Neverland is an evolutionally conserved Rieske-domain protein that is essential for ecdysone synthesis and insect growth

Steroid hormones are responsible for the coordination and regulation of many aspects of development, growth and differentiation of multicellular organisms. In insects and other arthropods, a strict regulation of titers of ecdysteroids, especially ecdysone (so called α-ecdysone) and 20-hydroxyecdysone (20E), plays central roles in development, especially in guiding transition from one developmental stage to the next via molting and metamorphosis (Thummel, 2001; Gilbert et al., 2002; Thummel and Chory, 2002). During larval and pupal development of insects, ecdysone is synthesized from dietary cholesterol or phytosterols via a series of hydroxylation and oxidation steps in the small endocrine organ called the prothoracic gland (PG)(Gilbert et al., 2002).

Studies using Drosophila have identified several molecules that are involved in ecdysteroid biosynthesis in the PG. For example, ecdysone synthesis is regulated by itpr, which encodes inositol 1,4,5-trisphosphate receptor (Venkatesh and Hasan, 1997; Venkatesh et al., 2001); dare, which encodes adrenodoxin reductase(Freeman et al., 1999); ecdysoneless (ecd), which encodes an evolutionally conserved protein with no known motifs (Warren et al., 1996; Gaziova et al.,2004); and without children, which encodes a putative transcriptional regulator (Wismar et al.,2000; Warren et al.,2001). Ras-dependent signaling cascade and insulin-dependent PG cell growth are also essential for the ecdysone production and/or release(Caldwell et al., 2005; Mirth et al., 2005).

Recently, five hydroxylase genes that are essential for ecdysteroid biosynthesis have been identified in Drosophila. All of the hydroxylase genes, Cyp306a1/phantom (phm), Cyp302a1/disembodied (dib), Cyp315a1/shadow(sad), Cyp314a1/shade (shd) and Cyp307a1/spook (spo), are named the Halloween genes and encode cytochrome P450 mono-oxygenases(Chávez et al., 2000; Warren et al., 2002; Petryk et al., 2003; Niwa et al., 2004; Warren et al., 2004; Namiki et al., 2005). A combination of molecular and biochemical experiments have shown that Phm, Dib,Sad and Shd play pivotal roles in the final four steps of ecdysteroidogenesis,namely the conversion of 5 β-ketodiol to 20E(Gilbert and Warren, 2005). Orthologs of the Halloween P450 genes have also been identified in the silkworm Bombyx mori and the tobacco hornmoth Manduca sexta.The expression patterns of these lepidopteran hydroxylase genes are spatially restricted to the PG and are temporally correlated with the ecdysteroids titer during larval development (Niwa et al.,2004; Warren et al.,2004; Namiki et al.,2005; Niwa et al.,2005; Rewitz et al.,2006). The identification of these genes provides the basis for investigating the regulation of insect hormone production in more detail. For example, it has been shown that the expression of phm and dib is regulated by the βFTZ-F1 transcription factor in Drosophila (Parvy et al.,2005). Similarly, molting defective, which encodes a putative transcription factor in Drosophila, influences the expression level of some Halloween P450 genes(Neubueser et al., 2005). In Bombyx, the expression of the dib ortholog is significantly induced by steroidogenic neuropeptide prothoracicotropic hormone in cultured PGs (Niwa et al., 2005).

Although the enzymes involved in the final biochemical steps of ecdysteroid biosynthesis are relatively well characterized, little is known about the molecules involved in earlier steps(Gilbert and Warren, 2005). Dietary cholesterol (C) is first converted to 7-dehydrocholesterol (7dC) by 7,8-dehydrogenation in the endoplasmic reticulum. Conversion of the 7dC to theΔ ⁴-diketol constitutes the so called `black box'. Subsequently, cytosolic 5 β-reduction and microsomal 3 β-reduction steps convert the Δ<sup>4</sup>-diketol to the 5 β-ketodiol. Further identification and characterization of these `early' ecdysteroidogenic genes are important for understanding the mechanisms by which ecdysteroidogenesis and developmental timing are precisely controlled in arthropods.

To facilitate identification and characterization of components responsible for ecdysteroid biosynthesis, we have used Bombyx to identify genes predominantly expressed in the PG (Niwa et al., 2004; Namiki et al.,2005; Niwa et al.,2005; Yamanaka et al.,2005). Based on gene expression analysis using BombyxcDNA microarrays that we have previously performed(Niwa et al., 2004), we now describe a novel gene named neverland (nvd), the expression of which is specifically enriched in ecdysone-synthesizing tissues, including the PG. We show that loss of nvd function in the PG causes growth arrest at the larval stages, and this phenotype is rescued by application of 20E or 7dC. Our results suggest that Nvd plays a pivotal role in the metabolism of cholesterol and steroid intermediates during ecdysteroidogenesis. Considering that the nvd gene family is evolutionally conserved, we propose that the nvd family of proteins is an essential regulator of steroid biosynthesis in various animal phyla.

Culture and staging of silkworms, B. mori (KINSHU χ SHOWA F1 hybrid), have been described previously(Niwa et al., 2004). All D. melanogaster flies were reared on a standard agar-cornmeal medium at 25°C under a 12-hour light/12-hour dark photoperiod unless otherwise specified. 2-286-GAL4 (Timmons et al., 1997; Andrews et al.,2002), AUG21-GAL4(Siegmund and Korge, 2001) and AKH-GAL4 (Lee and Park,2004) were kindly provided from C. S. Thummel, G. Korge and J. H. Park, respectively. woc^rgl(Wismar et al., 2000) was generous gift from J. T. Warren and L. I. Gilbert. breathless-GAL4 (Shiga et al., 1996), elav^c155-GAL4(Luo et al., 1994) and teashirt-GAL4 (Shiga et al., 1996) were provided from Genetic Strains Research Center in National Institute of Genetics, Japan. scabrous-GAL4(Klaes et al., 1994) and decapentaplegic-GAL4(Staehling-Hampton et al.,1994) were obtained from the Bloomington stock center. Lsp2-GAL4 (Cherbas et al.,2003) and pGawB⁵⁰¹⁵ were provided from the Drosophila Genetic Resource Center, Kyoto Institute of Technology. Although a previous study shows that a GAL4 in pGawB⁵⁰¹⁵ is active in lymph gland and hematopoietic cells(Sinenko et al., 2004), the GAL4 was also expressed in the larval ring gland containing the PG cells (see Fig. S1 in the supplementary materials; J.-B. Peyre and T. Aigaki,personal communication). All flies were maintained in yw background. Cyo[y⁺] and TM3[y⁺] were used as balancers.

Because an EST clone prgv0382 from the Bombyx EST project(Mita et al., 2003) lacked the 5 ′ region of full-length cDNA of Bombyx neverland(nvd-Bm), the 5 ′ end of nvd-Bm cDNA was obtained by the 5 ′ Rapid Amplification of cDNA ends (5 ′ RACE) method using the GeneRacer Kit (Invitrogen). The first 5 ′ RACE product was amplified with a gene-specific primer (5 ′-GGGCAGAAGTAAGGAGCGCCATCTCTGTG-3′)and GeneRacer 5 ′ Primer. Then we performed the nested PCR with another gene-specific primer (5 ′-CCGCTGTAAAGAAGCCAATTAAGGTGGCGC-3′) and GeneRacer 5 ′ nested primer. The nucleotide sequence of Drosophila neverland (nvd-Dm) was identified from the DrosophilaEST database (GenBank Accession Number BT021261). The cDNA containing the entire open reading frame (ORF) for nvd-Dm was amplified by PCR from wild-type Drosophila ring gland-derived cDNA using the following primers: forward, 5 ′-ATGACGAGCTACAGTTTATTTTGGATGTC-3′; reverse, 5′-CTACCAACCAATATTGGTTGCTTCAG-3 ′. The DNA sequence data of nvd-Bm and nvd-Dm were deposited in GenBank (AB232986 and AB232987, respectively).

For analyzing tissue expression pattern of nvd-Bm, total RNA was derived from tissue of second-day wandering (W1) fifth instar larvae of Bombyx as described (Niwa et al.,2004). Specific primers for nvd-Bm (forward, 5′-AGATGGCGCTCCTTACTTCTG-3′; reverse, 5′-TCAGACACTTGGTCACTCCATC-3′) were used. The expression level of nvd-Dm was examined by RT-PCR using the following specific primers:forward, 5 ′-CGAGCTACAGTTTATTTTGGATGTCATTGC-3′; reverse, 5′-GGGCATATAACACAGTCGTCAGC-3 ′. Quantitative RT-PCR was performed as described (Niwa et al.,2005). rpL3 and rp49 were used as loading controls for nvd-Bm and nvd-Dm, respectively(Foley et al., 1993; Matsuoka and Fujiwara,2000).

Digoxigenin (DIG)-labeled RNA probes were synthesized using the DIG RNA labeling kit (Roche) and T3 or T7 RNA polymerase (Invitrogen). For the nvd-Bm and nvd-Dm probes, prgv0382 and a pBluescript containing nvd-Dm (nvd-Dm-pBluescript), respectively, were used as templates. Northern blot and hybridization using a DIG-labeled probe was performed as described (Charles et al.,1999). In situ hybridization was performed as described(Lehmann and Tautz, 1994; Buszczak et al., 1999; Niwa et al., 2004).

Overexpression studies and RNAi experiments using hairpin double-stranded RNAs was performed using GAL4/UAS system(Brand and Perrimon, 1993). The construct for nvd-Dm overexpression was generated by ligation of a BamHI/XhoI fragment isolated from nvd-Dm-pBluescript into the BglII/XhoI site of pUAST vector (Brand and Perrimon,1993). To carry out a transgenic RNAi(Kennerdell and Carthew,2000), we generated two distinct UAS-nvd-Dm-Inverted Repeat constructs, designated UAS-nvd-Dm-IR-1 and UAS-nvd-Dm-IR-2. UAS-nvd-Dm-IR-1 was a genomic-cDNA fusion construct(Kalidas and Smith, 2002)comprising the 488-1131 bp region of the nvd-Dm gene. This genomic fragment contained a part of the fourth exon downstream of an intrinsic BspHI site, the entire fourth intron and a NotI site at the 3 ′ end (see Fig. S2 in the supplementary materials). These fragments amplified by PCR were ligated into pUAST. To make UAS-nvd-Dm-IR-2 (a cDNA-cDNA RNAi construct), the 77-757 bp region was amplified and cloned into pUAST as a tail-to-tail inverted repeat (see Fig. S2 in the supplementary material). The Drosophila transformants were obtained using standard protocols.

To assess the phenotype of nvd-Dm RNAi individuals, yw;UAS-nvd-Dm-IR males of each RNAi strain were crossed with yw;2-286-GAL4/TM3[y+] females and 0-12 hours after egg laying (AEL)eggs were collected. The animals were reared on agar-apple juice plates with yeast paste. nvd-Dm RNAi progeny carrying both UAS and GAL4 transgenes (yw; UAS-nvd-Dm-IR/+; 2-286-GAL4/+) were distinguished from the others not carrying either UAS or GAL4 by the yellowish color of their mouth hooks and denticle belts,because all of UAS and GAL4 strains were balanced by balancers with y⁺ marker. Progeny carrying either UAS or GAL4 transgene alone were used as a control. Control and nvd-Dm RNAi larvae were transferred to fresh plates 36±6 hours AEL. After 48 hours AEL, we checked the lethality and phenotype of larvae every 24 hours. Larvae were staged by the morphology of the mouth hook as described (Roberts and Standen,1998).

Control and nvd-Dm RNAi larvae were collected 30-42 hours AEL and stored in 1.5 ml tubes that were weighed before and after the addition of larvae in order to determine the weight of the larvae. Sample preparation for ecdysteroid titer measurement was done as described(Bialecki et al., 2002). The amounts of ecdysone were measured using the ecdysteroid radioimmunoassay (RIA)as described (Mizoguchi et al.,2001). As 20-hydroxyecdysone (20E, Sigma) was used as the standard, the ecdysteroid amount was expressed in 20E equivalents.

Control and nvd RNAi larvae were collected 30±6 hours AEL and placed on agar-apple juice plates. These larvae were fed on yeast paste with a final concentration of 1 mg/ml or zero 20E in 3.3% ethanol 30-42, 54-66 and 72 hours AEL in order to stimulate the hormone pulse that triggers the molt (Bialecki et al., 2002; Warren et al., 2006). Lethality was scored at 12-hour intervals. Cholesterol (C) and 7-dehydrocholesterol (7dC) were generous gifts from Y. Fujimoto. Control and nvd-Dm RNAi larvae were also fed on yeast paste containing 0.5% wet weight of C and 7dC (50 mg dry yeast, 95 μl water, 5 μl of 100% ethanol and 0.75 mg of each ecdysteroid intermediate)(Warren et al., 2001) from 30 hours AEL until all animals had emerged or died. The experiment using 7dC was carried out under constant dark conditions because 7dC is unstable in light.

To identify components for ecdysteroidogenesis, we have used a Bombyx cDNA microarray (Niwa et al., 2004). We have compared gene expression profiles of the PGs from first day of fifth instar (V0) larvae, in which ecdysone is not produced,with that of second-day of wandering (W1) larvae, in which ecdysone is actively synthesized. The microarray data have yielded 1883 out of 5760 nonredundant cDNA clones showing at least a twofold increased expression(Niwa et al., 2004). Of these clones, prgv0382 (hereafter referred to as nvd-Bm) showed a 4.7-fold higher expression in the PGs of second-day wandering (W1) larvae than those of first-day fifth instar (V0) larvae. Consistent with the microarray data,Northern blot analysis revealed that a change in mRNA expression level of nvd-Bm was positively correlated with the changes of hemolymph ecdysteroid titers during Bombyx development(Fig. 1A)(Satake et al., 1998; Mizoguchi et al., 2001). Furthermore, expression of nvd-Bm in various tissues of the W1 Bombyx larvae was examined by RT-PCR. nvd-Bm was expressed predominantly in the PG, while weak nvd-Bm was also detected in brain and malpighian tubules (Fig. 1B). These results suggest that nvd-Bm is involved in the temporal regulation of ecdysteroid biosynthesis in the PG.

The full-length nucleotide sequence of nvd-Bm was obtained by modified 5 ′ RACE. The predicted ORF for the full-length clone of nvd-Bm encodes a protein of 453 amino acids in length. A BLAST search using the deduced Nvd-Bm protein sequence revealed that putative nvd-Bm orthologs are present in several animal species(Fig. 2A). These Nvd proteins all contain a Rieske [2Fe-2S] center binding motif(C-X-H-X_16-17-C-X₂-H) that is known to function as an electron acceptor and is involved in electron transfer to other proteins(Link, 1999). The putative nvd family of proteins also have a highly conserved domain in the C-terminal region that contains a mononuclear, non-heme iron-binding motif(E/D-X₃-D-X₂-H-X₄-H), which is thought to be involved in oxygen binding (Mason and Cammack, 1992). The primary structure of the Nvd proteins, which has both the Rieske domain and the non-heme iron-binding motif, is typically conserved in class IA terminal oxygenases from prokaryotes(Jiang et al., 1996), such as KshA (3-ketosteroid 9 α-hydroxylase) of Rhodococcus erythropolis (van der Geize et al.,2002) and PrnD (aminopyrrolnitrin dioxygenases) of Pseudomonas fluorescens (Kirner et al.,1998). In the case of Nvd proteins from Bombyx, Drosophila, C. elegans and zebrafish, putative transmembrane regions are predicted in their N-terminal regions (Fig. 2A). Only one homolog existed in the genome of each of these eukaryotic species. We could not find nvd orthologs from mammalian or plant genomes using a standard BLAST search.

The Drosophila homolog of neverland (nvd-Dm) was not originally annotated in the Drosophila euchromatin genome database. Our BLAST search revealed that nvd-Dm is composed of six exons distributed over 76 kb and located on the third chromosome heterochromatin scaffold (Hoskins et al.,2002). The 5 ′ region of nvd-Dm overlapped with a predicted gene designated CG40050 that might have been mis-annnotated in FlyBase (see Fig. S3 in the supplementary materials).

To analyze the expression pattern of nvd-Dm during Drosophila development, in situ RNA hybridizations were carried out on embryos, larvae and adult ovaries. We detected no nvd-Dm mRNA in unfertilized eggs, suggesting that there is no maternal contribution of nvd-Dm (Fig. 3A), as was previously noted for phm, dib and sad(Chávez et al., 2000; Warren et al., 2002; Niwa et al., 2004; Warren et al., 2004). Moreover, no nvd-Dm expression was detected at the blastoderm,gastrulation, germ band elongation and retraction stages(Fig. 3A,B; see Fig. S4 in the supplementary materials). nvd-Dm expression was first seen at stage 14 in the primordia of the ring gland, which contains Drosophila PG cells (see Fig. S4 in the supplementary material). nvd-Dm expression in the ring gland became higher at stage 15, and continued through the remainder of embryogenesis (Fig. 3C). nvd-Dm mRNA was also expressed specifically in the ring gland at the larval stage (Fig. 3D,E). In the ring gland, this expression was exclusively observed in the PG, but not in the corpus allatum or corpus cardiacum cells(Fig. 3D,D′). nvd-Dm expression in the PG was downregulated after ecdysis from the second instar to the third instar, and then was significantly upregulated at the late third instar larval stage (Fig. 3F). These results suggest that the transcriptional activity of nvd correlates with ecdysone titer changes during the larval molting cycle, similar to that of phm, dib, sad and Start1(Drosophila homolog of StAR)(Roth et al., 2004; Gilbert and Warren, 2005).

In adult females, nvd-Dm mRNA was expressed in the nurse cells of developing egg chambers (Fig. 3G-I). The nurse cells are considered to be the source of ecdysteroids in adult females (Riddiford,1993; Gilbert et al.,2002) and a number of ecdysteroidogenic genes, such as dare,Start1 and ecdysoneless, are known to be expressed in the nurse cells (Freeman et al., 1999; Gaziova et al., 2004; Roth et al., 2004). These results indicate that the expression of nvd-Dm is enriched in tissues that synthesize steroid hormones.

To assess the importance of nvd-Dm during development, we examined phenotypes of loss- or gain-of-nvd-Dm function in developing flies. As nvd-Dm is located in heterochromatin region (see Fig. S3 in the supplementary materials), it is difficult to isolate or create genetic mutations within the nvd-Dm locus by genetic mutant screens or homologous recombination techniques. We therefore examined the effects of overexpression or knock down of nvd-Dm in developing flies using the GAL4/UAS system (Brand and Perrimon,1993). In a wild-type background, overexpression of nvd-Dm using any of GAL4 drivers in Table 1 had no visible effect on development (data not shown). To knock down nvd-Dm in developing flies, we used transgenic RNA interference (RNAi), known to be an effective method of degrading endogenous target mRNAs in Drosophila(Kennerdell and Carthew, 2000; Kalidas and Smith, 2002). We established transgenic lines in which double-stranded RNA molecules corresponding to nvd-Dm mRNA were generated using an inverted repeat construct under the control of the UAS promoter(UAS-nvd-Dm-IR; see Fig. S2 in the supplementary materials). We found that all of the RNAi animals failed to develop into adults when the UAS lines were crossed with two GAL4 lines, 2-286-GAL4 and pGawB⁵⁰¹⁵, in which GAL4transgenes are active in the PG cells(Table 1). By contrast, the progeny resulting from crosses between UAS-nvd-Dm-IR flies and GAL4 lines, in which the GAL4 transgenes are not active in the PG, were fully viable (Table 1), consistent with the expression pattern of nvd-Dm in wild-type animals. This lethal phenotype by PG-GAL4 strains was observed using two distinct nvd-Dm RNAi constructs(UAS-nvd-Dm-IR-1 and UAS-nvd-Dm-IR-2), which target different regions of nvd-Dm mRNA(Table 1; see Fig. S2 in the supplementary material). The reduction of nvd-Dm mRNA level in the RNAi larvae was confirmed by RT-PCR (Fig. 4A). These results suggest that nvd-Dm plays an essential role in the PG during fly development.

The lethal phase for animals carrying UAS-nvd-Dm-IR-1 and 2-286-GAL4 was examined in more detail. For simplicity, we refer to these animals as `nvd-Dm RNAi animals'. nvd-Dm RNAi animals completed embryogenesis and hatched normally. Following hatching, the first instar larvae of nvd-Dm RNAi animals showed no apparent morphological or behavioral defects (Fig. 4C-E; data not shown). However, ∼48 hours after egg laying(AEL), nvd-Dm RNAi animals showed apparent growth arrest in body size compared with control animals (Fig. 4E-G). To examine whether larval molting occurs in nvd-DmRNAi animals, the larval stages were determined by stage-specific size and morphology of mouth hook (Roberts and Standen, 1998). By 48 hours AEL, both the control and RNAi animals had the small, one-toothed mouth hook that is characteristic of the first instar larvae (Fig. 4H,I,M,N). Forty-eight hours AEL, the control animals underwent the first molt and showed the second instar larva-type mouth hook with two to five teeth(Fig. 4O,P). After 72 hours AEL, the control animals molted to the third instar larvae and possessed larger mouth hooks with numerous small teeth(Fig. 4Q). By contrast, nvd-Dm RNAi animals still retained the mouth hook of the first instar larva-type, even 96 hours AEL (Fig. 4J-L). At around 72 hours AEL, most of the nvd-Dm RNAi animals gradually ceased feeding and became stationary. Ninety-eight percent(54/55) of these animals died before 108 hours AEL as first instar larva, as judged from the mouth-hook morphology (Fig. 4B, Fig. 6I). We noted that all of the nvd-Dm RNAi animals failed to show the double-mouth-hook phenotype that frequently occurs in mutants with defects in ecdysone signaling (Li and Bender,2000; Gaziova et al.,2004), implying that loss of nvd-Dm function prevents larval growth prior to the initiation of the molting process. These results suggest that nvd-Dm plays a pivotal role in larval development in Drosophila. On this basis of the prolonged first instar larval phenotype, we named this gene `neverland', which is the fictional island featured in J. M. Barrie's play Peter Pan, where children cease to age.

The PG-specific expression of nvd-Dm suggests that nvd-Dmcould function in ecdysteroidogenesis in the PG and that nvd-Dm RNAi animals might have a reduced ecdysteroid titer during development. As a direct test of this hypothesis, we measured the ecdysteroid titer in these animals using a radioimmunoassay. nvd-Dm RNAi and control animals were collected 36±6 hours AEL, at which time wild-type first instar larvae show a peak of 20-hydroxyecdysone (20E)(Kraminsky et al., 1980; Sullivan and Thummel, 2003; Warren et al., 2006). Organic extracts prepared from nvd-Dm RNAi animals had a significantly reduced level of ecdysteroid compared with extracts from control animals(Fig. 5), indicating that nvd-Dm activity is required for ecdysteroid biosynthesis in PG.

To address whether the larval arrest of nvd-Dm RNAi animals is due to reduced ecdysteroid titers, we attempted to rescue the larval arrest phenotype by feeding ecdysteroids to nvd-Dm RNAi animals. When newly-hatched nvd-Dm RNAi animals were fed yeast paste supplemented with active ecdyseroid (20E) 30-42 hours AEL in the first instar period, the arrest phenotype was rescued and the animals grew to the second instar larvae. However, the animals later died as prolonged second instar larvae(Fig. 6A,B,I). Similarly, when the rescued second instar larvae were fed the 20E-containing food during 54-66 hours AEL in the second instar period, they molted into the third instar larvae but failed to become pupae (Fig. 6C,D,I). Furthermore, when the rescued third instar animals received food with 20E after 78 hours AEL, these RNAi animals were partially rescued and developed into pupae. A number of the pupae completed metamorphosis and emerged as fertile adults(Fig. 6E,I), although almost all died within one week of eclosion. Like the nvd-Dm animals without 20E, second and third instar larvae fed with 20E for 12 hours during each instar also showed the prolonged larval phenotype, surviving for 3-4 days after each molting. In spite of the prolonged phenotype, both the rescued second and third instar larvae did not overgrow, and their body sizes were almost the same as mature second and third instar wild-type larvae,respectively (Fig. 6A,C). We confirmed that PG cells looked normal in the prolonged nvd-Dm RNAi larvae (see Fig. S5 in the supplementary materials), indicating that the nvd-Dm RNAi phenotype is not due to loss of PG cells. Taken together,these results suggest that nvd-Dm is essential for insect molting,metamorphosis and body growth throughout development via the regulation of ecdysteroid biosynthesis in the PG.

To examine which ecdysteroid conversion step is affected by loss-of-function of nvd-Dm, we performed a feeding experiment with precursors of ecdysone biosynthesis. If Neverland functions in a specific conversion step of the ecdysteroid biosynthesis pathway, we would expect that an exogenously applied intermediate, which is downstream of the conversion step by Nvd, would overcome the developmental arrest phenotype observed in the nvd-Dm RNAi animals. We performed the feeding experiment with two precursors of ecdysteroid biosynthesis, cholesterol (C) and 7-dehydrocholesterol (7dC) (Gilbert et al., 2002). C and 7dC are involved in the first step of ecdysteroidogenesis in the PG; 7dC is synthesized from C by 7,8-dehydrogenation by an uncharacterized enzyme(Grieneisen et al., 1993; Gilbert et al., 2002; Gilbert and Warren, 2005). When 7dC was added to the food, all nvd-Dm RNAi animals were completely rescued and developed into adults(Fig. 6F,I). The food with 7dC also restored the ecdysone titer in nvd-Dm RNAi animals(Fig. 5), suggesting that the nvd-Dm mutants do not have a defect in the conversion steps from 7dC to ecdysone in its PG. However, nvd-Dm RNAi animals raised on food containing excessive C were partially rescued; about 47% (18/38) of these animals molted to the second instar larvae(Fig. 6G,H,I), but no nvd-Dm RNAi animal was observed to molt to the third instar larvae. It should be noted that although the nvd-Dm RNAi larvae rescued by C arrested in the second instar larvae, they survived for about 3 weeks and their body size reached almost the same size as wild-type third instar larvae(Fig. 6G). These results strongly suggest that the Nvd proteins act upstream of 7dC production in the ecdysteroid synthesis pathway in the PG.

In this study, we identified the neverland genes from Bombyx (nvd-Bm) and Drosophila (nvd-Dm),which belong to an evolutionally conserved family of Rieske-domain proteins and are expressed specifically in tissues that synthesize ecdysone, including the PG. We also showed through molecular genetics and feeding assays in Drosophila that loss of nvd function in the PG causes growth arrest during larval stages that is due to a reduced ecdysteroid titer. All of our results are consistent with the hypothesis that the nvd family of proteins plays a pivotal role in steroidogenesis. The primary protein structure of Nvd is completely different from that of proteins encoded by other known genes expressed predominantly in the PG, i.e. Halloween P450s(Gilbert and Warren, 2005),Dare (Freeman et al., 1999)and Start1 (Roth et al.,2004).

The spatial and temporal expression pattern of nvd during late embryogenesis and larval development is similar to that of Halloween P450 genes, i.e., phm, dib and sad, in both Bombyx and Drosophila. The expression of nvd-Bm is detected specifically in the PG, and nvd-Dm mRNA is enriched in both the PG and ovary, tissues that synthesize ecdysteroids. Temporal changes in expression are also observed in both nvd-Bm and nvd-Dm, and are correlated with changes in ecdysone titers during larval development. However, it should be emphasized that the early embryonic expression pattern of nvd-Dm is completely different from the patterns of the Halloween P450 genes. During Drosophila embryogenesis, ecdysteroid levels begin to rise around the onset of gastrulation (stages 6-7) and peak at stages 11-12 during germ band retraction (Kraminsky et al., 1980; Moróy et al., 1988; Sullivan and Thummel, 2003). The expression levels of the Halloween genes are strongly correlated with embryonic ecdysteroid titers, and mutations in any of these genes cause ecdysteroid deficiency in the embryo(Gilbert and Warren, 2005; Namiki et al., 2005). By contrast, nvd-Dm expression is not detected at this early embryonic stage, suggesting that nvd-Dm is not necessary to regulate ecdysone production during embryogenesis. Although it is known that maternally synthesized ecdysteroids are provided in Drosophila eggs(Bownes et al., 1988), it has been unclear what type of ecdysteroid precursor is required for ecdysone production during embryogenesis(Chávez et al., 2000). Identifying the exact steps in which Nvd is involved, as discussed below,would shed light on how ecdysone synthesis varies across different developmental stages.

The Nvd proteins have strong similarities to class IA oxygenases, which have both a consensus [2Fe-2S] Rieske-type domain and a domain that is highly similar to the proposed mononuclear non-heme Fe (II) binding motif(Mason and Cammack, 1992). Among the enzymes in this class, the protein encoded by Rhodococcus kshA functions in steroid hydroxylation(van der Geize et al., 2002),raising the possibility that the Nvd proteins could catalyze oxidation or hydroxylation reactions in the ecdysteroid biosynthesis pathway.

In which step of ecdysteroid biosynthesis does Nvd function? Considering the results of our rescue experiments with cholesterol (C) and 7-dehydrocholesterol (7dC), it is likely that the Nvd protein might play a role in transport and/or metabolism of cholesterol and/or its derivatives. This idea is also supported by the recently reported phenotype of Drosophila NPC1 mutants (Fluegel et al., 2005; Huang et al.,2005). Drosophila NPC1 is an ortholog of mammalian Niemann-Pick type C disease genes that are well known to encode cholesterol-binding proteins (Chang et al.,2005). NPC1 mutants show a larval arrest phenotype similar to that of the nvd-Dm RNAi animals discussed above, and this phenotype is rescued to varying degrees by feeding with 20E, C or 7dC.

Our feeding experiments reveal that food containing 7dC completely rescues the nvd-Dm RNAi phenotype, indicating that Nvd acts upstream of 7dC synthesis. Given the fact that such complete rescue activity is not observed when food supplemented with excessive C was used, it is possible that Nvd might function in the conversion of C to 7dC. This hypothesis is consistent with the following points.

1. nvd-Dm is not expressed during mid-embryogenesis. Although ecdysteroidogenesis takes place at this stage, embryos do not consume food containing dietary C. By contrast, the larval expression of nvd-Dmcoincides with the uptake of C, suggesting a role for nvd in C metabolism.

2. The nvd ortholog is conserved in the nematode C. elegans,in which the C to 7dC conversion takes place during steroid biosynthesis even though C. elegans does not have ecdysteroids(Chitwood, 1999).

3. A mutant phenotype of C. elegans ortholog of nvd, as discussed below, is rescued by feeding with 7dC, but not with C(Rottiers et al., 2006).

It is possible, however, that Nvd itself does not directly catalyze C to 7dC. Previous biochemical studies have shown that the conversion of C to 7dC is mediated by an enzyme with cytochrome P450 characteristics under the control of the Zn-finger protein Without Children (Woc)(Grieneisen et al., 1993; Warren et al., 1995; Warren et al., 2001). Our RT-PCR analysis did not detect a significant difference of nvd-Dmexpression level between woc mutant and wild type (see Fig. S6 in the supplementary materials). We also tested whether the Nvd proteins can catalyze cholesterol and cholesterol derivatives (22-hydroxycholesterol,25-hydroxycholesterol and 7dC) using a S2 cell system previously described in biochemical studies of Halloween P450s(Warren et al., 2002; Niwa et al., 2004). However,no metabolites have yet been detected (data not shown).

Although the proposed function for Nvd on the conversion from C to 7dC is most likely, we should point out that food containing excessive C partially rescues the defect of nvd-Dm RNAi animals. This raises the possibility that the partial rescue activity by C might be due to a hypomorphic nature of the nvd-Dm RNAi. In nvd-Dm RNAi animals, we detect a significantly reduced but specific level of nvd-Dm mRNA (Fig. 4A)and substantial 20E (Fig. 5),suggesting that the nvd-Dm RNAi animals might still possess low levels of Nvd activity and thus these animals may be able to produce small amounts of 7dC from C. An alternative explanation is that Nvd would function in the `black box' in which 7dC is converted to theΔ ⁴-diketol by an uncharacterized mechanism, because 7dC is a slightly unstable chemical and it is possible that the rescuing activity of 7dC might be caused by a contaminant oxidated in the food. In order to determine the function of Nvd in more detail, isolation of genetic null mutants of nvd and the biochemical properties of Nvd proteins will need to be investigated.

Our current studies do not address the functional role of the Nvd proteins in organisms other than insects. Notably, a nvd ortholog is found in the C. elegans genome. According to recent advances in taxonomy, both insects and nematodes belong to Ecdysozoa, which are characterized by molting(ecdysis) behavior (Aguinaldo et al.,1997). In C. elegans, it has been suggested that uncharacterized steroid hormones play a role in both molting and dauer formation (Thummel, 2001; Entchev and Kurzchalia, 2005). Therefore, we postulate that Nvd may be important in controlling proper larval development in nematode species. Indeed, during the revision of this paper, it has been demonstrated that the daf-36 gene, which encodes the C. elegans homolog of nvd, is required to bypass the C. elegans dauer diapause (Rottiers et al., 2006). It is thought that daf-36 mutants cannot produce a steroid-like hormone that prevents dauer formation, suggesting that nvd/daf-36 is also essential for a steroid-like synthesis pathway in nematodes. daf-36 mutants are also shown to displays adult aging phenotypes (Rottiers et al.,2006). As reduced 20E in Drosophila is associated with expanded lifespan (Tu et al.,2002; Simon et al.,2003; Colombani et al.,2005), it would be intriguing to explore the role of nvdon insect aging.

Interestingly, no ortholog of nvd has been found in genomes of mammalian species, while nvd orthologs are conserved in other chordates, such as ascidian, zebrafish and chicken. This suggests that nvd orthologs might have been lost during mammalian evolution. Currently, we do not know the exact differences of steroid biosynthesis pathways between mammals and other animals. Future studies on the nvdfamily could provide new insight into the similarities and differences between steroid hormone biosynthesis among animal clades.

Supplementary materialSupplementary material for this article is available at http://dev.biologists.org/cgi/content/full/132/13/2565/DC1

We are grateful to Toshiro Aigaki, Adam Antebi and Veerle Rottiers for informing us their unpublished data. We also thank Yoshinori Fujimoto,Lawrence I. Gilbert, Günter Korge, Joe H. Park, Carl S. Thummel, Tadashi Uemura, James T. Warren, the Bloomington stock center, National Institute of Genetics of Japan and Drosophila Genetic Resource Center of Kyoto Institute of Technology for stocks and reagents; and Frank J. Slack, Diya Banerjee,Katherine Olsson Carter, Aurora Esquela-Kerscher and Helge Großhans for critical reading of the manuscript. R.N. was a recipient of SPD research fellowship of the Japan Society for the Promotion of Science (JSPS). This work was supported by grants to K.M. from the Program for Promotion of Basic Research Activities for Innovative Biosciences and the Insect Technology Project of the Ministry of Agriculture, Forestry and Fisheries of Japan; and to H.K. from Research for the Future Program of the JSPS.