To identify genes that alter their expression in synchrony with the late third instar and prepupal pulses of 20E, RNA was isolated from w¹¹¹⁸ animals staged at -18, -4, 0, 2, 4, 6, 8, 10, and 12 hours relative to pupariation, labeled, and hybridized to Affymetrix Drosophila Genome Arrays. The sensitivity and accuracy of the array data were determined by comparing the expression patterns of known 20E-regulated genes with previously published developmental Northern blot data 2728. A subset of this analysis, depicted in Figure 1, reveals that the temporal expression pattern of key regulatory genes - EcR, usp, E74A, DHR3, FTZ-F1, and DHR39 - are faithfully reproduced in the temporal arrays, as well as the 20E-regulated switch from Sgs glue genes to L71 late genes in the larval salivary glands, and the expression of representative IMP and Edg genes in the imaginal discs and epidermis. This comparison demonstrates that the microarrays accurately reflect the temporal patterns of 20E-regulated gene expression at the onset of metamorphosis and have sufficient sensitivity to detect rare transcripts such as EcR and E74A.

EcR mutants die during early stages of development, complicating their use for studying receptor function during metamorphosis. To circumvent this problem, we employed a transgenic system that allows heat-induced expression of double-stranded RNA corresponding to the EcR common region to disrupt EcR function at puparium formation (EcRi) 29. RNA was harvested for array analysis from EcRi animals staged at -4, 0, and 4 hours relative to pupariation. All EcRi animals formed arrested elongated prepupae, consistent with an effective block in 20E signaling and highly reduced EcR protein levels (Figures 2 and 3c in 29). Data obtained from these arrays were compared to our array data from control animals at the same stages of development to identify EcR-dependent genes. The initial effect of EcR RNA interference RNA (RNAi) is significant upregulation of gene expression in late third instar larvae, followed by a switch at puparium formation such that the majority of genes are not properly induced (Figure 2a). These data are consistent with genetic studies of usp that define a critical role for this receptor in repressing ecdysone-regulated genes during larval stages 30, and provide further evidence that one essential function for the EcR-USP heterodimer is to prevent premature maturation through the repression of select 20E target genes during larval stages.

A total of 4,188 genes change their expression at least 1.5-fold in at least one time point in EcRi animals (Additional data file 1), suggesting that almost a third of all genes require EcR, either directly or indirectly, for their proper regulation at the onset of metamorphosis. This number is consistent with the 2,268 genes that have been reported to change their expression at pupariation in one of five tissues examined: midgut, salivary gland, wing disc, epidermis, and central nervous system 20. It is also similar to the 4,042 genes that change their expression at least 1.5-fold at pupariation in our temporal arrays. Of these 4,042 genes, 2,680 are affected in EcRi animals, supporting the proposal that EcR plays a major role in coordinating transcriptional responses at the onset of metamorphosis. Not all genes that change their expression at pupariation, however, are dependent on EcR. Several such transcripts were selected for validation by Northern blot hybridization (Additional data file 3). This is consistent with an earlier microarray study of EcR-regulated genes in the larval midgut 20. This study found that of 955 genes that change their expression in wild-type midguts at the onset of metamorphosis, 672 genes are affected by an EcR mutation while 283 genes are unaffected, close to the proportion of EcR-independent genes identified by our study. This is also consistent with earlier studies that indicate that other signaling pathways are active at this stage in development. For example, the miR-125 and let-7 microRNAs are dramatically induced at puparium formation, in tight temporal synchrony with the 20E primary-response E74A mRNA, but do so in a manner that is independent of either 20E or EcR 24. Similarly, α-ecdysone, the immediate upstream precursor of 20E, has critical biological functions 233132, can activate the DHR38 nuclear receptor 22, and can induce genes in Drosophila third instar larvae that are distinct from those that respond to 20E (RBB, GL and CST, unpublished results). The sesquiterpenoid juvenile hormone can also function with 20E to direct specific transcriptional responses during early metamorphosis 333435. The results of the study described here, however, indicate that most genes that change their expression at the onset of metamorphosis do so in an EcR-dependent fashion, and pave the way for future studies that integrate these responses with those of other signaling pathways.

To identify 20E-regulated genes, wandering third instar larvae were dissected and their organs cultured in the presence of either no hormone, 20E alone, cycloheximide alone, or 20E plus cycloheximide for 6 hours. RNA extracted from these samples was analyzed on Affymetrix Drosophila Genome Arrays. Comparison of the no hormone and 20E-treated datasets led to the identification of 20E-regulated genes, while comparison of the cycloheximide dataset with data derived from organs treated with 20E and cycloheximide led to the identification of a set of genes we refer to as 20E primary-response genes. In comparing these datasets, it is important to note that cycloheximide treatment alone can stabilize pre-existing mRNAs and thus mask their induction by 20E 61736. These transcripts would not be identified by our experiments. In addition, some 20E-inducible genes are expressed at higher levels in the absence of protein synthesis, due to the lack of 20E-induced repressors 617. The addition of cycloheximide thus provides a means of detecting 20E-regulated transcripts that might otherwise be missed. In this study, 743 20E-regulated genes were identified (Figure 2b), with 555 genes responding to 20E alone, 345 genes responding to 20E in the presence of cycloheximide (Additional data file 2), and 159 genes overlapping between these two datasets.

Comparison of the 20E-regulated genes to those genes that require EcR for their proper regulation at the onset of metamorphosis led to a final list of 20E-regulated, EcR-dependent genes. Only those genes that are upregulated by 20E in culture and downregulated in at least one of the EcRi time points, or downregulated by 20E in culture and upregulated in at least one of the EcRi time points, were considered for further analysis, leading to the identification of 479 genes (20E-final; Figure 2c). As depicted in Figure 2d, the majority of 20E-final genes that are upregulated by 20E are induced in -4 hour late larvae and/or early prepupae, in apparent response to the late larval 20E pulse, while many genes downregulated by 20E are repressed at these times. The downregulated 20E-final genes that peak in 4 to 6 hour prepupae could be repressed by 20E and thus expressed during this interval of low 20E titer.

We compared our EcR-dependent genes and the 20E-final gene set to data from two microarray studies that examined 20E-regulated biological responses - either EcR-dependent genes expressed in the larval midgut at pupariation 20 (Table 1, rows A and B), or changes in gene expression that occur during 20E-induced larval salivary gland cell death 37 (Table 1, rows C and D). As expected, many genes that are normally downregulated in the midgut at pupariation are upregulated in our EcRi gene set (113 genes; Table 1, row B), and genes that are normally upregulated in the midgut at pupariation are downregulated in our EcRi gene set (120 genes; Table 1, row A). Similarly, we see significant overlaps between our 20E-final set and midgut genes that change their expression at pupariation (65 genes upregulated and 10 genes downregulated; Table 1, rows A and B). Statistically significant overlaps were also observed with genes that change their expression during salivary gland cell death, consistent with a critical role for 20E in directing this response 37 (Table 1, rows C and D). These correlations validate our datasets and support the conclusion that our results represent 20E responses in multiple tissues at the onset of metamorphosis.

Those genes that are upregulated by 20E twofold or higher and dependent on EcR are listed in Table 2. An examination of this list reveals several known key mediators of 20E signaling during development. These include three classic ecdysone-inducible puff genes, E74A, E75, and E78 7, as well as Kr-h1, which encodes a family of zinc finger proteins required for metamorphosis 38, the DHR3 nuclear receptor gene 39, and Cyp18a1 17. Expanding our list by including all 20E-regulated genes (Additional data file 2), results in the identification of the DHR39, DHR78, and FTZ-F1 nuclear receptor genes, as well as the L71 (Eip71E) late genes, IMP-E2, IMP-L3, Fbp-2, Sgs-1, urate oxidase, and numerous genes identified in other studies as changing their expression at the onset of metamorphosis 202140. The identification of well-characterized 20E-regulated genes within our datasets suggests that the other genes in these lists are also likely to function in 20E signaling pathways, and thus provide a foundation to extend our understanding of 20E action in new directions.

In an effort to identify biological pathways that might respond to 20E at the onset of metamorphosis, we compared our EcRi and 20E-final datasets with published microarray studies of circadian rhythm, starvation, stress, and immunity 4142434445. No statistically significant overlaps were seen with the circadian rhythm gene sets examined; however, we did observe significant overlaps with genes that are expressed during starvation, stress, or an innate immune response. For the starvation response, we examined genes that change their expression upon starvation for 4 hours or starvation in the presence of sugar for 4 hours 45. We observed 120 genes induced under these conditions that are upregulated in EcRi animals, and 90 genes that are repressed upon starvation and downregulated in EcRi animals (Table 1, rows E and F). As shown in Figure 3a, the starvation-regulated genes are part of an EcR-dependent switch that occurs at puparium formation, where many of the induced genes are normally downregulated at puparium formation, and many starvation-repressed genes are upregulated at puparium formation. These genes include eight members of the cytochrome P450 family, three triacylglycerol lipase genes, α-trehalose-phosphate synthase, and a fatty-acid synthase gene that are downregulated at the onset of metamorphosis, while lipid storage droplet-1, pumpless, a UDP-galactose transporter, a lipid transporter, and phosphofructokinase are upregulated at this stage. Similarly, genes that change their expression in response to oxidative or endoplasmic reticulum stress 43 are significantly upregulated in EcRi animals at puparium formation (Table 1, rows G and H), reflecting their normal coordinate downregulation at puparium formation (Figure 3b), and demonstrating that this response is mediated by EcR. Within the 87 genes that overlap between the downregulated stress response genes and the upregulated EcR-dependent genes, we identified 14 of the 17 Jonah genes that encode a family of coordinately regulated midgut-specific putative proteases 46. Six genes that encode trypsin family members are also within this gene set, indicating that many peptidase family members are regulated by EcR. Taken together with the data on EcR-regulated starvation genes, these results indicate that EcR plays a central role in controlling metabolic responses at pupariation, directing the change from a feeding growing larva to an immobile non-feeding pupa.

Genes that change their expression upon microbial infection 4447 are also significantly upregulated in EcRi animals at puparium formation (Table 1, rows I and J), and coordinately downregulated at pupariation (Figure 3c). Interestingly, we identified both the Toll ligand-encoding gene dorsal and the key Toll effector gene spätzle as downregulated at the onset of metamorphosis in a EcR-dependent manner, suggesting that central regulators of the Toll-mediated immune response pathway are under EcR control 48. In addition, well studied immune response genes are downregulated by 20E, including Cecropin C, Attacin A, Drosocin, Drosomycin, and Defensin (Additional data file 2). These observations indicate that many metabolic and immunity-regulated genes are part of the genetic program directed by 20E at the onset of metamorphosis, and that these genes are normally coordinately downregulated at puparium formation in an EcR-dependent manner.

We selected all potential transcriptional and translational regulators from the list of most highly induced 20E primary-response genes that are EcR-dependent (Table 2) and not yet implicated in 20E signaling pathways, identifying seven genes: sox box protein 14 (sox14), cabut, CG11275, CG5249, vrille, hairy, and brain tumor (brat). Northern blot hybridization was used to validate the transcriptional responses of these genes to 20E (Figure 4a). All seven genes are induced by 20E in larval organ culture, with CG5249 displaying a very low level of expression and hairy showing only a modest approximately twofold induction. Several transcripts are increased upon treatment with cycloheximide alone, consistent with its known role in stabilizing some mRNAs 61736. Addition of 20E and cycloheximide, however, resulted in higher levels of transcript accumulation, similar to the response seen when E74A is used as a control 649. Their temporal patterns of expression at the onset of metamorphosis also reveal brief bursts of transcription that correlate with the 20E pulses that trigger puparium formation and adult head eversion (Figure 4b). These seven genes thus appear to represent a new set of 20E primary-response regulatory genes that could act to transduce the hormonal signal during metamorphosis.

We examined roles for brat during metamorphosis because, unlike the other six 20E primary-response genes described above, a brat mutant allele is available (brat^k06028) that allows an assessment of its functions during later stages of development 2550. The brat^k06028 P-element maps to the fourth exon of the brat gene. Precise excisions of this transposon result in viable, fertile animals, demonstrating that the transposon is responsible for the mutant phenotype 25. Lethal phase analysis of brat^k06028 mutants revealed that 61% of the animals survive to pupariation, with the majority of these animals pupariating 1 to 2 days later than their heterozygous siblings (n = 400). Of those mutants that pupariated, 11% died as prepupae, 8% died as early pupae, 46% died as pharate adults, and the remainder died within a week of adult eclosion. Phenotypic characterization of brat^k06028 mutant prepupae and pupae revealed defects in several ecdysone regulated developmental processes, including defects in anterior spiracle eversion (29%; Figure 5b–d), malformed pupal cases (15%, Figure 5b–d), and incomplete leg and wing elongation (12%). Northern blot hybridization of RNA isolated from staged brat^k06028 mutant third instar larvae (Figure 5e, -18 and -4 hour time points) or prepupae (Figure 5e, 0 to 12 hours) revealed a disruption in the 20E-regulated transcriptional hierarchy. In wild type animals, brat mRNA is induced in late third instar larvae and 10 hour prepupae, similar to the temporal profile determined by microarray analysis (Figures 4b and 5e), with reduced levels of brat mRNA in brat^k06028 mutants, consistent with it being a hypomorphic allele 25. βFTZ-F1 is unaffected by the brat mutation in mid-prepupae, while E74 mRNA is reduced at 10 hours after pupariation (Figure 5e). BR-C, E93, EcR, DHR3, and L71-1 are expressed at higher levels in late third instar larvae and early prepupae (Figure 5e), with significant upregulation of BR-C. In addition, the smallest BR-C mRNA, encoding the Z1 isoform, is under-expressed in brat mutant prepupae (Figure 5e). It is unlikely that brat exerts direct effects on transcription since it encodes a translational regulator 26. Nonetheless, these effects on 20E-regulated gene expression are consistent with the late lethality of brat^k06028 mutants. In particular, the rbp function provided by the BR-C Z1 isoform is critical for developmental responses to 20E, and overexpression of BR-C isoforms can lead to lethality during metamorphosis 1351. Thus, not only are the brat mutant phenotypes consistent with it playing an essential role during metamorphosis, but it may exert this function through the regulation of key 20E-inducible genes. Efforts are currently underway to address the roles of the remaining six new 20E primary-response regulatory genes in transducing the hormonal signal at the onset of metamorphosis.

w¹¹¹⁸ animals were used for phenotypic, array, and Northern blot studies. brat^k06028/CyO, kr-GFP was used to analyze brat function. Third instar larvae were staged by the addition of 0.05% bromophenol blue to the food as previously described 53, or synchronizing animals at pupariation. brat^k06028 animals were identified by the loss of the kr-GFP marker associated with the CyO balancer chromosome. For EcR RNAi, hs-EcRi-11 third instar larvae were heat-treated twice at 37°C, each time for 1 hour, at 24 hours and 18 hours prior to pupariation, as described 29. RNA was harvested for microarray analysis at -4, 0, and 4 hours relative to pupariation from three independent collections of animals for each time point.

Partial blue gut third instar larvae were staged by the addition of 0.05% bromophenol blue to the food, as previously described 53. Eight animals were dissected in each well of a nine-well glass dish (Corning, Corning, NY, USA) and cultured in approximately 100 μl oxygenated Schneiders Drosophila Medium (Invitrogen, Carlsbad, CA, USA) at 25°C. Cultures were incubated in a styrofoam box under a constant flow of oxygen. Following an initial incubation of 1 hour, the medium was removed and replaced with either fresh Schneiders Drosophila Medium (no hormone), medium plus 8.5 × 10^-5 M cycloheximide (Sigma-Aldrich, St. Louis, MO, USA), medium plus 5 × 10^-6 M 20-hydroxyecdysone (Sigma), or medium plus cycloheximide and 20E, each for 6 hours at 25°C. Organs were collected and RNA was extracted as described below. All experiments were done in triplicate and harvested separately for microarray analysis.

All experiments for microarray analysis were performed independently, in triplicate, to facilitate statistical analysis. Total RNA was isolated using TriPure (Roche, Indianapolis, IN, USA) followed by further purification with RNAeasy columns (Qiagen, Valencia, CA, USA). Probe labeling, hybridization to Affymetrix GeneChip^® Drosophila Genome Arrays (Affymetrix, Santa Clara, CA, USA), and scanning, were performed by the University of Maryland Biotechnology Institute Microarray Core Facility. dChip1.2 was used to normalize the raw data and determine gene expression values 54. Statistically significant changes between sample sets were identified using significance analysis of microarray (SAM) with a delta value to give a <10% false discovery rate 55. Further analysis and comparisons between datasets were performed using Access (Microsoft Corporation, Redmond, WA, USA). Cluster analysis was performed using dChip1.2. A cutoff of 1.3-fold change in expression level was used to restrict the 20E-regulated gene set (organ culture data), with a 1.5-fold cutoff for the EcRi gene sets. These fold cutoffs were chosen in order to restrict the datasets to those genes that are most significantly affected by 20E and EcRi. The lower fold cutoff for the organ culture data reflects the observation that 20E responses tend to be reduced in organ culture when compared to the intact animal 17 (unpublished results). This is, most likely, due to the stress of dissection and in vitro culture. Microarray data from this study can be accessed at the National Center for Biotechnology Information Gene Expression Omnibus website 56, with accession numbers as follows: GSE3057 for the temporal array series; GSE3060 for the organ culture array series; and GSE3069 for the EcRi array series.

Comparisons were made between our microarray datasets and previously published microarray datasets using Microsoft Access. Each dataset was split into genes that are either upregulated or downregulated, represented by up or down arrows in Table 1. The EcRi datasets were constrained to at least a twofold change in expression level in order to reduce the 4,188 genes to more manageable numbers, resulting in 634 upregulated genes and 924 downregulated genes (Table 1). For the stress-regulated genes, only those genes that show at least a twofold change in expression with either paraquot, tunicamycin, or H₂O₂ treatment were used for comparison 43. The 400 immune genes were split into 221 genes that showed consistent upregulation and 148 genes that showed consistent downregulation, eliminating 31 genes that showed an inconsistent profile under the conditions tested 4447. All other microarray datasets used for our comparison studies were as published (see Table 1 legend).

Total RNA was isolated using Trizol (Gibco) from staged animals or organ cultures, fractionated by formaldehyde gel electrophoresis, transferred to nylon membranes, and probed with radioactively labeled probes 27. To facilitate comparisons, w¹¹¹⁸ blots and brat^k06028 blots where probed, washed, and exposed together. rp49 was used as a loading control. Probes were generated by PCR from genomic DNA using the following pairs of primers: hairy forward (F), 5'CAAATTGGAAAAGGCCGACA3'; hairy reverse (R), 5'AGAGAAACCCTAAGCGGCTT3'; cabut F, 5'CTCTTCTAGTAGCCAAGACG3'; cabut R, 5'GAGATTGGTTCTGATGCTGC3'; sox box protein 14 F, 5'TCAGCAAGGACGATAAGCAGC3'; sox box protein 14 R, 5'AGCTCCGTTGTTATCGTGTGC3'; CG5249 F, 5'ACGATGTGGATCCTGAGACG3'; CG5249 R, 5'GCTCCATCATCAGCATGTGC3'; CG11275 F, 5'ACACAGATTGCGTGTTCCACG3'; CG11275 R, 5'TGGACAACGTGACTCCATACG5'; brain tumor F, 5'TCTCCACGAACTGGAGAACG3'; brain tumor R, 5'TGATGGTGTGACTGTTGGTGG3'; vrille F, 5'ACAGTTGTTGGCATCGCTGC3'; vrille R, 5'GACAACAAGAAGGACGAGAGC3'.