The Drosophila nuclear receptors DHR3 and βFTZ-F1 control overlapping developmental responses in late embryos

Animals develop through a series of distinct life stages separated by developmental transitions, including birth and sexual maturation. In Drosophila, pulses of the steroid hormone 20-hydroxyecdysone (20E) direct each of the major developmental transitions in the life cycle, including molting and metamorphosis. Each pulse of 20E is followed by a stereotypic sequence of 20E-responsive transcription factors. The regulatory interactions between these transcription factors have been elucidated through detailed studies of the onset of metamorphosis, when the late larval 20E pulse triggers pupariation (Fig. 1A, reviewed in Henrich et al., 1999; Riddiford et al., 2001; Thummel, 2001). The 20E signal, acting through its EcR/USP receptor, directly induces E74A and E75A transcription, and upregulates the early-late gene DHR3 synergistically with 20E-induced protein synthesis. This additional level of regulation leads to DHR3 accumulation at later times than transcription of direct primary-response genes such as E74A and E75A (Horner et al., 1995). DHR3 then directly induces βFTZ-F1 expression in mid-prepupae (Kageyama et al., 1997; Lam et al., 1997; White et al., 1997). βFTZ-F1, in turn, functions as a competence factor for genetic responses to the prepupal 20E pulse, which occurs ~10 hours after puparium formation, directing adult head eversion and the prepupal-pupal transition (Woodard et al., 1994; Broadus et al., 1999; Yamada et al., 2000). These responses include reinduction of E74A and E75A, as well as stage-specific expression of the 20E primary-response gene E93 (Thummel, 2001). These 20E-regulated genes are all widely expressed, suggesting that their primary function is to transduce temporal information provided by the sequential pulses of 20E. E74A encodes an ETS-domain transcription factor, whereas EcR, usp, E75A, DHR3 and βFTZ-F1 all encode members of the nuclear receptor superfamily. Nuclear receptors are defined by a zinc-finger DNA-binding domain and a C-terminal ligand-binding domain that can interact with small lipophilic compounds. Within this group of receptors, however, only EcR and E75A have known ligands (20E and heme/monoatomic gases, respectively; Koelle et al., 1991; Reinking et al., 2005).

Although the regulatory interactions between DHR3, βFTZ-F1, E74A and E75A at the onset of metamorphosis have been well defined, their individual developmental functions remain less clear. One model postulates that a central role for this transcriptional cascade is to provide an appropriate delay in βFTZ-F1 expression, enabling it to regulate responses to the subsequent prepupal pulse of 20E (White et al., 1997; Lam et al., 1999). An alternative model is that the sequential expression of DHR3 and βFTZ-F1 directs successive temporal programs of gene expression that specify appropriate progression through prepupal development, similar to the genetic network that controls patterning of the Drosophila anteroposterior axis in early embryos (Arnone and Davidson, 1997). The similarity between EcR, DHR3 and βFTZ-F1 mutant phenotypes during metamorphosis, which include lethality at the prepupal-pupal transition with defects in head eversion and gas bubble translocation, supports the first model (Bender et al., 1997; Lam et al., 1997; Yamada et al., 2000). This interpretation, however, is complicated by the use of hypomorphic alleles or rescuing constructs that are required to overcome the early lethality associated with null mutations in EcR, DHR3 and βFTZ-F1. Complete loss of function in any of these genes results in lethality during embryogenesis (Bender et al., 1997; Carney et al., 1997; Yamada et al., 2000). Moreover, DHR3, βFTZ-F1, E74A and E75A are sequentially expressed during mid-embryogenesis in a pattern that is essentially identical to that seen in prepupae, starting a few hours after a high titer 20E pulse that occurs approximately 8 hours after egg laying (AEL) (Maroy et al., 1988; Sullivan and Thummel, 2003). These observations suggest that investigating the functions of DHR3 and βFTZ-F1 during embryonic development might help to discriminate between these two proposed models, and will provide further insights into the regulation and function of this transcriptional cascade during development.

Null mutations in the genes that encode components of the 20E biosynthetic pathway result in embryonic lethality, with defects in head involution, dorsal closure, gut development and cuticle deposition (Chavez et al., 2000; Ono et al., 2006). Inactivating EcR function leads to similar embryonic lethal phenotypes, although with earlier lethality and more severe defects, including aberrant germ band retraction (Bender et al., 1997; Kozlova and Thummel, 2003). Null mutations in either DHR3 or βFTZ-F1 also lead to fully penetrant embryonic lethality, although their developmental phenotypes are less well defined. DHR3 null mutants die at the end of embryogenesis with relatively minor developmental defects (Carney et al., 1997). Approximately 50% of the embryos contain peripheral nervous system (PNS) defects, including loss of neurons, pathfinding and fasciculation defects, cluster organization defects, and displacement of neurons to inappropriate positions within the PNS. DHR3 mutants also display a low frequency of defects in muscle patterning (5-15%). These defects, however, are unlikely to account for the embryonic lethality of DHR3 mutants, as they are only partially penetrant and other mutants with similar phenotypes die later during development (Kolodkin et al., 1993). The FTZ-F1 locus encodes two protein isoforms, α and β, which share a common C-terminal region with both a DNA-binding and ligand-binding domain (Lavorgna et al., 1991). αFTZ-F1 is maternally deposited and functions as a cofactor for the fushi tarazu (ftz) segmentation gene (Guichet et al., 1997; Yu et al., 1997). Consistent with this, FTZ-F1 maternal mutants die early during embryogenesis, with segmentation defects that resemble those of ftz mutants. In contrast, the βFTZ-F1 isoform is zygotically expressed during late embryonic, larval and pupal stages. βFTZ-F1 zygotic null mutants die during embryogenesis, with defects that remain to be characterized (Yamada et al., 2000).

We show that the regulatory interactions between DHR3, βFTZ-F1, E74A and E75A during embryogenesis are identical to those defined at the onset of metamorphosis. In addition, DHR3 and βFTZ-F1 null mutants not only exhibit common phenotypes, including a failure to fill their tracheal system with air, but also display gene-specific effects, with βFTZ-F1 controlling the size and pigmentation of denticles, and DHR3 directing proper ventral nerve cord (VNC) condensation. A microarray analysis of DHR3 mutant embryos provides a molecular context for these phenotypic effects, revealing genes that are regulated by either DHR3 and βFTZ-F1, or by both receptors together. This study defines the lethal phenotypes of DHR3 and βFTZ-F1 mutants, and provides evidence for functional bifurcation in the 20E-responsive transcriptional cascade.

Fly stocks were maintained at 18-25°C on standard cornmeal-agar-yeast food. The w¹¹¹⁸ strain was used as a control unless noted otherwise. dib^F8 is a null allele described in Chavez et al. (Chavez et al., 2000). Two DHR3 null alleles were used in this study: w⁻; DHR3^G60S and w⁻; DHR3^22-35 (Carney et al., 1997). DHR3^G60S encodes a protein with a single amino acid change at a highly conserved position within the DNA-binding domain, whereas DHR3^22-35 has a rearrangement that involves the exon that encodes the DNA-binding domain. Mutant phenotypes were examined in animals carrying a DHR3 mutation in combination with Df(2R)12, a small X-ray-induced deficiency that removes the DHR3 locus (Weber et al., 1995). FTZ-F1^ex7 is an embryonic lethal allele that was generated by P-element excision from a maternal effect FTZ-F1 mutation (Guichet et al., 1997; Yamada et al., 2000). Two heat-inducible rescue constructs were used: P[w⁺; hs-DHR3] (Lam et al., 1999) and P[y⁺w⁺; hs-βFTZ-F1] (Yamada et al., 2000), neither of which express detectable amounts of protein in the absence of heat shock. A breathless-GAL4 transgene (a gift from M. Metzstein) and elav-GAL4 were used to drive UAS-GFP expression in the developing trachea and nervous system, respectively. Balancer chromosomes marked with Dfd-GMR-YFP and twi-Gal4 UAS-GFP were used to identify mutant embryos (Halfon et al., 2002; Le et al., 2006).

For hs-βFTZ-F1 rescue experiments, flies of the following genotype were crossed: DHR3/CyO, P[Dfd-GMR-YFP]; P[y⁺w⁺, hs-βFTZ-F1] with DHR3/CyO twi-Gal4 UAS-GFP, and βFTZ-F1^ex7 P[y⁺w⁺, hs-βFTZ-F1]/TM3, P[Dfd-GMR-YFP] with βFTZ-F1^ex7/TM3 twi-Gal4 UAS-GFP. Four-hour egg lays were collected on molasses agar plates supplemented with yeast paste and allowed to develop at 25°C for 12 hours. The dishes were floated in a water bath at 35°C for one hour, after which mutant embryos were transferred to fresh plates. Rescue was assessed by counting living first-instar larvae on the plate two days AEL. Similarly, rescue of DHR3 lethality by hs-DHR3 was tested in embryos derived from DHR3/CyO, P[Dfd-GMR-YFP]; P[w⁺, hs-DHR3]/TM3, Sb parents. Embryos were collected, aged 6-10 hours AEL, and heat shocked at 35°C for 30 mins.

Control embryos with air-filled tracheae and mutant embryos at least 27 hours AEL were dechorionated, hand picked and mounted in a drop of cuticle-mounting medium: 50 μl glacial acetic acid, 25 μl CMCP10 (Masters Co.), 25 μl 85% lactic acid. The slides were incubated overnight at 60°C with a weight on the coverslip, and observed under dark field and DIC optics on a Zeiss Axioskop 2 microscope.

Developing tracheae were labeled using rhodamine-conjugated chitin-binding protein (CBP) at 1:500 dilution (New England Biolabs). GFP was detected using a rabbit anti-GFP primary antibody (1:500, MBL). A rabbit polyclonal anti-DHR3 antibody was used at 1:50 dilution (Lam et al., 1997). Secondary antibodies were Cy2- and Cy3-conjugated anti-rabbit and anti-mouse antibodies (1:200, Jackson ImmunoResearch). Staged embryos were stained as described by Reuter and Scott (Reuter and Scott, 1990), except that the primary antibody and CBP incubations were performed at room temperature. Embryos were visualized using confocal microscopy.

RNA was isolated from 16-20 hour AEL w¹¹¹⁸ and DHR3^G60S/Df(2R)12 embryos using TriPure isolation reagent (Roche) and purified on RNeasy columns (Qiagen). All samples were prepared in four replicates to facilitate subsequent statistical analysis. Probe labeling, hybridization to two-color Agilent Drosophila 44K arrays and scanning were performed by the University of Utah Microarray Core Facility. The data were quantile normalized using R, and the fold changes in gene expression and t-statistics were determined using GeneSifter (VizX Labs, Seattle, WA). p-values were calculated using the Benjamimi and Hochberg correction for false-discovery rate. Comparison between microarray datasets was performed using Microsoft Access. Microarray data from this study can be accessed at NCBI GEO (accession number: GSE18577). Northern blot analysis was conducted essentially as described by Karim and Thummel (Karim and Thummel, 1991).

Northern blot hybridization of staged embryos revealed that E75A is expressed in synchrony with the 20E pulse that occurs ~8 hours AEL (Fig. 1B) (Maroy et al., 1988). This is followed by transient bursts of DHR3 and βFTZ-F1 expression, and then co-expression of E74A and E75A in late embryos (Fig. 1B). A similar result was seen in an independent study (Sullivan and Thummel, 2003). These temporal profiles of expression raise the possibility that a functional transcriptional cascade triggered by 20E and involving DHR3 and βFTZ-F1 could play a role in embryonic development. As a first step toward testing this hypothesis, we examined DHR3 expression in mutants deficient for 20E synthesis. Embryos carrying a null mutation for disembodied (dib), a cytochrome P450 required for 20E synthesis (Chavez et al., 2000), were immunostained with anti-DHR3 antibodies. DHR3 expression is absent in stage 14 dib^F8 mutants when compared to a heterozygous control (Fig. 2A-B), indicating that embryonic DHR3 expression depends on 20E. We also analyzed the patterns of βFTZ-F1, E74A and E75A transcription in staged control embryos, DHR3 mutants and βFTZ-F1 mutants by northern blot hybridization (Fig. 2C). The DHR3^G60S null allele was used in combination with a deficiency that removes the DHR3 locus (Df(2R)12), together with animals homozygous for the FTZ-F1^ex7 mutation. βFTZ-F1 expression is reduced in DHR3 mutant embryos, similar to the effect seen in DHR3 mutant prepupae, and undetectable in FTZ-F1^ex7 mutants, consistent with the nature of this null allele. In addition, E74A and E75A expression is significantly reduced in both DHR3 and βFTZ-F1 mutants during late embryogenesis (Fig. 2C). Similar results were observed with a second null allele for DHR3, DHR3^22-35, in combination with the Df(2R)12 deficiency (data not shown). These results support the hypothesis that DHR3 is required for proper βFTZ-F1 induction in mid-embryos, and that the reduced expression of βFTZ-F1 in DHR3 mutants leads to reduced levels of E74A and E75A expression, paralleling the regulatory interactions seen in prepupae (Fig. 1A).

If the major role of DHR3 during embryogenesis is to ensure proper temporal expression of βFTZ-F1, then ectopic βFTZ-F1 expression should rescue the embryonic lethality of DHR3mutants. To test this hypothesis, we used an hs-βFTZ-F1 transgene to drive βFTZ-F1 expression in DHR3 mutant embryos during the short period of time when it is normally expressed. As expected, heat-induced βFTZ-F1 expression 6-10 hours AEL in βFTZ-F1 mutant embryos efficiently rescues their lethality (Fig. 3A), consistent with previous observations (Yamada et al., 2000). In contrast, no rescue was observed when the same treatment was applied to DHR3 mutant embryos (Fig. 3A). It is possible that the inability of βFTZ-F1 to rescue DHR3 mutants is due to lethality unrelated to the DHR3 mutation. Heat-induced DHR3 expression, however, is sufficient to rescue DHR3 mutants, indicating that this is not the case (Fig. 3B). Interestingly, no heat treatment is needed for this construct to rescue DHR3 lethality, demonstrating that basal DHR3 expression from this hs-DHR3 transgene is sufficient to direct proper embryonic development. Taken together, these data suggest that DHR3 exerts both βFTZ-F1-dependent and βFTZ-F1-independent functions to direct proper progression through embryogenesis.

As a first step toward identifying developmental functions for DHR3 and βFTZ-F1 during embryogenesis, we examined cuticle formation in embryos mutant for these two nuclear receptors. Genetic and biological studies have demonstrated that 20E signaling plays a central role in cuticle synthesis and deposition during each stage in the life cycle (Chavez et al., 2000; Riddiford et al., 2001; Kozlova and Thummel, 2003). Examination of cuticle preparations from DHR3 and βFTZ-F1 mutants at low magnification revealed normal patterning of the embryonic cuticle, with the correct number and position of the ventral denticle belts (Fig. 4A-C). The dorsal cuticle is also normal in DHR3 and βFTZ-F1 mutants, indicating that dorsal closure occurred properly. Mouth hooks and denticle belts are fully developed in control embryos by 19 hours AEL; however, this process is not complete in DHR3 and βFTZ-F1 mutants until a few hours later, indicating a developmental delay in cuticle differentiation (data not shown). In addition, the denticles in βFTZ-F1 mutants, but not DHR3 mutants, are abnormally small and unpigmented (Fig. 4D-F). The number of denticles per belt appears to be wild type in βFTZ-F1 mutant embryos, indicating that patterning is normal, but the denticles do not appear to fully differentiate. The specificity of this βFTZ-F1 phenotype was confirmed by observing the same defects in animals carrying the βFTZ-F1^ex7 mutation in combination with Df(3L)Cat^DH104, a deficiency that removes the FTZ-F1 locus (Broadus et al., 1999) (data not shown). The absence of a cuticular phenotype in DHR3 mutants was confirmed using a second allelic combination (DHR3^G60S/Df) and is in agreement with previous reports (Carney et al., 1997). Taken together, these results define a βFTZ-F1-specific function in controlling the size and pigmentation of ventral denticles.

To further define functions for DHR3 and βFTZ-F1 during embryonic development, we examined the major developmental events that occur after the peak expression of DHR3 and βFTZ-F1: tracheal system maturation, initiation of muscular movements and central nervous system (CNS) condensation, three processes that remain relatively poorly understood (Campos-Ortega and Hartenstein, 1985). Tracheal development begins during the second half of embryogenesis, when 20 metameric placodes invaginate from the epidermis and undergo stereotypic branching and fusion events to form the tracheal tubular network. These tubes form as liquid-filled structures, with the epithelial cells depositing an apical chitinous matrix into the lumen that coordinates uniform tube growth (reviewed by Swanson and Beitel, 2006). The tubular network fills with gas approximately one hour before hatching. This event is easy to follow using bright-field microscopy because of the different refractive index of the gas compared to the surrounding liquid. Examination of the tracheae in DHR3 and βFTZ-F1 late-stage mutant embryos revealed that air filling fails to occur in both genetic backgrounds, compared to the clearly visible air-filled tracheae present in controls (Fig. 5A-C). This phenotype is fully penetrant in animals of both genotypes (Table 1). The specificity of the DHR3 mutant phenotype was confirmed by using a second allele, DHR3^G60S, in combination with a deficiency that removes the DHR3 locus (Table 1). Interestingly, a gas bubble reminiscent of the first step of the air-filling process was observed in a few animals of this genotype (Table 1). Ectopic expression of βFTZ-F1 in DHR3^G60S mutants had no effect on the tracheal air-filling defect, suggesting that this represents a function of DHR3 that is not solely dependent on its downstream target, βFTZ-F1 (data not shown).

Failure to air fill the tracheal network could result from defects in earlier tracheal morphogenesis or from specific disruption of this late developmental event. To examine the tracheal morphology in DHR3 and βFTZ-F1 mutant embryos, we used rhodamine-conjugated CBP to specifically label the tracheal network. This showed that the branching pattern and size of the lumen in the developing tubes is indistinguishable from that seen in control animals (Fig. 5D-F). Closer examination of the tracheal cells, using btl-GAL4 to mark these cells with GFP in combination with staining for CBP, confirmed that the development of the tracheal network, including differentiation, lumen formation and extension, chitin deposition and branch patterning, is unaffected in both DHR3 and βFTZ-F1 mutants (Fig. 5G-L and data not shown). The air-filling defect observed in DHR3 and βFTZ-F1 mutant embryos therefore arises from a specific inability to perform the last steps of tracheal development required for air filling of the tracheal network.

Observation of late DHR3 and βFTZ-F1 mutant embryos also indicated that organogenesis is apparently normal. Specifically, the midgut displayed its normal looping pattern (data not shown) and the four differentiated Malpighian tubules were clearly visible (Fig. 5B-C, yellow arrowheads). In addition, most DHR3 and βFTZ-F1 mutant embryos exhibited normal muscular movements, as shown in Fig. 5B (red arrowhead) by the gap observed between the cuticle and epidermis (quantified in Table 1). These results indicate that development is not simply arrested in DHR3 and βFTZ-F1 mutant embryos, but rather that specific developmental processes are selectively disrupted.

Similar to the tracheal system, the CNS undergoes terminal differentiation during the final stages of embryonic development (Campos-Ortega and Hartenstein, 1985). The CNS acquires its overall final morphology during germ-band retraction, when the germ-band neuroblasts follow the movement of the epidermis to generate the VNC, a neuronal structure that underlies the entire ventral side of the embryo. The VNC later increases in cellular density and decreases in size in a process called condensation. This process starts at late stage 15, with the majority of condensation occurring during stage 16 and early stage 17 (Campos-Ortega and Hartenstein, 1985), after the peaks of DHR3 and βFTZ-F1 expression. To visualize VNC condensation in living embryos, we drove GFP expression specifically in the nervous system using an elav-GAL4 driver, and assessed anteroposterior condensation by measuring the percentage of the embryo length occupied by the VNC in control and mutant animals (Fig. 6A). Condensation results in a shortening of the VNC to an average value of 43% of embryo length at the end of stage 17 (air-filled tracheae) in control animals (Fig. 6C). Two DHR3 mutant combinations displayed low penetrance VNC condensation defects, with 25% of DHR3^G60S/Df embryos and 11% of DHR3^22-35/Df animals exhibiting incomplete condensation (Fig. 6B-C). In contrast, VNC condensation is essentially normal in βFTZ-F1 mutant embryos (Fig. 6C).

Microarray studies were performed to investigate the molecular mechanisms by which DHR3 contributes to late embryonic development. RNA was extracted from control and DHR3 mutant embryos staged 16-20 hours AEL, a time period following DHR3 and βFTZ-F1 expression, and during which the affected developmental processes occur. RNA was labeled and hybridized to two-color Agilent Drosophila 44K arrays. All experiments were conducted in four replicates to facilitate statistical analysis. The raw data were quantile normalized using R, and the fold changes in gene expression and t-statistics were determined using GeneSifter. We chose E74 as a positive control to determine the fold change cut-off and p-value for this analysis (no isoform-specific probe was available on the array for the other known DHR3-regulated gene, E75A). Our study revealed that 1352 transcripts are significantly affected in DHR3 mutant embryos (≥2.1-fold, adjusted P-value<0.03), with 746 genes upregulated and 606 genes downregulated (see Table S1 in the supplementary material).

We used the web-based software DAVID to identify gene functions (gene ontologies) corresponding to transcripts that are misregulated in DHR3 mutants (Dennis et al., 2003; Huang da et al., 2009). This analysis revealed that many genes involved in the innate immune and defense responses are expressed at increased levels in DHR3 mutants, together with reduced expression of genes encoding cytochrome P450 enzymes and proteolytic enzymes, and genes that act in metabolism (data not shown). This is in agreement with the functions of the DHR3 vertebrate homologs RORα and RORβ in controlling energy homeostasis and lipid metabolism in mice (Jetten, 2004). To determine whether metabolic dysfunction might contribute to the lethality of DHR3 mutant embryos, we measured the levels of the major forms of stored energy and ATP. No significant difference, however, was detected between mutant and control animals (see Fig. S1 in the supplementary material).

The gene ontology analysis also identified 128 genes associated with development that are misregulated in DHR3 mutants (see Table S2 in the supplementary material). These include several genes that have been well characterized as being directly induced by 20E, including E74A, E75 and IMPE3, which are expressed at reduced levels in DHR3 mutant embryos. An embryo-specific isoform of E93, another known 20E target gene, is normally repressed at the end of embryogenesis and is expressed at higher levels in DHR3 mutants (Baehrecke and Thummel, 1995). Additional targets of 20E signaling are found among the list of DHR3-regulated genes and display increased levels of expression in DHR3 mutants, including Fbp1, Sgs4, Sgs5 and Lsp1γ. We also see reduced expression (2.6-fold) of the key ecdysone biosynthetic gene phantom. Although the time point we examined is significantly later than the known 20E pulse in embryos, this observation nonetheless raises the possibility that DHR3 mutants may suffer from a hormone deficiency. To test this hypothesis, we determined whether exogenous 20E can rescue the embryonic lethality of DHR3 mutants. Although exogenous 20E application is sufficient to rescue embryos that are mutant for the ecdysone-synthesis gene spook (11 viable larvae out of 126 embryos, 9% rescue), consistent with previous studies (Ono et al., 2006), this approach had no effect on DHR3 mutants (0 viable larvae out of 319 embryos). This result suggests that hormone deficiency is not the sole cause of DHR3 embryonic lethality.

Northern blot hybridizations were conducted to test the response of selected DHR3-regulated genes in DHR3 and βFTZ-F1 mutant embryos (Fig. 7). This analysis identified three major classes of genes. First, some genes that are misregulated in DHR3 mutants are similarly affected in βFTZ-F1 mutant animals, such as retn (retained), E93 (Eip93F) and kkv (krotzkopf verkehrt). Second, E74A and E75A are submaximally expressed in DHR3 mutants, but are not detectably expressed in βFTZ-F1 mutant embryos. Finally, we identified one gene, Idgf5 (imaginal disc growth factor 5), that is downregulated in DHR3 mutants, but not significantly affected by the βFTZ-F1 mutation. Taken together, these results validate the DHR3 target genes identified in the microarray analysis, and provide a molecular basis to support the existence of overlapping developmental functions for DHR3 and βFTZ-F1 during embryogenesis.

Studies of the onset of metamorphosis have identified a transcriptional cascade that leads to the sequential expression of the transcription-factor-encoding genes DHR3, βFTZ-F1, E74A and E75A. Although the regulatory interactions between these genes have been characterized, the functional consequences of their expression remain more poorly understood. In this study, we show that the DHR3-βFTZ-F1 transcriptional cascade is functional in embryos and is required for progression through this stage of development. We describe the essential roles of DHR3 and βFTZ-F1, and demonstrate their contributions to both common and distinct developmental and transcriptional responses during embryogenesis. This study also reveals an evolutionarily conserved role for DHR3 and its vertebrate homolog, RORα, in nervous system development.

The regulatory interactions between DHR3, βFTZ-F1 and E74A/E75A that are described here in embryos are indistinguishable from those seen in prepupae. First, DHR3 expression in embryos is dependent on 20E signaling. Second, DHR3 mutants display reduced levels of βFTZ-F1, E74A and E75A expression at both stages in the life cycle, and βFTZ-F1 mutants have reduced levels of E74A mRNA and no detectable E75A expression (Broadus et al., 1999; Lam et al., 1999) (Fig. 2). Taken together with studies that show that ectopic βFTZ-F1 is sufficient to drive maximal expression of E74A and E75A (Woodard et al., 1994), our results indicate that DHR3 exerts its effect on these genes through its induction of βFTZ-F1 in embryos. Third, a loss of DHR3 function during embryogenesis does not eliminate βFTZ-F1 expression (Lam et al., 1999). This is probably due to other upstream factors that contribute to this response. One candidate for this function is the DHR4 nuclear receptor, which is coexpressed with DHR3 in both embryos and prepupae (Sullivan and Thummel, 2003). DHR4 mutants have no effect on DHR3 expression, but display significantly reduced levels of βFTZ-F1 mRNA in prepupae (King-Jones et al., 2005). These mutants, however, have no effect on embryonic development, suggesting that DHR4 does not play a major role in βFTZ-F1 induction at this early stage in the life cycle.

The late larval pulse of 20E both directly and indirectly induces DHR3 and represses βFTZ-F1 (Woodard et al., 1994; Horner et al., 1995). Taken together with the inductive effect of DHR3 on βFTZ-F1 expression, this regulation ensures that the peak of βFTZ-F1 expression will be delayed until the proper time during development (White et al., 1997; Lam et al., 1999). The observation that the embryonic 20E pulse, at ~8 hours AEL, immediately precedes DHR3 expression suggests that similar regulatory interactions are acting in embryos. However, unlike prepupae, there is no known hormone peak in late embryos that could account for the coordinated induction of E74A and E75A mRNA at this time, as is known to occur in late prepupae (Thummel, 2001). It is possible that these transcripts are fully dependent on trans-acting factors such as βFTZ-F1 for their expression in embryos. Alternatively, these 20E primary-response genes might be induced by a novel temporal signal that remains to be identified.

It is interesting to note that a similar temporal profile of DHR3, βFTZ-F1 and E74A/E75A expression is also seen in larvae. A burst of DHR3 expression in mid-second instar larvae immediately follows the peak in the 20E titer and precedes the transient expression of βFTZ-F1, which is followed by co-expression of E74A and E75A at the end of the instar (Sullivan and Thummel, 2003). Curiously, E75A, but not E74A, is expressed at an earlier time as well, in apparent synchrony with the 20E pulse, recapitulating the timing seen in embryos (Sullivan and Thummel, 2003) (Fig. 1B). It is thus likely that a common set of regulatory interactions function in both embryos and larvae to dictate the precise timing of these expression patterns at each stage in the life cycle, prior to the third instar. Moreover, the observation that EcR, E75A and βFTZ-F1 mutants display defects in larval molting indicates that their expression is essential for proper progression through these stages in development (Li and Bender, 2000; Yamada et al., 2000; Bialecki et al., 2002).

DHR3 and βFTZ-F1 null mutations lead to fully penetrant embryonic lethality, with relatively minor and partially penetrant phenotypes reported in DHR3 mutant embryos and no phenotypic description of βFTZ-F1 mutant embryos (Carney et al., 1997; Yamada et al., 2000). The studies described here define both common and unique functions for these two nuclear receptors during embryogenesis. DHR3 and βFTZ-F1 null mutants both display a highly penetrant defect in air filling of the tracheal tree (Fig. 5). In addition to this common function, βFTZ-F1 is required for the proper differentiation of the denticles in the ventral cuticle and DHR3 is required for VNC condensation (Figs 4, 6). Both DHR3 and βFTZ-F1 mutants display apparently normal muscle movements at the end of embryogenesis, indicating that only some developmental responses are blocked at this stage (Table 1). These processes of cuticle differentiation, tracheal air filling, muscular movements and VNC condensation represent the major developmental events that can be described in late embryos. Defects in three of these four pathways thus define a central role for DHR3 and βFTZ-F1 in late embryonic development. In addition, unlike prepupae, in which DHR3 and βFTZ-F1 mutants have essentially identical phenotypes, these studies establish independent functions for these two nuclear receptors during development. Together with the previously identified early embryonic roles of the 20E receptor EcR in dorsal closure, head involution and midgut morphogenesis, these data indicate that each step of the 20E-induced transcriptional cascade controls sequential developmental programs during embryogenesis (Chavez et al., 2000; Kozlova and Thummel, 2003). Moreover, as mentioned above, the observation that this transcriptional cascade is also required for larval molting suggests that it represents a stereotypic 20E response that is required for progression through each major transition in the life cycle.

Ectopic expression of wild-type βFTZ-F1 is sufficient to rescue the lethality of βFTZ-F1 mutants, but has no effect on the viability of DHR3 mutants, indicating that DHR3 exerts essential functions independently of its downstream partner (Fig. 3). The causes of lethality in DHR3 and βFTZ-F1 mutant embryos, however, remain unclear. Strong loss-of-function mutations in the signal peptide peptidase (Spp) gene result in tracheal air-filling defects; however, Spp mutant embryos hatch normally and die as first or second instar larvae (Casso et al., 2005). Similarly, embryos with severe defects in VNC condensation can hatch into first instar larvae and survive to later stages of development (Page and Olofsson, 2008). These results indicate that the lethality of DHR3 and βFTZ-F1 mutant embryos cannot be directly attributed to defects in these pathways. Rather, DHR3 and βFTZ-F1 may participate in a developmental checkpoint necessary to trigger the last steps of embryogenesis required for hatching and survival.

Our microarray study revealed that a number of 20E-responsive genes are misregulated in DHR3 mutants, consistent with studies in prepupae that indicate a crucial role for DHR3 in 20E signaling (Lam et al., 1999). The microarray analysis also identified several genes that are involved in chitin metabolism and protein secretion, which could account for the defects in tracheal gas filling seen in DHR3 mutants. These included the chitinase genes Idgf5 (−8.6-fold) and kkv (+2.4-fold) (Devine et al., 2005), the CBP Cht12 (+2.6-fold) and the COPII coat subunit sec13 (+2.5-fold) (Tsarouhas et al., 2007). This study also identified a number of genes that play a role in axon guidance (Table 2). Interestingly, most of these genes have dose-dependent effects, whereby either reduced or increased expression can disrupt nervous system development (see, for example, Winberg et al., 1998). Failure of DHR3 mutant embryos to express these genes at normal levels could thus contribute to the PNS defects described previously (Carney et al., 1997).

Northern blot hybridization studies to examine the effects of DHR3 and βFTZ-F1 mutants on selected DHR3-regulated genes confirm and extend our phenotypic studies of these mutants. Some genes, such as retn, E93 and kkv, display similar transcriptional responses in DHR3 and βFTZ-F1 mutants, whereas E74A and E75A are more significantly affected in βFTZ-F1 mutants and Idgf5 is selectively reduced in DHR3 mutants (Fig. 7). These transcriptional effects support our phenotypic studies and provide further evidence that DHR3 and βFTZ-F1 exert common and independent regulatory roles during embryogenesis. This conclusion is consistent with experimental and theoretical studies of gene regulatory networks, which indicate that transcriptional cascades provide an effective means of amplifying signals and integrating multiple cues to provide specificity in biological responses. Transcriptional cascades can also direct temporal programs of successive gene expression, as observed in the formation of flagella in Escherichia coli (Kalir et al., 2001) and the specification of anteroposterior patterning in the Drosophila embryo (Arnone and Davidson, 1997). In addition, the DHR3-βFTZ-F1 transcriptional cascade involves nuclear receptors that could potentially act as ligand-regulated transcription factors, introducing an additional level of control by small lipophilic compounds. These observations support the proposal that the sequential expression of DHR3 and βFTZ-F1 at multiple stages of development can specify successive biological programs that promote appropriate progression through the life cycle. By combining insect endocrinology with the predictive power of genetics, the 20E-triggered transcriptional cascades in Drosophila provide an ideal context to define how a repeated systemic signal can be refined into precise stage-specific temporal responses during development.

DHR3 is required for VNC condensation, a terminal step in embryonic nervous system morphogenesis that is dependent on nervous system activity, glial cell function and apoptosis (Page and Olofsson, 2008). In addition, previous studies have identified roles for DHR3 in PNS development (Carney et al., 1997). Interestingly, these functions, which are specific for DHR3 and are not shared with its direct target, βFTZ-F1, parallel the role of the mammalian DHR3 homolog RORα in brain development. RORα was initially identified as the gene associated with the spontaneous staggerer mutation in mice, which display ataxia associated with cerebellum developmental defects and degeneration (Hamilton et al., 1996; Gold et al., 2007). The cerebellum in staggerer mutants is dramatically smaller than in controls, containing fewer of the two major cell types: granule cells and Purkinje cells (Sidman et al., 1962). Further investigation showed that this phenotype arises primarily from reduced expression in Purkinje cells of Sonic hedgehog (Shh), a mitogenic signal for granule cells (Gold et al., 2003). These data support the hypothesis that there is an evolutionarily conserved role for the ROR/DHR3 family of nuclear receptors in nervous system development and suggest that further functional studies of DHR3 may provide new insights into its ancestral functions in this pathway.

We thank H. Ueda for providing the βFTZ-F1^ex7 mutant and the hs-FTZ-F1 transformant line, as well as the Bloomington Drosophila Stock Center for several stocks. We thank the University of Utah Microarray Core Facility and B. Milash for help with analyzing the microarray data, J. Evans for technical help, M. Metzstein for strains and advice, and M. Horner, D. Seay and J. Tennessen for critical reading of the manuscript. A.F.R. was supported by a Fondation pour la Recherche Médicale postdoctoral fellowship. This research was supported by National Institute of Health grant 1R01DK075607. Deposited in PMC for release after 12 months.