shown in X-Gal staining of (E) early and (F) late third-instar CNS (ring gland removed in F). (G) mAb staining of l(2)3909 wandering third-instar CNS/ring gland complex.

The developmental and morphogenetic defects in l(2)6353 are similar to those described for weak DC0 mutants in Drosophila (Lane and Kalderon 1993). It seemed likely that l(2)6353 would be a null mutant since the enhancer trap is inserted into the 5′ transcribed but untranslated region of DC0, thereby separating the transcriptional start site from the coding region by about 14 kb that has previously been shown to contain termination and aberrant splicing signals (Horowitz and Berg 1995). Perhaps a less efficient secondary start site is utilized or a less translatable message is produced from the DC0 locus containing the PZ insert, resulting in a weak rather than a null mutation.

l(2)1275: A phenotype suggestive of an endocrine defect was observed in line l(2)1275, in which the PZ insertion is 30 bp 5′ to the transcriptional start site of the gene encoding the translational elongation factor EF-1α F1 (Hovemannet al. 1988). Homozygotes exhibit slow growth, and molt to third instar approximately one day later than sibling heterozygotes. Ten to 15% of those surviving to third instar remain alive for up to 5 days and initiate wandering but never pupariate. There are two EF-1α genes in Drosophila but only the F1 gene potentially disrupted here is expressed at high levels during larval development (Hovemannet al. 1988).

l(2)4524: Four HindIII fragments representing 3.3 kb adjacent to the PZ insertion were subcloned from the genomic DNA plasmid rescued from line l(2)4524. The two proximal fragments of 581 bp and 921 bp and the distal fragment of 396 bp were sequenced, as were the proximal 216 bp and distal 600 bp of the fourth fragment (Figure 4A). A BLAST analysis revealed three regions of 255, 451, and 473 bp whose conceptual translations show colinear homology to a Caenorhabditis elegans 66.5-kD hypothetical protein (Wilsonet al. 1994). The 255-bp region may represent a complete exon since the entire 255 bp are contained in an open reading frame (ORF) that is bounded by splice donor and acceptor sequences, and a potential TACTAAC splicing branch point (Mountet al. 1992) is located 35 bp upstream. The 85-amino-acid sequence encoded by this region is 55% identical to a portion of the C. elegans protein (Figure 4C). The other regions of homology can be resolved into a single 924-bp ORF by removal of a putative 53-bp intron that contains splice donor and acceptor sequences and a potential TACTAAC box (Figure 4B). The resulting “spliced” ORF encodes a 308-amino-acid sequence which is 61% identical to the C-terminal region of the C. elegans protein (Figure 4C). Approximately 8 × 106 plaques from each of two cDNA libraries, one made from the CNS of third-instar larvae and one made from the CNS and ring gland of wandering third-instar larvae, were screened using the 3 kb adjacent to the enhancer trap as a probe. Two clones were plaque purified from each library (584 and 314 bp, and 563 and 600 bp, respectively), sequenced, and shown to be from the same gene and identical to the 3′ ORF region of the rescued genomic DNA (Figure 4A). All of the cDNA clones contain a 5′ ORF followed by a TAA stop codon and a 303-bp 3′untranslated region followed by a poly-A tail. The longest cDNA contains a 279-bp ORF encoding a 93-amino-acid sequence that is 50% identical to the C-terminal region of the C. elegans protein (Figure 4C).

Figure 4.

(A) Diagram of plasmid-rescued genomic DNA adjacent to the l(2)4524 PZ insertion. Regions sequenced are boxed. Regions homologous to a C. elegans hypothetical 66.5-kD protein are shaded. The sizes of the homologous regions are given in bases inside the boxes. H, HindIII site; H/N, HindIII or NotI site; *, putative intron. The position of the cDNA relative to the genomic DNA and the ORF, 3′ UTR, and poly-A sizes are indicated below the diagram (double line). (B) Sequence of the putative 53-bp intron. Acceptor and donor splice sequences are double underlined. A potential TACTAAC box is in bold. (C) Macaw alignments (Schuleret al. 1991) of regions of homology between conceptual translations of l(2)4524 and C. elegans sequences. Darkness of shading indicates the significance of the match. The putative intron sequence (B) has been removed from the 3′ sequence. The region corresponding to the longest cDNA is underlined.

l(3)3540 and l(3)6015: We also identified enhancer traps inserted into or very near to two prevously characterized genes on the third chromosome, tramtrack (ttk; Giesenet al. 1997) and couch potato (cpt; Bellenet al. 1992; Table 3). These two genes encode a DNA-binding and an RNA-binding protein respectively, both of which are involved in peripheral nervous system development.

l(2)2278: PZ homozygotes of l(2)2278 developed normally until third instar. This last instar, however, was prolonged up to 8 days. Between days 4 and 8 of the extendedinstar, approximately90% of the homozygotes pupariated but died before metamorphosing. Imaginal discs of these animals showed hyperplastic overgrowth, which might account for the prolonged third instar. Complementation experiments indicate that l(2)2278 represents an allele of the expanded locus (Boedigheimer and Laughon 1993; data not shown).

l(2)4012: In this line the enhancer trap is inserted into a “1360” middle repetitive element (Kholodilovet al. 1987) but there are no known transcriptional units close to the insertion. The mutation causes leg, wing, and bristle polarity defects and a behavioral phenotype suggestive of nervous system defects that cannot easily be attributed to an endocrine deficiency. The fact that the enhancer trap is inserted into a 1360 repetitive element has made any molecular characterization of this line difficult.

l(2)6072: Restriction enzyme analysis of the genomic DNA obtained by plasmid rescue from line l(2)6072 indicated that the clone might contain concatamers of unrelated DNA. Sequence and PCR analyses showed that the first 308 bp of the clone does represent genomic DNA immediately adjacent to the PZ insertion, so this was used to screen a genomic library constructed from this line. A single genomic clone was isolated that contained most or all of the PZ insertion and 1.4 kb of flanking DNA. Sequencing confirmed that this 1.4-kb fragment was immediately adjacent to the insertion. Approximately 8 × 106 plaques of a wandering third-instar CNS/ring-gland cDNA library were screened using the 1.4-kb fragment as a probe, and two cDNA clones of 1.5 and 1.8 kb were plaque purified. The 5′ end of the longer cDNA has a 158-bp overlap with the 3′ end of the 308 bp adjacent to the PZ insertion, indicating that the insertion is 150 bp 5′to the start of the longer cDNA. A BLAST analysis of the cDNA showed that it encodes the Drosophila homolog of the C subunit of V-ATPase and has high homology to tick, human, slime mold, and yeast C subunits (Figure 5). The conceptual translation of the Drosophila cDNA reported here is most homologous to the tick protein (Luo 1993).

Comparison of expression patterns of reporter lacZ as detected by X-Gal and antibody staining, and of nearby genes as detected by RNA in situ hybridization: We compared lacZ expression patterns in the ring gland, CNS, imaginal discs, lymph gland, fat body, and salivary glands of third-instar larvae of six lines, l(2)1275, l(2)1857, l(2)2278, l(2)3909, l(2)4524, and l(2)6072, as detected enzymatically using X-Gal and immunohistochemically using a monoclonal antibody to β-galactosidase. In most cases the overall staining patterns were very similar to each other although X-Gal staining was often more diffuse than antibody staining (Figure 3, A–D). This difference was most evident in lines such as l(2)3909, which show a dynamic expression pattern during the third instar (Figure 3, E–G). Within the ring gland, lacZ-positive LC were detected by both methods. However, in five of the fifteen cases where the MC also stained and in the three instances where the CCC also stained, expression in these non-LC ring gland cells was detected only immunohistochemically (Table 2; Figure 2, A and B). These discrepancies presumably reflect different sensitivities of the methods used.

Cloning of the genes affected by the PZ insertions made it possible to compare the expression of the reporter gene with that of the affected endogenous gene. We performed RNA in situ hybridization on whole mounts of ring glands, CNS, imaginal discs, and lymph glands of heterozygous third-instar larvae of four lines, l(2)1275, l(2)3909, l(2)4524, and l(2)6072, using probes corresponding to transcribed regions of genes adjacent to the PZ insertions (Table 3). The probes used were a 539-bp PCR product from exon 2 of EF-1α F1 (Hovemannet al. 1988), a 568-bp fragment of the calmodulin gene containing 3′ untranslated sequences common to both known transcripts (Kovalick and Beckingham 1992), a 920-bp genomic fragment containing coding and 3′untranslated regions of a novel Drosophila gene, and a 1.8-kb cDNA clone for the Drosophila C subunit of vacuolar ATPase (V-ATPase), respectively. The RNA staining patterns for lines l(2)1275, l(2)3909, and l(2)4524 closely matched the lacZ expression patterns detected by X-Gal and β-galactosidase antibody staining. For lines l(2)1275, l(2)3909 and l(2)4524 endogenous gene expression was detected in both the LC and the MC of the ring gland, which is in agreement with the antibody results noted above (see, for example, Figure 2, A–C). However, for line l(2)6072, the expression pattern for the V-ATPase C subunit gene was very different from the lacZ pattern. In third-instar larvae, reporter gene expression as assayed by X-Gal staining (Figure 2D), mAb to β-galactosidase (Figure 2E), or RNA in situ hybridization using a lacZ probe (Figure 2F) was limited to the MC of the ring gland. In contrast to these results, in situ hybridization using a probe for the C subunit gene revealed nearly ubiquitous expression in the ring gland (Figure 2G) as well as in the CNS, imaginal discs, and lymph gland. The expression pattern of the lacZ reporter gene differs significantly from that of other insertions and from that of an endogenous gene located less than 200 bp from the PZ insertion site.

Figure 5.

Macaw alignments (Schuleret al. 1991) of the C subunits of V-ATPase from various sources. Darkness of shading indicates the significance of the match. References: fly, this study; tick (Luo 1993); Dictyostelium (Gerisch and Westphal, unpublished results), Genbank accession no. 1718089; yeast (Beltranet al. 1992); human (Van Hilleet al. 1993).


We have identified 76 genes that are strongly expressed in the Drosophila ring gland during development. For nine of these, further studies of expression pattern, mutant phenotype and molecular nature identify the genes as strong candidates for playing an important role in endocrine functions controlling development. In some cases, the molecular nature of the predicted gene product is quite consistent with conclusions drawn from physiological studies of the endocrine system of larger insects. In other cases, our data suggest significant extension of present models for the regulation of insect endocrine glands.

Two of the genes we have identified encode products that have already been implicated in the functioning of prothoracic glands of other insects. The insert in l(2)3909 is 60 bp 5′ to the transcription start site of the calmodulin gene (Doyleet al. 1990). The gene is expressed exclusively and at high levels in the ring gland of third-instar larvae, suggesting an important, presumably endocrine function for calmodulin in that tissue as has already been suggested for lepidopterans (Gilbertet al. 1988; Grangeret al. 1995). Calmodulin and other Ca2+-binding proteins are integral to the transduction of a wide range of Ca2+-dependent signals (Nikiet al. 1996) and there is clear evidence for the Ca2+ dependence of EC production in the Manduca larval prothoracic gland (PTG), at least for the commitment peak early in the last larval instar. Studies of these glands in vitro show that changes in intracellular Ca2+ concentrations are both necessary and sufficient for the generation of the commitment peak of EC (Smithet al. 1986) and that PTTH-mediated stimulation of EC production requires extracellular Ca2+ (Rybcyznski and Gilbert 1994). Stimulation of EC production by brain extracts on isolated Drosophila ring glands is also Ca2+-dependent (Henrich 1995). A simple interpretation is that binding of PTTH to its receptor initiates an influx of Ca2+ into the cell and that this influx activates downstream elements of the Ca2+-cAMP-dependent signaling pathway. It is known that Ca2+ activates PTG adenylate cyclase both directly and as a complex when bound to calmodulin (Melleret al. 1990). Since cAMP phosphodiesterase activity is low at this stage (Smith and Pasquarello 1989), cAMP is expected to accumulate. Both large and small PTTH stimulate increased cAMP levels in PTG (Smithet al. 1984; Smith and Pasquarello 1989; Watsonet al. 1993), and a rise in cAMP levels occurs with PTTH-stimulated EC production in early last-instar PTG.

The enhancer trap of l(2)6353 is inserted into the 5′ untranslated region of the DC0 gene, which encodes the catalytic subunit of protein kinase A (PKA or cAMP-PK; Lane and Kalderon 1993). This protein probably functions downstream of cAMP in the Ca2+-cAMP-dependent signaling pathway. PKA is activated in M. sexta PTGs by PTTH immediately prior to EC production (Smithet al. 1986). This is consistent with the idea that activation of the Ca2+-cAMP-dependent signaling pathway by PTTH leads to PKA-dependent phosphorylation of key proteins including ribosomal protein S6 (Rountreeet al. 1987; Combest and Gilbert 1992), and that this causes changes in selective translation leading to increased EC production.

The enhancer trap in l(2)1275 is inserted 30 bp 5′ to the transcription start site of the gene encoding the translation elongation factor EF-1α F1 (Hovemannet al. 1988). A role for this factor in hormone production and/or secretion has not been previously suggested, but it is plausible that it plays a role downstream of ribosomal protein S6 in the Ca2+-cAMP-dependent signaling pathway. The EF-1α F2 gene is expressed at high levels during metamorphosis (Walldorfet al. 1985), a time of higher and prolonged levels of EC (Hodgettset al. 1977). Studies in M. sexta have shown that EC production is under translational control (Smith and Pasquarello 1989) and that certain proteins are selectively translated and phosphorylated in response to PTTH (Rybczynski and Gilbert 1994). This selective translation could result from the production and/or activation by phosphorylation (Venemaet al. 1991) of EF-1α which has been shown to be a key regulator of translational control in other systems. Rapamycin, an inhibitor of S6 phosphorylation, dramatically inhibits selective translation of both EF-1α and EF-2 in mammalian cells (Teradaet al. 1994), suggesting that synthesis of these elongation factors is selectively enhanced by S6 phosphorylation.

There is another possible function for EF-1α in the regulation of hormone titers. This factor is structurally conserved among diverse species including Drosophila and probably has similar functions in all organisms (Hovemannet al. 1988). In Tetrahymena, EF-1α has two entirely separate functions. In addition to its role in directing the binding of aminoacyl-tRNAs to the ribosome during translation, EF-1α can function as a Ca2+/calmodulin-dependent F-actin bundling factor (Numata 1996). Changes in the actin cytoskeleton have been proposed to mediate neuropeptide and hormonal secretion (Trifaro and Vitale 1993). Ultrastructural studies have shown an increase in smooth endoplasmic reticulum and secretory vesicles throughout the final instar in EC-producing cells of both Drosophila ring glands (Aggarwal and King 1969) and Manduca PTG (Hantonet al. 1993). In flies, there is a 50-fold increase between 50 and 94 hr of development, followed by an additional 10-fold increase over the last 4 hr of the third instar. It is possible, therefore, that the hemolymph titer of EC is regulated both by biosynthetic rates and by control of secretion, and that EF-1α may be involved in regulating one or both of these processes.

The gene identified in l(2)4524 and its C. elegans homolog do not show significant homology to any other known proteins. They do have some limited homology to a family of carboxylases, particularly to propionyl-CoA carboxylases, but it is not strong enough to support conjecture about function. However, the behavioral phenotype is very similar to that seen in l(2)3909 and in weak Cam mutants.

We also identified enhancer traps inserted into or very near to two previously characterized genes on the third chromosome, tramtrack (ttk) and couch potato (cpo). Although these two genes are known from their roles in peripheral nervous system development (Bellenet al. 1992; Guoet al. 1995), it is likely that they have other functions as well. Amorphic cpo alleles are embryonic lethal, but the homozygous embryos show no obvious developmental abnormalities (Bellenet al. 1992). Ttk is required for embryonic glial cell development (Giesenet al. 1997) and it also functions in the assignment of cell fates during sensory organ development (Xiong and Montell 1993; Guoet al. 1995). If expressed early enough, both of these genes could play roles in cell fate determination during ring gland development.

Complementation experiments indicate that l(2)2278 represents an allele of the expanded locus, which controls wing size (Boedigheimer and Laughon 1993). The prolonged third instar observed in l(2)2278 may be the result of the hyperplastic overgrowth of imaginal discs, rather than a direct effect of the mutation on ring gland function. Other imaginal disc overgrowth mutants also show prolonged larval life, apparently because overgrowing discs prevent the endocrine system from initiating metamorphosis, rather than because of a direct effect of the mutation on the endocrine system (Bryant and Levinson 1985). The significance of expanded expression in the ring gland is unknown at present. l(2)4012 also shows leg, wing, and bristle abnormalities and a behavioral phenotype (uncoordinated movement, inability to fly) that cannot easily be attributed to any endocrine deficiency. The fact that the enhancer trap is inserted into a 1360 repetitive element has made any molecular characterization of this line difficult.

Our screen identified one enhancer trap, l(2)6072, with strong expression in the MC of the ring gland, which is thought to be the source of JH. Whether assayed for β-galactosidase activity with X-Gal or β-galactosidase protein production with a monoclonal antibody, the only significant lacZ reporter gene expression in l(2)6072 larvae is in the MC during the second and third instars. With prolonged X-Gal incubation, low levels of expression are detected in the midgut and brain as well. lacZ expression increases in the pupal brain but in the ringgland it remains restricted to the MC. In adults, strong lacZ expression occurs in the rectal papillae.

The enhancer trap of l(2)6072 is inserted 5′ to the coding sequence of the gene encoding the C subunit of V-ATPase (Van Hilleet al. 1993). The main known function of V-ATPase in insects is to act as a proton pump to energize active transport at the apical plasma membrane of ion-transporting epithelia, for example the rectal papillae, midgut and malpighian tubules (Harvey 1992). That the reporter gene expression in l(2)6072 represents the action of a legitimate C-subunit enhancer is supported by the strong lacZ expression seen in the rectal papillae. However, the specific expression of lacZ in the larval CA may represent a different C subunit/V-ATPase function in those cells during development. V-ATPases are known to play an important role in neurotransmission by providing the energy for the uptake of neurotransmitters into synaptic vesicles, and they may also be important in synaptic vesicle formation and in neurosecretion (Nelson 1993). It is therefore possible that the ring gland V-ATPase functions in the uptake of neuropeptides in the MC of the ring gland.

Since lacZ expression in this initial screen was detected by X-Gal staining, which often diffused throughout the ring gland, it was not always evident for the remaining 75 ring gland-positive lines whether the EC-producing LC were the only ring gland cells expressing the reporter gene or whether the MCs and/or the CCC expressed lacZ as well. This issue was resolved for a subset of these lines by detecting beta-galactosidase immunohistochemically. In eight of nine instances where different ring gland patterns were observed with X-Gal and antibody staining, the antibody was more sensitive. In line l(2)10280, however, nuclear localization indicative of authentic reporter lacZ expression was detected in the LCs by X-Gal staining but antibody staining revealed only a light, diffuse staining pattern in the same cells.

Most ring gland-positive lines expressed lacZ in other tissues as well. The most common additional pattern was a uniform, usually diffuse staining of the optic lobes and ventral nerve cord, sometimes darker in the neuropile region of the optic lobes, that was observed in over half of the lines assayed including those lines that were ring gland-negative. This staining may represent, as suggested by Mlodzik and Hiromi (1992) a chronic pattern resulting from sequences contained within the PZ construct itself rather than from sequences in the flanking genomic DNA. Punctate staining in the CNS occurred at a much lower frequency and probably represents legitimate externally driven reporter-gene expression. The second most common additional lacZ-positive tissues were imaginal discs. Various combinations of disc types stained but usually the eye discs were included and often were the predominant or only discs to stain. Eye-disc staining typically occurred at and behind the morphogenetic furrow (Readyet al. 1976). This eye disc reporter gene activity has been noted previously for this and similar enhancer-trap constructs (Mlodzik and Hiromi 1992) and again may represent a chronic pattern resulting from sequences within the construct itself.

Another problem with enhancer-trap screening is that the reporter construct may detect only a subset of the regulatory elements controlling nearby genes. This is dramatically illustrated in line l(2)6072 where, even though the enhancer trap is inserted within 150 bp of the 5′ end of the C subunit V-ATPase gene, lacZ is expressed almost exclusively in the MCs of the ring gland whereas the endogenous gene is ubiquitously expressed.

We have identified nine genes expressed in the ring gland during development and have suggested ways in which they could function in the neuroendocrine control of development. Quantification of hormone levels from these mutants as well as rescue experiments will be required to determine the precise nature of the endocrine defects, if any, in these mutants, and to define the normal functions of the set of gene products they identify.


The authors thank Dr. John Tower (University of Southern California) for providing the stocks used in these experiments, Drs. Rudi Gram and Gunter Korge (Freie University), and Dr. Frantisek Sehnal (Czech Academy of Sciences) for the generous gifts of cDNA libraries. This work was supported by National Science Foundation Grant IBN-9210656 (to P.B.) and by National Institutes of Health Training Grant GM-07134 (to P.H.).


  • Communicating editor: S. Henikoff

  • Received October 12, 1997.
  • Accepted February 2, 1998.


View Abstract