Temporal profiles of nuclear receptor gene expression reveal coordinate transcriptional responses during Drosophila development.

The recent completion of the Drosophila genome sequence revealed 21 members of the nuclear receptor superfamily. Many of these genes are transcriptionally regulated by the steroid hormone ecdysone and play a role during the onset of metamorphosis, including the EcR/USP ecdysone receptor heterodimer. As a first step toward a genomic analysis of this gene family, we have characterized the temporal patterns of expression for all detectable nuclear receptor transcripts throughout major ecdysone-regulated developmental transitions in the life cycle: embryogenesis, a larval molt, puparium formation, and the prepupal-pupal transition. We find an unexpected close temporal relationship between DHR3, E75B, and betaFTZ-F1 expression after each major ecdysone pulse examined, reflecting the known cross-regulatory interactions of these genes in prepupae and suggesting that they act together at other stages in the life cycle. In addition, E75A, E78B, and DHR4 are expressed in a reproducible manner with DHR3, E75B, and betaFTZ-F1, suggesting that they intersect with this regulatory cascade. Finally, we find that known ecdysone-inducible primary-response transcripts are coordinately induced at times when the ecdysteroid titer is low, implying the existence of novel, as yet uncharacterized, temporal signals in Drosophila.

The recent completion of the Drosophila genome sequence revealed 21 members of the nuclear receptor superfamily. Many of these genes are transcriptionally regulated by the steroid hormone ecdysone and play a role during the onset of metamorphosis, including the EcR/USP ecdysone receptor heterodimer. As a first step toward a genomic analysis of this gene family, we have characterized the temporal patterns of expression for all detectable nuclear receptor transcripts throughout major ecdysone-regulated developmental transitions in the life cycle: embryogenesis, a larval molt, puparium formation, and the prepupal-pupal transition. We find an unexpected close temporal relationship between DHR3, E75B, and ␤FTZ-F1 expression after each major ecdysone pulse examined, reflecting the known cross-regulatory interactions of these genes in prepupae and suggesting that they act together at other stages in the life cycle. In addition, E75A, E78B, and DHR4 are expressed in a reproducible manner with DHR3, E75B, and ␤FTZ-F1, suggesting that they intersect with this regulatory cascade. Finally, we find that known ecdysone-inducible primaryresponse transcripts are coordinately induced at times when the ecdysteroid titer is low, implying the existence of novel, as yet uncharacterized, temporal signals in Drosophila. ( S MALL LIPOPHILIC HORMONES, including retinoic acid, thyroid hormone, and steroids, function as key regulators of development and adult physiology in higher organisms. These signals are transduced by ligand-dependent transcription factors that comprise the nuclear receptor superfamily (1). Nuclear receptors are defined by a highly conserved DNA-binding domain that consists of two zinc fingers, as well as a C-terminal ligand-binding domain that serves multiple functions, including hormone binding, dimerization, and ligand-dependent transactivation. The Drosophila genome encodes 21 members of the nuclear receptor superfamily, many of which have close homologs in mammals and Caenorhabditis elegans (2)(3)(4). The ecdysone receptor (EcR) binds the steroid hormone 20hydroxyecdysone (hereafter referred to as ecdysone) and transduces this signal as a heterodimer with the Drosophila retinoid X receptor ortholog, Ultraspiracle (USP) (5)(6)(7)(8)(9). DHR38, which encodes the ortholog of vertebrate NGFI-B/Nurr family members, can also heterodimerize with USP and act as an ecdysteroid receptor, although by a novel mechanism that does not involve direct hormone binding (10,11). USP has been proposed to act as a receptor for the sesquiterpenoid hormone, juvenile hormone (12). The remaining Drosophila nuclear receptors are classified as orphan receptors having, as yet, no known ligands (13).
Ecdysone functions as a critical temporal signal in Drosophila, triggering the major developmental transitions in the life cycle (14). In embryos, ecdysone is required for morphogenetic events as well as cuticle deposition (15). Pulses of ecdysone during the first and second larval instars define the duration of these developmental stages, triggering molting of the cuticle (14). A high titer pulse of ecdysone at the end of the third larval instar triggers puparium formation, signaling metamorphosis, and the onset of prepupal development. This stage is terminated by another ecdysone pulse, approximately 10 h after puparium formation, that triggers eversion of the adult head, marking the prepupal-to-pupal transition. During metamorphosis, ecdysone coordinates the destruction of obsolete larval tissues by programmed cell death and their replacement by adult tissues, directing the formation of the adult fly (16).
Ecdysone exerts its effects on development by triggering cascades of gene expression through the EcR/ USP receptor heterodimer (9,17,18). EcR encodes two protein isoforms, with EcR-B, but not EcR-A, acting as a potent hormone-dependent transcriptional activator in tissue culture assays (19,20). The hormone-receptor complex directly induces a set of primary-response genes, including the Broad-Complex (BR-C) and E74 early puff genes (21,22) as well as a subset of orphan nuclear receptors (23). These genes encode transcription factors that transduce and amplify the ecdysone signal, regulating the expression of large batteries of downstream secondary-response late genes. BR-C encodes a family of zinc finger transcription factors while E74 encodes two ETS-domain proteins, E74A and E74B, from distinct transcripts. BR-C, EcR, and E74B mRNAs are induced by lower concentrations of ecdysone than E74A and thus provide a sensitive indicator of hormone titers, as assayed in late third-instar larval organs (24,25). Higher ecdysone concentrations further up-regulate BR-C transcription while repressing EcR and E74B and inducing E74A.
Some nuclear receptors function during embryogenesis and/or in neuronal development. These include the tailless (tll) and knirps (kni) gap genes that contribute to embryonic pattern formation (41-44). Knirps-related (knrl) functions together with its closely related partner kni in both head and wing development (45-48). FTZ-F1 also encodes an embryonic-specific isoform, ␣FTZ-F1, that acts as a critical cofactor for the FTZ homeotic transcription factor (49-51), whereas DHR3 null mutants die during embryogenesis with defects in peripheral nervous system development (52). Three nuclear receptor genes, dissatisfaction (dsf), seven-up (svp), and eagle (eg), are required for neuronal function. Dsf is expressed in a subset of central nervous system (CNS) neurons and is required for appropriate adult sexual behavior (53). Svp (the fly ortholog of COUP-TF) contributes to a number of developmental pathways, including photoreceptor determination in the adult eye (54-57), whereas eg is required for the development of serotonergic neurons in the CNS (58)(59)(60). Finally, six nuclear receptor genes have not yet been subjected to specific mutational analysis: DHR4, DHR96, dERR (the fly ortholog of vertebrate ERR), dHNF-4 (the fly ortholog of vertebrate HNF-4), DHR83, and CG16801 (the fly ortholog of C. elegans FAX-1).
The genome sequence of D. melanogaster allows, for the first time, a genomic approach toward studying the nuclear receptor superfamily in this organism.
Comparing and contrasting the regulation and function of Drosophila nuclear receptor genes should provide a better understanding of how these factors act together to define specific developmental pathways in the organism. As a first step toward this goal, we have determined the temporal profiles of nuclear receptor gene expression during major ecdysone-triggered developmental transitions in the fly life cycle: embryogenesis, a larval-to-larval molt, puparium formation, and the prepupal-pupal transition. This study reveals a repeated cascade of nuclear receptor transcription, in which a subset of nuclear receptor genes are expressed in a defined sequential pattern in apparent response to each ecdysone pulse examined. A previously uncharacterized orphan nuclear receptor gene, dERR, is expressed in embryos and at the onset of metamorphosis in a pattern that correlates with ecdysone pulses, suggesting critical roles at these stages. Finally, BR-C and E74 mRNAs, which were used as markers for ecdysone pulses, are expressed at times during development when the ecdysone titer is low, suggesting that distinct, as yet unidentified, temporal signals may contribute to progression through the Drosophila life cycle.

Temporal Patterns of Nuclear Receptor Transcription during Embryogenesis
Total RNA was isolated from two independent collections of embryos staged at 2-h intervals throughout the 24 h of Drosophila embryonic development. Five Northern blots were prepared using equal amounts of RNA from each time point. These blots were sequentially hybridized, stripped, and rehybridized with radioactive probes derived from each of the 21 nuclear receptor genes encoded by the Drosophila genome (Table 1). This approach allowed us to generate time courses of nuclear receptor gene expression that could be directly compared between family members. The transcripts detected are consistent with reported sizes ( Fig. 1 (2-4 h) suggests that these mRNAs are maternally loaded and rapidly degraded ( Fig. 1). EcR-B and usp transcripts are induced in early embryos, up-regulated at 6-8 h after egg laying (AEL), and maintain expression through the end of embryogenesis, with down-regulation of EcR-B in late embryos ( Fig. 1). EcR-A, in contrast, is expressed for a relatively brief temporal window, at 8-14 h AEL. These patterns of EcR transcription are very similar to those reported by Talbot et al. (64), although earlier repression of EcR-A was seen in our study.
Six nuclear receptor genes are expressed in brief intervals during midembryonic stages. DHR39 and E75A are initially induced at 4-6 and 6-8 h AEL, respectively, and peak at 8-12 h AEL (Fig. 1). This is followed by induction of DHR3, DHR4, and E75B at 8-12 h AEL, followed by ␤FTZ-F1 expression at 12-18 h AEL. DHR39 appears to exhibit an expression pattern reciprocal to that of ␤FTZ-F1, with lowest levels of mRNA at 14-16 h AEL and reinduction at 16-18 h as ␤FTZ-F1 is repressed. This is followed by a second peak of E75A transcription at 18-22 h AEL.
A second group of nuclear receptors, DHR78, DHR96, dHNF-4, and dERR, is more broadly expressed at low levels throughout embryogenesis. DHR78 accumulates above its constant low level of expression between 8 and 14 h AEL. dERR exhibits an apparent mRNA isoform switch between 14 and 18 h AEL. dHNF-4 regulation also appears complex, with two size classes of mRNA induced at approximately 8-10 h AEL. While the 4.6-kb dHNF-4 mRNA is expressed throughout embryogenesis, the 3.3-kb mRNA is down-regulated at 14-16 h AEL. This timing is consistent with work by Zhong et al. (65), who showed that dHNF-4 is expressed primarily in the embryonic midgut, fat body, and Malpighian tubules. Finally, nuclear receptors known to exert essential functions in patterning the early embryo, tll, kni, and knrl, are expressed predominantly during early stages (Fig. 1).
Two genes that are not members of the nuclear receptor superfamily, BR-C and E74, were also examined in this study, as transcriptional markers for ecdysone pulses during development. Unexpectedly, both of these genes are induced late in embryogenesis, several hours after the rise in ecdysone titer at 6 h AEL (66, 67) (Fig. 1). An approximately 7-kb BR-C transcript is induced at 10-12 h AEL and is present through the end of embryogenesis while E74B is induced at 14-16 h AEL and repressed as E74A is expressed from 16-20 h AEL. This BR-C expression pattern is consistent with the identification of the BR-C Z3 isoform in specific neurons of the embryonic CNS (68, 69).

Temporal Patterns of Nuclear Receptor Transcription during a Larval Molting Cycle
First-instar larvae were synchronized as they molted to the second instar, aged and harvested at 4-h intervals throughout second-instar larval development. Two Northern blots were prepared using equal amounts of total RNA isolated from a single collection of animals. Each blot was sequentially hybridized, stripped, and rehybridized to detect nuclear receptor transcription (Fig. 2, left). The following transcripts were not detectable during the second instar: E75C, dERR, CG16801, DHR83, dsf, eg, svp, tll, kn, and knrl ( Fig. 2 and data not shown).
EcR-B expression is induced in mid-second-instar larvae, but does not reach maximum levels until 68-72 h AEL, just before the molt (Fig. 2). In contrast, usp is expressed throughout the instar. A sequential pattern of nuclear receptor expression is observed that resembles the pattern seen in midembryogenesis. DHR39 and E75A are expressed in the early second instar. This is followed by induction of E75B, E78B, DHR3, and DHR4, followed by expression of ␤FTZ-F1 at the end of the instar. DHR39 again shows a pattern that is approximately reciprocal with ␤FTZ-F1, with highest levels during the first half of the instar. Similarly, DHR78, DHR96, and dHNF-4 exhibit broad expression patterns throughout second-instar larval development. E74A, E75A, and DHR38 are coordinately upregulated with EcR-B at the end of the instar, between 64-72 h AEL. Finally, an approximately 9-kb BR-C transcript is detected throughout the second-larval instar.

Temporal Patterns of Nuclear Receptor Transcription during the Onset of Metamorphosis
We also examined nuclear receptor gene expression throughout the third larval instar and into the early stages of metamorphosis, encompassing the ecdysone-triggered larval-to-prepupal and prepupal-topupal transitions. Third-instar larvae were staged relative to the molt from the second instar and harvested at 4-h intervals throughout the 48 h of the instar. Prepupae were synchronized relative to puparium formation (Ϯ15 min) and harvested at 2-h intervals up to 16 h after puparium formation (APF). Total RNA was isolated from whole animals and analyzed by Northern blot hybridization. Five blots were prepared from two independent collections of animals. These blots were sequentially hybridized, stripped, and rehybridized to detect nuclear receptor gene expression (Fig. 2). The following transcripts were not detectable during thirdinstar larval or prepupal stages: CG16801, DHR83, dsf, eg, svp, tll, kn, and knrl (data not shown).
Most nuclear receptor genes show little or no detectable expression in early and mid-third-instar larvae, a time when the ecdysone titer is low (70). Similar to the pattern seen in second-instar larvae, usp is expressed at relatively low levels throughout the instar and up-regulated at puparium formation, while EcR-B is induced at approximately 100 h AEL and rapidly down-regulated at puparium formation (Fig. 2). This is followed by a sequential pattern of nuclear receptor expression similar to that seen at earlier stages. Equal amounts of total RNA isolated from embryos staged at 2-h intervals were analyzed by Northern blot hybridization to determine the temporal patterns of nuclear receptor gene expression. Time points in hours AEL are shown at the top with a black box representing the approximate timing of the ecdysone titer peak (see Fig. 3). Arrows designate E74A and E74B mRNAs. Detection of rp49 mRNA was used as a control for loading and transfer.

Fig. 2. Temporal Profiles of Nuclear Receptor Gene Expression during Larval and Prepupal Development
Samples of total RNA were isolated at 4-h intervals throughout second-and third-instar larval development and at 2-h intervals throughout prepupal stages. Equal amounts of RNA from each time point were analyzed by Northern blot hybridization to determine the temporal patterns of nuclear receptor gene expression. Time points of larval development in hours AEL, and hours APF for prepupal and early pupal stages, are shown at the top along with the major ecdysone-triggered developmental transitions represented by lines. The approximate times of peak ecdysone titer are marked with boxes (see text for details). Arrows designate E74A and E74B mRNAs. For BR-C transcripts, the arrow marks the 9-kb RNA (68). The time courses were divided into separate blots at the mid-third instar. The white line down the middle denotes the boundary between these two blots. Although the signal intensity for each gene is approximately equal between the blots, they cannot be directly compared. The apparent increased expression of many genes at 92-96 h AEL is not reproducible. Hybridization to detect rp49 mRNA was used as a control for loading and transfer. The 88-to 92-h AEL, 92-to 96-h AEL, and 14-h APF lanes are underloaded. DHR39, E75A, and E78B are induced at 116-120 h AEL, in concert with the late larval ecdysone pulse, followed by maximum accumulation of E75B, DHR3, and DHR4 at 0-4 h APF. ␤FTZ-F1 is expressed from 6-10 h APF, with a pattern that is approximately reciprocal to that of DHR39. EcR-A is expressed in parallel with E75B, DHR3, and DHR4 in midprepupae, similar to their coordinate expression during embryogenesis.
DHR78, DHR96, and dHNF-4 continue to exhibit broad expression profiles throughout third-instar larval and prepupal development (Fig. 2). An E75 isoform not detected in embryos or second-instar larvae, E75C, is also detectable at low levels throughout most of the third instar and up-regulated in correlation with the late-larval and prepupal pulses of ecdysone. DHR38 is detectable at very low levels in early third-instar larvae, in synchrony with the early induction of E74B and BR-C. E74B is repressed, E74A is induced, and BR-C transcripts are up-regulated in late third-instar larvae, in synchrony with the late-larval ecdysone pulse (Fig.  2). The prepupal pulse of ecdysone occurs at 10-12 h APF, marking the prepupal-to-pupal transition. EcR-A, E75A, E78B, DHR4, dERR, E75C, dHNF-4, and E74A are all induced at 10-12 h APF, in apparent response to this hormone pulse. These results are consistent with a microarray analysis of gene expression at the onset of metamorphosis where the temporal profiles of about half of these genes were reported (71).

DISCUSSION
The completion of the Drosophila genome sequence has revealed 21 members of the nuclear receptor superfamily, significantly less than the approximately 270 nuclear receptor genes identified in C. elegans and 49 nuclear receptor genes in humans (3,4,72,73). The definition of this family allows us, for the first time, to undertake a genomic approach toward studying this class of ligand-regulated transcription factors. This study represents a first step in that direction, determining the temporal profiles of expression for all detectable Drosophila nuclear receptor transcripts during each major ecdysone-triggered developmental transition in the life cycle: embryogenesis, a larval molt, puparium formation, and head eversion. Most nuclear receptors can be divided into one of four classes based on this study: 1) those that are expressed exclusively during early embryogenesis (kni, knrl, tll); 2) those that are expressed throughout development (usp, DHR78, DHR96, dHNF-4); 3) those that are expressed in a reproducible temporal cascade at each stage tested (E75A, E75B, DHR3, DHR4, FTZ-F1, DHR39); and 4) those that are undetectable in our assays (CG16801, DHR83, dsf, eg, svp). Below, we discuss these classes and propose new cross-regulatory interactions between orphan nuclear receptors. In addition, we describe unexpected coordinate patterns of gene expression during the life cycle, provid-ing evidence for novel temporal signals that may act independently of the well characterized ecdysone signaling pathway.

Nuclear Receptor Expression Patterns Reveal Distinct Temporal Classes
Three nuclear receptor genes appear to be expressed exclusively during early embryogenesis: kni, knrl, and tll. This restricted pattern of expression fits well with the functional characterization of these genes, which have been shown to act as key determinants of embryonic body pattern (41, 43, 44, 47, 48). We have also identified eight genes (usp, EcR, FTZ-F1, DHR39, DHR78, DHR96, dERR, and dHNF-4) that appear to have maternally deposited transcripts and thus possible embryonic functions. Indeed, maternal functions have been defined for usp, EcR, and ␣FTZ-F1 (15, 49, 51, 61-63).
Four nuclear receptor genes are broadly expressed through all stages examined: usp, DHR78, DHR96, and dHNF-4. This is consistent with an analysis of usp expression by Henrich et al. (74), as well as the profiles of DHR78 and DHR96 expression during the onset of metamorphosis (75). dHNF-4 mRNA is first detectable at 6-10 h AEL (Fig. 1), as the ecdysone titer begins to rise (Fig. 3A). In addition, peaks of dHNF-4 expression are seen at 0, 12, and 16 h APF, in synchrony with the E74 and E75C early ecdysone-inducible genes. These observations raise the interesting possibility that this orphan nuclear receptor is regulated by ecdysone.
DHR38 transcripts are difficult to detect in our assays. This is consistent with earlier studies which used RT-PCR or riboprobes for this purpose (31,75). Nonetheless, we can detect DHR38 mRNA during thirdinstar larval development, consistent with the widespread expression reported in earlier studies. Kozlova et al. (31) have shown that DHR38 expression peaks at late pupal stages, consistent with its essential role in adult cuticle formation.
dERR and E75C display related temporal profiles of expression that do not fit with other nuclear receptor genes described in this study. Both of these genes are specifically transcribed during prepupal development, with increases in expression at 0 and 10-12 h APF. dERR, but not E75C, is also expressed during embryogenesis, with an initial induction at approximately 6 h AEL. These increases occur in synchrony with ecdysone pulses, suggesting that these orphan nuclear receptor genes are hormone inducible, although in a stage-specific manner. Further studies of dERR regulation, as well as a genetic analysis of this locus, are currently in progress (Sullivan, A. A., and C. S. Thummel, unpublished results).
Finally, we did not detect transcripts from five nuclear receptor genes during the stages examined: CG16801, DHR83, dsf, eg, and svp. Interestingly, these genes are implicated in specific neuronal functions during development. Three of these genes, dsf, eg, and svp, are expressed in small subsets of neurons and play roles in neuronal specification (eg and svp) (57)(58)(59)(60)76) or adult behavior (dsf) (53). It is likely that these highly restricted patterns of expression preclude the detection of their corresponding transcripts by Northern blot analysis. CG16801 and DHR83 were identified by virtue of the genome sequence (2). Both of these genes are most similar to C. elegans FAX-1, which is required for neuronal pathfinding and neurotransmitter expression (3,77). This gene is expressed in specific neurons during development, consistent with the possibility that its fly homologs may also be expressed in a highly cell type-specific manner.

A Recurring Cascade of Orphan Nuclear Receptor Gene Expression Follows the Midembryonic, Second-Instar, and Third-Instar Ecdysone Pulses
Interactions between the DHR3 and E75B orphan nuclear receptors contribute to appropriate ␤FTZ-F1 regulation during the onset of metamorphosis. DHR3 is both necessary and sufficient to induce ␤FTZ-F1 and appears to exert this effect directly, through two response elements in the ␤FTZ-F1 promoter (26,27,78,79). E75B can heterodimerize with DHR3 and is sufficient to block the ability of DHR3 to induce ␤FTZ-F1 (27). These three factors thus define a cross-regulatory network that contributes to the timing of ␤FTZ-F1 expression in midprepupae. ␤FTZ-F1, in turn, acts as a competence factor that directs the appropriate genetic and biological responses to the prepupal pulse of ecdysone (28,80,81). The patterns of DHR3, E75B, and ␤FTZ-F1 expression that we observe at the onset of metamorphosis are consistent with these regulatory interactions as well as the expression patterns reported in earlier studies (27,29,71,82).
Unexpectedly, the tight linkage of DHR3, E75B, and ␤FTZ-F1 expression seen at the onset of metamorphosis is recapitulated at earlier stages, after each of the major ecdysone pulses examined, in midembryogenesis and second-instar larval development (Fig. 4). This observation suggests that the regulatory interac- tions between these receptors is not restricted to metamorphosis, but rather may recur in response to each ecdysone pulse during development. It is possible that this regulatory cascade contributes to cuticle deposition, which is dependent on ecdysone signaling in embryos, larvae, and prepupae (14,15). In support of this proposal, DHR3 and ␤FTZ-F1 mutants exhibit defects in larval molting, suggesting that they act together to regulate this early ecdysone response.
Three other orphan nuclear receptor genes, E75A, DHR4, and DHR39, are expressed in concert with DHR3, E75B, and ␤FTZ-F1, after the embryonic, second-instar, and third-instar ecdysone pulses (Fig. 4). A peak of E75A expression marks the start of each genetic cascade, correlating with the rising ecdysone titer in 6-to 8-h embryos, the first half of the second instar, and in late third-instar larvae (Figs. 1, 3, and 4). This is followed by DHR3, E75B, and DHR4 expression which, in turn, is followed by a burst of ␤FTZ-F1 expression (Fig. 4). E78B is expressed in synchrony with DHR4 in late second and third-instar larvae (Fig. 2), but not in embryos. These patterns of expression raise the interesting possibility that E75A, DHR4, and E78B may intersect with the cross-regulatory network defined for DHR3, E75B, and ␤FTZ-F1. E75B and E78B are related to the Rev-erb vertebrate orphan nuclear receptor and are both missing their DNA binding domain (32,40,83). E75B and E78B null mutants are viable and fertile, suggesting that they exert redundant regulatory functions (33,40). E75A mutants die during larval Northern blot data for E75A, DHR3, DHR4, E75B, FTZ-F1, and DHR39 were quantitated by laser scanning densitometry and converted into histograms. The plots are aligned using different gray tones for each time point, with the times listed at the bottom in hours AEL for embryonic (embryo), second-instar (L2), and third-instar (L3) stages, and hours APF for prepupal stages (prepupa). stages, with no known direct regulatory targets (33). DHR4 mutants have not yet been described, although recent work indicates that this gene exerts essential roles in genetic and biological responses to the late larval ecdysone pulse (King-Jones, K., J.-P. Charles, and C. S. Thummel, unpublished results). Further functional studies of these nuclear receptor genes should provide insight into their possible contribution to the regulatory circuit defined by DHR3, E75B, and ␤FTZ-F1.
Interestingly, DHR39 displays a reproducible pattern of expression that is inversely related to that of ␤FTZ-F1, defining possible repressive interactions (Figs. 1,  2, and 4). DHR39 and ␤FTZ-F1 have a similar DNA binding domain (63% identity) and bind to identical response elements, suggesting that they may exert cross-regulatory interactions (36,37,84). Moreover, DHR39 can repress transcription through the same response element that is activated by ␤FTZ-F1 (36). It would be interesting to determine whether the reciprocal patterns of DHR39 and ␤FTZ-F1 expression during development is of functional significance.
Recent work has defined a cross-regulatory cascade of vertebrate nuclear receptors that regulate bile acid synthesis, indicating that these interactions are not unique to insects (85,86). Binding of bile acids to the FXR nuclear receptor directly induces expression of an atypical nuclear receptor that lacks a DNA binding domain, SHP. This factor, in turn, heterodimerizes with an orphan nuclear receptor, LHR-1, inhibiting its transactivation function and thereby down-regulating both LHR-1 and the rate-limiting enzyme in the bile acid synthetic pathway, cholesterol 7␣-hydroxylase. Regulatory cascades of nuclear receptor expression may thus provide a more general means of obtaining specificity and feedback regulation in hormone-controlled biological pathways.

Evidence for New Temporal Signals during the Drosophila Life Cycle
The transcription of BR-C, EcR, E74, and E75 has been extensively characterized during the onset of metamorphosis, due to their rapid and direct regulation by the steroid hormone ecdysone at this stage in development (24,25,32,64,68,82). Surprisingly, however, their expression appears to be disconnected from the high-titer ecdysteroid pulses during embryonic and second-instar larval stages. As expected, EcR is induced early in embryonic development, in coincidence with the rising ecdysone titer at 4-10 h AEL, with EcR-B transcripts appearing first followed by EcR-A (Figs. 1 and 3A). BR-C mRNA, however, is not seen until 10-12 h AEL and E74B mRNA is induced even later, at 14-16 h AEL, when the ecdysteroid titer has returned to a basal level (Fig. 3A). Both EcR-B and E74B are repressed from 16-20 h AEL as E74A and E75A are induced, a switch that has been linked to the high-titer ecdysone pulse in late third-instar larvae (24,25); however, this response occurs during late embry-ogenesis when the ecdysteroid titer is low (Figs. 1 and  3A). A similar observation has been made for E75A expression in the Manduca dorsal abdominal epidermis, where a brief burst of E75A mRNA is detected immediately before pupal ecdysis, after the ecdysteroid titer has returned to basal levels (87).
Ecdysteroid titer measurements for the second larval instar in Drosophila (Fig. 3B) are in more dispute than those determined by Maroy et al. (66) and Kraminsky et al. (67) for embryonic development (Fig.  3A). Both Kraminsky et al. and Bialecki et al. (33) report a peak of ecdysteroids between 4 and 12 h after the first-to-second instar larval molt (Fig. 3B). The data from Maroy et al. (88) show a much later ecdysteroid peak, at the end of the instar (Fig. 3B). This conclusion is based, however, on only five data points through the second instar. Moreover, Kraminsky et al. and Maroy et al. (88) synchronized their collections at hatching, rather than the molts, and did not report on the synchrony of their animals as they progressed through the instars. In contrast, Bialecki et al. (33) staged their animals at the first-to-second instar larval molt and followed their progression into the third instar. It thus seems likely that the second instar ecdysone pulse occurs during the first half of the instar. This profile is consistent with the early induction of E75A (Fig. 2). EcR-B and E74A, however, are not induced until the second half of the second instar, with a peak at the end of the instar. BR-C mRNA levels remain steady throughout the second instar. Finally, as reported earlier, EcR-B, E74B, and BR-C are induced in early to mid-third-instar larvae, a time when one or more lowtiter ecdysone pulses may occur (64,82,89). It is curious that E74B is poorly expressed relative to E74A during embryonic and second-instar larval stages, disconnecting its expression from that of EcR. This pattern is not seen in studies that focused on the onset of metamorphosis (25,82,89). Taken together, the temporal profiles of early gene expression (EcR, BR-C, E74, E75A) during late embryonic and late secondinstar larval stages appear to be unlinked to the known ecdysteroid pulses at these stages. This could indicate that these promoters are activated in a hormoneindependent manner at these stages in the life cycle. Alternatively, these ecdysone primary-response genes may be induced by a novel temporal signal that remains to be identified.
Several lines of evidence indicate that 20-hydroxyecdysone is not the only temporal signal in Drosophila. A major metabolite of this hormone, 3-dehydro-20hydroxyecdysone, was shown to be as effective as 20-hydroxyecdysone in inducing target gene transcription in the hornworm, Manduca sexta (90). Similarly, 3-dehydro-20-hydroxyecdysone is more efficacious than 20-hydroxyecdysone in inducing Fbp-1 transcription in the Drosophila larval fat body (91). Champlin and Truman (92) have shown that a hightiter pulse of ␣-ecdysone, the precursor to 20hydroxyecdysone, can drive the extensive proliferation of neuroblasts during early pupal development in Manduca. This is the first evidence that ␣-ecdysone is responsible for a specific response in insects. It is unlikely, however, that this signal is transduced through the EcR/USP heterodimer, which shows only very low transcriptional activity in response to this ligand (93). Rather, recent evidence indicates that ␣-ecdysone may activate DHR38 through a novel mechanism that does not involve direct hormone binding (11).
Studies of ecdysteroid-regulated gene expression in Drosophila have also provided evidence for hormone signaling pathways that may act independently of 20hydroxyecdysone. Several studies have identified a large-scale switch in gene expression midway through the third larval instar, an event that has been referred to as the mid-third-instar transition (94). It is not clear whether this response is triggered by a low-titer ecdysteroid pulse, another hormonal signal, or in a hormone-independent manner (89). Similarly, the let-7 and miR-125 micro-RNAs are induced at the onset of metamorphosis in Drosophila in tight temporal correlation with the E74A early mRNA, but not in apparent response to 20-hydroxyecdysone (95). These studies, taken together with the data presented in this paper, indicate that 20-hydroxyecdysone cannot act as the sole temporal regulator during the Drosophila life cycle.

Developmental Staging
Canton S wild-type flies were maintained at 25 C. Embryos were collected on molasses/agar plates supplemented with yeast paste. After 2 d of conditioning and discarding an initial egg lay, embryos were collected at 2-h intervals and harvested either immediately or after aging for 2-22 h at 25 C. To harvest, embryos were washed with water over a nylon mesh, dechorionated in 50% bleach, rinsed, and frozen at Ϫ80 C until all 12 time points were collected. Larvae were staged by placing either first or second instar larvae in petri dishes on moist black filter paper supplemented with yeast paste and maintaining them at 25 C. The plates were examined 4 h later for larvae that had molted. Newly molted larvae were collected immediately or aged in 50-ml glass beakers with fresh yeast paste in a humidified chamber. They were subsequently harvested at 4-h intervals and frozen at Ϫ80 C. Prepupae were staged relative to puparium formation and aged at 2-h intervals for up to 16 h APF, and then frozen at Ϫ80 C. Embryos and third-instar larvae advanced to the next developmental stage at the appropriate time, indicating that the staging was accurate.

Northern Blot Hybridizations
Total RNA (10 or 15 g) was fractionated on formaldehyde 1% agarose gels for embryonic or larval/prepupal samples, respectively. RNA was transferred to nylon membranes and cross-linked by UV irradiation as described (24). Blot hybridization, washes, and stripping were performed as described (89), except blots were stripped by boiling for 5 min instead of 30 min. Many hybridizations were confirmed by probing an independent blot, including all nuclear receptor genes that gave no detectable signal. The expression patterns of some nuclear receptor genes (see Fig. 4) were quantitated by laser scanning densitometry using ImageQuant software (Amersham Pharmacia Biotech, Arlington Heights, IL). Values were converted to percentages of the highest value for each autoradiograph and plotted into histograms. Thus, although each histogram represents the relative levels of expression for one transcript at one stage, the relative levels of expression between different histograms are not meaningful.

Probe Preparation
DNA templates used to prepare labeled probes are listed in Table 1. Most probes were prepared by gel purifying a restriction fragment and labeling it with random oligonucleotide primers (Prime-It, Stratagene, La Jolla, CA). dERR, CG16801, DHR4, dsf, eg, kni, and knrl probe templates were prepared by PCR amplification from genomic DNA isolated from adult flies, except dERR, which was amplified from a cDNA clone. DHR4 and DHR83 probes were a gift from K. King-Jones. The DHR38 probe was a gift from T. Kozlova. The PCR primers used to generate a probe for dERR were ATAGATCTGC-CACTTAACGAC and TCAGCTGGAGCGTCAGGATCT (a gift from T. Kozlova). The primers used for CG16801 were GAT-GTAACCAACGACAACGAGGAGCCGC and AGAGCACCTT-TTCCATGGGCGTGTTCCC. The primers used for dsf were CCGTCTCGCTGGTGACCAATGTCTCGG and CGCAAATCA-TGAGAGAGGCGTGGCTGGG. The primers used for eg were AGTCTGGAGCAGCTGGCGCGATTACAGCandCCTGTCAA-CGTTTGTGACCTCACCAGGG. The primers used for kni were TACCGGCAGGAGATGTACAAGCACCGCC and CGA-ATATTCCCCTCATGGCACTAGCCGC. The primers used for knrl were TGACGCCCGGCAGACCGCCACAGATGCG and TTCAATGCTGGTCTTTGCTGTCTCTGTGG.