In C. elegans, steroid hormones function in conjunction with insulin/IGF-1-like signaling in promoting reproductive development over entry into the diapausal dauer stage. The NCR-1 and -2 (NPC1-related) intracellular cholesterol transporters function redundantly in preventing dauer arrest,presumably by regulating the availability of substrates for steroid hormone synthesis. We have identified hsd-1 as a new component of this cholesterol trafficking/processing pathway, using an ncr-1 enhancer screen. HSD-1 is orthologous to 3β-hydroxysteroid dehydrogenase/Δ5-Δ4 isomerases(3β-HSDs), which are key steroidogenic enzymes in vertebrates, and is exclusively expressed in two neuron-like XXX cells that are crucial in preventing dauer arrest, suggesting that it is involved in biosynthesis of dauer-preventing steroid hormones. The hsd-1 null mutant displays defects in inhibiting dauer arrest: it forms dauers in the deletion mutant backgrounds of ncr-1 or daf-28/insulin; as a single mutant,it is hypersensitive to dauer pheromone. We found that hsd-1 defects can be rescued by feeding mutant animals with several steroid intermediates that are either downstream of or in parallel to the 3β-HSD function in the dafachronic acid biosynthetic pathway, suggesting that HSD-1 functions as a 3β-HSD. Interestingly, sterols that rescued hsd-1 defects also bypassed the need for the NCR-1 and/or -2 functions, suggesting that HSD-1-mediated steroid hormone production is an important functional output of the NCR transporters. Finally, we found that the HSD-1-mediated signal activates insulin/IGF-I signaling in a cell non-autonomous fashion, suggesting a novel mechanism for how these two endocrine pathways intersect in directing development.
Under environmental conditions that are hostile towards growth, C. elegans enters a diapausal dauer stage instead of the normal 3rd larval stage. The distinctive morphology and physiology of dauer larvae make them resistant to environmental stresses and thus survive until conditions improve,upon which they resume growth towards reproductive adulthood. Genetic analyses of dauer formation in the worm have revealed several conserved signaling pathways (Hu, 2007). Three types of endocrine signals are known to be involved: steroid hormones(Rottiers and Antebi, 2006),insulin/IGF-1-like peptides (Li et al.,2003; Pierce et al.,2001) and TGFβ-like ligands(Ren et al., 1996; Schackwitz et al., 1996). These signals act in a non-redundant fashion, i.e. active signaling by their respective hormone receptor is required to promote reproductive development and prevent dauer arrest.
The role of steroid-signaling in dauer formation became apparent with the identification of several protein components. These include DAF-12/nuclear hormone receptor (Antebi et al.,2000), two steroid-processing enzymes, DAF-9/cytochrome P450(Gerisch et al., 2001; Jia et al., 2002) and DAF-36/mono-oxygenase (Rottiers et al.,2006), and the probable intracellular cholesterol transporters NCR-1 and NCR-2 (Li et al.,2004). Recently, several steroids have been identified as DAF-12 ligands that act to prevent dauer arrest(Motola et al., 2006; Held et al., 2006).Δ 4- and Δ7-dafachronic acids are two such hormones (Motola et al.,2006). Additional hormones may also exist that prevent dauer arrest (Matyash et al.,2004).
A two-branch pathway has been proposed for the biosynthesis ofΔ 4- and Δ7-dafachronic acids from cholesterol (Rottiers et al.,2006). However, the enzymes for several biosynthetic steps remain unknown. Genetic identification of new enzymes will not only help validate/modify the putative dafachronic acid biosynthetic pathway, but also reveal new pathways/networks for the biosynthesis of additional steroid hormones in the worm.
The steroid and insulin/IGF-1-like pathways are likely to converge for coordinated regulation in vivo, as both pathways regulate dauer formation,lifespan and stress resistance (Baumeister et al., 2006; Finch and Ruvkun, 2001). The two pathways also resemble each other in mode of action, as both involve multiple ligands that target one receptor. There are 40 insulin-like peptides in C. elegans and several, including DAF-28, have been reported to have functional relationships with DAF-2/insulin-like receptor (Kimura et al., 1997; Pierce et al.,2001; Li et al.,2003). Although the current model is that steroid-signaling functions downstream of or in parallel to insulin-signaling in governing dauer formation (Rottiers and Antebi,2006), genetic evidence suggests that the two pathways may intersect in a complex manner (Gems et al., 1998). Given that insulin/IGF-1 and steroid pathways also function in concert to regulate development in flies and mice(Tatar et al., 2003),understanding potential crosstalk between the two endocrine pathways in the worm will help reveal conserved mechanisms for hormonal coordination in general.
Nematodes cannot synthesize steroid hormones de novo, but instead modify sterols, such as cholesterol, acquired from their environment(Chitwood, 1999). The C. elegans NCR-1 and NCR-2 (Li et al.,2004; Sym et al.,2000) are both orthologs of human NPC1, an intracellular cholesterol transporter that results in Niemann-Pick type C (NPC) disease when mutated (Carstea et al., 1997). ncr-1 and ncr-2 function redundantly in preventing dauer arrest, as knockout of both but neither alone causes dauer arrest. ncr-1 is thought to be more important for sterol trafficking as it is more broadly expressed than ncr-2(Li et al., 2004). We hypothesized that both NCR proteins are involved in providing cholesterol and/or other sterols for the production of dafachronic acids or related hormones to promote reproductive development.
We identified HSD-1 from an ncr-1 enhancer screen. HSD-1 is related to 3β-HSD enzymes that are essential for producing active steroid hormones in vertebrates. Based on protein similarity and the role of hsd-1 in promoting reproductive growth, we hypothesize that HSD-1 is involved in the biosynthesis of Δ4-dafachronic acid. To determine the function of hsd-1 and dissect the proposed dafachronic acid-biosynthetic pathway, we performed a series of sterol-feeding experiments on hsd-1 mutants. The results supported our initial hypothesis. In addition, our sterol-feeding experiments also addressed substrate specificity of the NCR transporters at the organismal level, which we believe will contribute to greater understanding and subsequent treatment of NPC disease. Furthermore, our findings suggest that the expression of hsd-1 is exclusive to the neuron-like XXX(L/R) cells, and that the HSD-1-mediated steroid signal intersects with the insulin/IGF-I pathway by globally regulating the translocation of the DAF-16/FOXO transcription factor.
MATERIALS AND METHODS
C. elegans strains
C. elegans strains were maintained using standard methods(Sulston and Hodgkin, 1988). The following parental strains were used in this report: N2 (wild type),CB4856, hsd-1(mg433) (6× outcrossed), ncr-2(nr2023),ncr-1(nr2022), daf-36(k114), unc-75(e950), daf-28(tm2308) (3×outcrossed), zIs356[Pdaf-16::DAF-16::GFP;rol-6(su1006)]. Molecular methods were used to construct strains, in addition to mutant phenotypes.
ncr-1 enhancer screen and mg433
An ncr-1 enhancer screen was designed to identify mutants that phenocopy ncr-2(nr2023); ncr-1(nr2022), which form transient, pale dauers at all temperatures. A synchronized L4 larval population of ncr-1(nr2022) (P0) was mutagenized with EMS(Sulston and Hodgkin, 1988). F1 and F2 animals were cultivated at 20°C. Candidate mutants from synchronized F2 animals, which represented the progeny of ∼9000 F1s, were divided into 18 pools and then subjected to a 30 minute treatment in 0.5% SDS solution. Pale dauers were then selected from surviving animals and recovered at 20°C. Sixteen mutant isolates were obtained from 10/18 pools, of these mg433 displayed the highest penetrance of dauer arrest and was selected for further analysis. unc-75(e950) was used as a positional marker to introduce a wild-type stretch of chromosome I to the left of mg433 for crossing out a background mutation.
Constructs for hsd-1 expression and rescue
The expression pattern of hsd-1 was determined with gfp(Chalfie et al., 1994) fused to the hsd-1 promoter, which contained 10.5 kb 5′ sequence of hsd-1, including the entire lagr-1 gene. This design was based on a previous prediction in WormBase that these genes form an operon. Phsd-1::GFP was generated by in vivo homologous recombination(Aroian et al., 1990) of two injected PCR fragments: the first contains 3.2 kb of promoter and most of lagr-1; the second includes the 3′ region of lagr-1,the intergenic region between lagr-1 and the ATG of hsd-1fused with gfp from pPD95.70, using a PCR-fusion method(Hobert, 2002). pha-1DNA was used as the co-injection marker and a pha-1temperature-sensitive mutant was used for injection. Psdf-9/eak-5::RFP is as described by Hu et al.(Hu et al., 2006).
Two hsd-1 transgenes were generated to test for rescue of hsd-1 defects: Psdf-9::hsd-1 and Phsd-1::hsd-1. Psdf-9::hsd-1 contains the 3.5 kb sdf-9 promoter fused to the hsd-1-coding region and 740 bp downstream sequences. Phsd-1::hsd-1 contains the 275 bp 5′ sequence of hsd-1, the coding region and 180 bp downstream sequence. The co-injection marker was sur-5::gfp(Yochem et al., 1998).
Dauer assays, sterol feeding and pheromone treatment
Two different methods were used to synchronize worm populations for dauer assays. The first was hypochlorite treatment of gravid adults as described by Sulston and Hodgkin (Sulston and Hodgkin,1988). The second was timed egg-laying with seven to eight hermaphrodites per plate, allowing them to lay eggs for 5-6 hours before removing them. The window of time for scoring dauers versus bypassors (L4 or adults) at 25°C is 48-72 hours after egg preparation or egg-laying (65-72 hours for strains containing the ncr-1 deletion and 48-55 hours for other strains). For dauer assays at 20°C and 15°C, worms were scored 70-76 and 94-100 hours after egg preparation/laying, respectively. The transgenic rescue of hsd-1; ncr-1 transient dauers was monitored using the developmental timing of synchronized populations, and the percentages of rescued transgenic animals were determined at a time point of 68 hours after egg preparation, when the control hsd-1; ncr-1 animals were still virtually all dauers and N2 animals were adults.
Agar plates containing different sterols were prepared according to the standard procedure for NGM plates (Sulston and Hodgkin, 1988), except that cholesterol was replaced by different sterols at the equivalent concentration of 12.9 μM (1×). Stock solutions of sterols, dissolved in ethanol, were added into liquid agar media at a high temperature to allow even distribution. Sterols used in this study are as follows: cholesterol, 4-cholesten-3-one, 5-cholesten-3-one,lathosterol, lathosterone, 7-dehydrocholesterol, β-sitosterol/sitosterin and ergosterol (Sigma-Aldrich, St Louis, MO, USA; Steraloids, Newport, RI,USA; Research Plus, Barnegat, NJ, USA).
Crude pheromone extract was obtained as described by Golden and Riddle(Golden and Riddle, 1984). Pheromone extract was diluted with 750 μl of OP50 culture and distributed over the entire surface of a 3 ml NGM plate. Plates were then dried in a tissue culture hood until no liquid was seen on the surface (40-70 minutes). The egg preparation method was used to obtain synchronized populations for the pheromone assay. Bacterial food was always in excess during experiments, which were carried out at 25°C.
Analysis of GFP patterns and imaging
DAF-16::GFP transgenic strains were propagated at 15°C. Five to eight young adults were allowed to lay eggs for 4 hours and develop at 20°C. Animals were scored for GFP subcellular localization under a fluorescent dissecting scope. The fluorescent images in Fig. 1A-C were obtained using a compound scope with 40× optics and Fig. 1D-F using a 16×lens.
Genetic identification of a steroid-processing enzyme that functions in the ncr-1 pathway
In order to identify new components of the cholesterol trafficking/processing pathway that regulates dauer arrest, we performed a genetic screen using the ncr-1(nr2022) deletion as a sensitized background. Although ncr-1 animals have overall wild-type development, they are thought to have attenuated cholesterol transport, based on reported minor phenotypes (Li et al.,2004). We performed an EMS screen for transient dauers resembling ncr-1; ncr-2 mutant animals. From this screen we isolated mg433 (for phenotypic analyses, see Table 1) and subsequently mapped it, using the SNP method (Wicks et al., 2001). The analyses of the SNP markers of 55 recombinants on chromosome I located mg433 to the interval between the cosmid M01E5 and the YAC Y71A12B, which are about 650 kb or 5 cM apart on the physical and genetic maps, respectively. We sequenced three candidates in this interval and found that mg433 is a mutation in the gene Y6B3B.11. This gene is predicted to encode a protein with extensive similarity to the 3β-hydroxysteroid dehydrogenase/Δ5-Δ4isomerase family of enzymes, thus we designated Y6B3B.11 as hsd-1. mg433 has a G to A transition that results in an early amber stop codon at Trp45, suggesting that mg433 is a molecular null. Transgenic expression of hsd-1 fully rescued the constitutive dauer formation phenotype (referred to as dauer arrest throughout this report) of hsd-1(mg433); ncr-1(nr2022) (see below), verifying that hsd-1 is involved in the inhibition of dauer formation and that mg433 impairs this function.
Using reciprocal BLASTP, we determined that hsd-1 is an ortholog of the predicted human gene HSPC105; they share 20.5% identity and 32.0%similarity. In addition, among the top hits from the BLASTP of HSD-1 against the human RefSeq protein database are the following 3β-HSDs: NSDHL,HSD3B7 and HSD3B1, which are important for the biosynthesis of cholesterol,bile acids and steroid hormones, respectively. These enzymes are 15.5%, 15.8%,16.7% identical and 27.0%, 25.7%, 26.5% similar to HSD-1, respectively. The 3β-HSD proteins are fairly short (359-462 amino acids) and share a number of very small conserved motifs across their whole length (data not shown). A major role of these mammalian enzymes is to generateΔ 4-3-keto-sterols and two are implicated in human diseases(Konig et al., 2000; Schwarz et al., 2000). The extensive similarity between HSD-1 and the human 3β-HSDs suggests that HSD-1 is involved in producing Δ4-3-keto-steroids such asΔ 4-dafachronic acid. HSD-1 also has two paralogs in C. elegans, HSD-2 (ZC8.1) and HSD-3 (ZC449.6).
The XXX cells serve as the endocrine tissue for HSD-1-mediated steroid signaling
Using the hsd-1 promoter to drive a GFP reporter, we determined that hsd-1 is exclusively expressed in a pair of neuron-like cells in the head. We surmised that these were the XXX(L/R) cells, because these cells are not only important for preventing dauer arrest(Hu et al., 2006; Ohkura et al., 2003), but are also thought to be involved in steroidogenesis, based on the expression patterns of NCR-1, NCR-2 and DAF-9(Gerisch and Antebi, 2004; Jia et al., 2002; Li et al., 2004; Mak and Ruvkun, 2004). To confirm this prediction, we examined whether the expression of Phsd-1::GFP co-localizes with that of Psdf-9/eak-5::RFP. The sdf-9 gene is expressed only in the XXX cells(Hu et al., 2006; Ohkura et al., 2003) and we observed co-localization of both markers(Fig. 1A-C). Consistent with this, a transgene containing the hsd-1-coding sequence driven by the sdf-9 promoter, rescued the dauer arrest of hsd-1(mg433);ncr-1(nr2022) (Table 2). The expression of Phsd-1::GFP was initially observed at the pretzel stage of embryogenesis. This XXX cell-specific expression persists through all four larval stages and becomes fainter in adults, which is consistent with the view that the XXX cells are an important endocrine tissue in regulating C. elegans larval development.
hsd-1 promotes reproductive growth
hsd-1(mg433); ncr-1(nr2022) animals undergo dauer arrest transiently at all temperatures tested(Table 1), similar to that seen in ncr-2(nr2023); ncr-1(nr2022)(Li et al., 2004) and the mutants of DAF-12 that are defective in steroid-ligand binding, such as rh273 (Antebi et al.,1998). hsd-1; ncr-1 dauers can resume reproductive growth at all temperatures, including the relatively high 25°C. We compared the dauer recovery rates for hsd-1; ncr-1 and ncr-2; ncr-1 at 25°C (Table 1) and found that hsd-1; ncr-1 dauers appeared to recover slightly slower. We found that hsd-1; ncr-1 dauers recovered poorly and had abnormalities such as protruding-vulvas and `bagging' (matricide by internal hatching) phenotypes (9/25 animals), similar to ncr-2; ncr-1 (8/25). Like ncr-2, ncr-1, we also found that hsd-1; ncr-1 has a reduced brood size (Fig. 2A). In the absence of dauer-preventing steroids, such as in daf-9 or ncr-2; ncr-1, the un-liganded form of DAF-12 receptor is known to be required for dauer formation (Rottiers and Antebi, 2006). Similarly, we found that hsd-1; ncr-1dauer arrest at 25°C can be suppressed by RNAi of daf-12(Table 1). Similar to mutants of other steroid-signaling components, except for null mutants of daf-12,hsd-1; ncr-1 is capable of forming natural dauers under the appropriate physiological conditions. To conclude, in a similar manner to DAF-9, the HSD-1-mediated steroid signal normally promotes reproductive growth.
The hsd-1(mg433) single mutant appears to develop normally;however, we observed that it forms dauers more readily than the wild-type N2 under the high population densities present in continuous plate culture. Dauer pheromone is constitutively produced and released by worms, serving as an external signal for judging and regulating population density(Golden and Riddle, 1982; Golden and Riddle, 1984). In order to determine whether the hsd-1 single mutant is more sensitive to induction by dauer pheromone, we examined the response of hsd-1 to a crude extract of dauer pheromone. We found that hsd-1 forms dauers at 98.2% and 56.7% with 8 and 4 μl of pheromone extract per plate,respectively (Fig. 2B). At the same pheromone concentrations, N2 did not form dauers, indicating that hsd-1 is hypersensitive to pheromone induction. In fact, it took approximately 30× higher pheromone concentration to induce dauers in N2 than hsd-1 (Table 1). We also tested whether ncr-1, for comparison with hsd-1, is hypersensitive to pheromone and found that although it does form some dauers,it is not as sensitive (Fig. 2B). This result suggests that hsd-1 prevents dauer arrest at population densities lower than those that would normally trigger dauer formation in wild-type worms, which is consistent with a role for hsd-1 in preventing dauer arrest in the ncr-1 background. However, in the range of population densities used in this report, hsd-1 animals rarely form dauers without pheromone(Fig. 3), suggesting that our dauer assays were not influenced by population density.
To investigate possible synergism between hsd-1(mg433) and other steroid-pathway null mutants that display only weak dauer arrest phenotypes,we constructed hsd-1(mg433); ncr-2(nr2023) and hsd-1(mg433);daf-36(k114) double mutants. In contrast to hsd-1(mg433);ncr-1(nr2022), hsd-1(mg433); ncr-2(nr2023) animals did not display dauer arrest, consistent with the notion that NCR-1 contributes more to cholesterol trafficking than does NCR-2, and/or suggesting that the functional output of NCR-2 completely relays into that of HSD-1. The putative daf-36 null allele k114 displays only a weak dauer arrest phenotype at 25°C(Rottiers et al., 2006). We found that hsd-1(mg433); daf-36(k114) displayed 100% dauer arrest at 25°C (Table 1). This result suggests that, compared with HSD-1, DAF-36 functions in a parallel biosynthetic pathway, a different tissue, or both. Rottiers et al.(Rottiers et al., 2006) have previously shown that DAF-36 is expressed in the intestine and may function in the Δ7-branch (Fig. 4A), which is consistent with the observed synergism.
We also investigated possible synergism between hsd-1(mg433) and daf-28(tm2308), a deletion mutant of an insulin. The transient dauer phenotype of the dominant-negative mutant daf-28(sa191) suggests that DAF-28 functions as an agonist of the DAF-2 receptor(Li et al., 2003; Malone and Thomas, 1994). We found that 28.5% of daf-28(tm2308) undergo dauer arrest at 25°C(Table 1). Unlike the single mutants of either hsd-1 or daf-28, hsd-1(mg433);daf-28(tm2308) animals form 89.6% transient dauers at 25°C(Table 1). This synthetic phenotype suggests that, although the insulin and steroid pathways are non-redundant with respect to each other, a subset of ligands from each pathway can be redundant in preventing dauer arrest. Interestingly, the transient dauers of hsd-1(mg433); daf-28(tm2308) have mixed characteristics of dauers induced by either reduced insulin or steroid signaling. They are large dauers, resembling the mutants of the DAF-2 receptor, but also pale and active, resembling ncr-2; ncr-1 dauers(data not shown).
HSD-1 functions as a 3β-hydroxysteroid dehydrogenase/Δ5-Δ4 isomerase inΔ 4-steroid signaling
Biochemical and structural studies on the 3β-HSD family of enzymes,especially on HSD3B1 (Thomas et al.,2003), have shown that they carry out a sequential two-step reaction: (1) the dehydrogenase activity oxidizes the 3β-hydroxyl group of a sterol; (2) the isomerase activity switches the position of the double carbon bond on the sterol ring from Δ5 to Δ4(Fig. 4A). Thus, we hypothesize that HSD-1 functions in the Δ4-dafachronic acid branch by converting cholesterol to 4-cholesten-3-one(Fig. 4A).
In order to test whether dietary supply of 4-cholesten-3-one could compensate for the lack of hsd-1 activity, we first examined the hypersensitivity of hsd-1(mg433) to dauer pheromone when grown on agar plates containing 4-cholesten-3-one instead of cholesterol. We observed substantial rescue of dauer arrest with 4-cholesten-3-one, at two different pheromone concentrations (Fig. 5A). We investigated whether 4-cholesten-3-one could rescue synthetic dauer arrest associated with the double mutants of hsd-1(mg433) with daf-28(tm2308), daf-36(k114) and ncr-1(nr2022), and we observed robust rescue(Fig. 5B; Fig. 6A,B). As a control, we also tested the predicted product of the first step of the HSD-1 reaction,5-cholesten-3-one, which contains the C-3 ketone group but still has theΔ 5-ring configuration(Fig. 4A). We found that 5-cholesten-3-one did not rescue hsd-1; ncr-1 as well as 4-cholesten-3-one (Fig. 6A,B),suggesting that both structural characteristics of 4-cholesten-3-one contribute to the rescue of hsd-1(mg433). The above results are consistent with the hypothesis that HSD-1 converts cholesterol to 4-cholesten-3-one in vivo, or alternatively, HSD-1 modifies a derivative of cholesterol into a Δ4-3-keto-sterol.
Δ7-Dafachronic acid, another dauer-preventing hormone, is thought to be produced in parallel to Δ4-dafachronic acid from cholesterol (Fig. 4A)(Motola et al., 2006). If hsd-1 is required by the Δ4-biosynthetic branch,raising the level of Δ7-signaling by increasing the levels ofΔ 7-intermediates may compensate for impairedΔ 4-dafachronic acid biosynthesis. We found lathosterol and lathosterone, two Δ7-intermediates, can suppress both hsd-1(mg433) dauer pheromone hypersensitivity and synthetic dauer arrest in the double mutants of hsd-1 with daf-28(tm2308),daf-36(k114) and ncr-1(nr2022)(Fig. 5A,B; Fig. 6A,B). Higher concentrations of lathosterol were required for full suppression of hsd-1;ncr-1 dauer arrest (Fig. 6A). By contrast, cholesterol was unable to suppress this phenotype at any of the concentrations tested(Fig. 6A). In a titration experiment to compare the efficacy of the sterols in alleviating hsd-1-associated dauer phenotypes, we found that, like 4-cholesten-3-one, lathosterone is quite potent at a quarter of the standard 12.9 μM concentration of cholesterol. Lathosterone still displayed rescue at an eighth this concentration (Fig. 6B). The physiological concentration forΔ 4-dafachronic acid is estimated to be 200 nM(Motola et al., 2006), and our experiment showed that lathosterone at a concentration ∼8-fold greater than this can be effective. Interestingly, 7-dehydrocholesterol, the first intermediate in the Δ7-branch, did not suppress hsd-1-associated dauer arrest(Fig. 5A,B; Fig. 6A), except in hsd-1;daf-36, where the suppression is probably due to the rescue of daf-36, as this is the predicted end product of the DAF-36 reaction(Fig. 4A). The above results suggest that some sterol intermediates in the parallelΔ 7-branch can compensate for the lack of HSD-1-mediated Δ4-dafachronic acid production.
The requirement for the NCR-1 and/or -2 transporter function can be bypassed by dietary supply of certain sterol intermediates
hsd-1(mg433) was identified as an enhancer of ncr-1(nr2022), with the double mutant forming transient dauers when grown on cholesterol-containing media. We considered the possibility that one role of NCR-1 or -2 transporters is to mobilize and display cholesterol for HSD-1-mediated 3β-dehydrogenation. According to this hypothesis,sterol intermediates beyond the point of the HSD-1 reaction, or those in the parallel Δ7-branch, should be able to bypass the requirement for the NCR-1 and NCR-2 transporters in preventing dauer arrest.
We found that dauer arrest in ncr-2(nr2023); ncr-1(nr2022) was suppressed when cultivated on the following intermediates: 4-cholesten-3-one,5-cholesten-3-one, lathosterol and lathosterone(Fig. 6C), suggesting that the NCR-1 and -2 transporters are only required for the transport/mobilization of certain sterols, such as cholesterol and perhaps 7-dehydrocholesterol. Alternative mechanisms may exist for mobilizing other types of sterols, such as lathosterol, for steroidogenesis. We also tested the major plant and fungal sterols, β-sitosterol and ergosterol, respectively (for structures, see Fig. 4B) and found that, like cholesterol, they did not suppress ncr-2(nr2023); ncr-1(nr2022),suggesting that these sterols are NCR-1/-2-dependent and thus likely occupy a similar position to cholesterol in the metabolic pathway. β-Sitosterol,ergosterol and 7-dehydrocholesterol are all known to be able to substitute for cholesterol in supporting the growth of wild-type animals(Chitwood, 1999; Hieb and Rothstein, 1968). Given that the sterols that alleviated hsd-1-associated defects also appeared to suppress ncr-2; ncr-1, HSD-1-mediated hormone production appears to be an important functional output of the NCR transporters.
The NCR-independent sterols block crosstalk between steroid and insulin signaling
In order to detect any potential crosstalk between steroid and insulin/IGF-1-like signaling, we examined the subcellular localization of DAF-16::GFP in hsd-1(mg433); ncr-1(nr2022) and ncr-2(nr2023);ncr-1(nr2022). Upon downregulation of DAF-2, DAF-16/FOXO, a transcription factor in the insulin pathway, is activated after translocating from the cytoplasm to the nucleus (Henderson and Johnson, 2001; Lee et al.,2001; Lin et al.,2001). In wild-type animals, DAF-16::GFP is localized to the cytoplasm, thus displaying a diffuse pattern(Fig. 1D). When DAF-16::GFP is in the nucleus, it displays a punctate pattern. We found that 29% of hsd-1; ncr-1 and 60.6% of ncr-2; ncr-1 animals displayed punctate DAF-16::GFP at the L2d stage, a preparatory stage prior to dauer formation (Table 3). The difference between the two double mutants suggests that the downregulation of steroid signaling is more severe in ncr-2; ncr-1 than in hsd-1;ncr-1, although dauer arrest between the two is comparable. We observed variable degrees of DAF-16 nuclear localization amongst individuals of both hsd-1; ncr-1 and ncr-2; ncr-1 mutants. For example, Fig. 1E shows a mix of diffuse and punctate DAF-16::GFP, whereas Fig. 1F displays a punctate pattern. Interestingly, we observed that in both double mutants, DAF-16::GFP is often more punctate in the posterior of the worm, as though insulin signaling is more active in the head compared with the tail (Fig. 1E, see Discussion). In addition, we observed that a complete punctate pattern is often found among the double mutants close to or during the molting period prior to dauer formation (Fig. 1F). However, DAF-16::GFP is primarily diffuse amongst dauers(Table 3). These results suggest that the insulin pathway is partially, or transiently, downregulated in both double mutant backgrounds prior to dauer formation.
When hsd-1(mg433); ncr-1(nr2022) and ncr-2(nr2023);ncr-1(nr2022) animals were grown on plates with either 4-cholesten-3-one(Table 3) or 2×lathosterol (data not shown), we observed diffuse DAF-16::GFP, similar to wild-type controls (Fig. 1D). Thus, rescue of steroid-signaling by 4-cholesten-3-one also results in restoring insulin-pathway activity to wild-type levels in the double mutants.
HSD-1 may function as a 3β-hydroxysteroid dehydrogenase/Δ5-Δ4 isomerase
From an enhancer screen for mutations that would further impair sterol trafficking/processing mediated by the probable intracellular cholesterol transporter NCR-1 in regulating dauer arrest, we identified HSD-1, which has extensive similarity to mammalian 3β-HSDs. The human HSPC105 protein appears to be the true ortholog of HSD-1, based on reciprocal BLASTP. Although very little is known about HSPC105, several other human 3β-HSDs are important for development or physiology. The major 3β-HSD function is to generate Δ4-3-keto-sterols, which is thought to be required for the formation of active vertebrate steroid hormones(Payne and Hales, 2004). Although vertebrate steroid hormones lack the long side chain that is present in the dafachronic acids, the presence of a C-3 ketone group appears to be a shared characteristic (Motola et al.,2006). The genetic identification of HSD-1 in C. eleganshas reinforced this notion.
Our sterol-feeding experiments showed that 4-cholesten-3-one rescued hsd-1- associated defects more effectively than 5-cholesten-3-one,suggesting that both structural changes enacted by the dehydrogenase and isomerase activities of 3β-HSD contribute to rescuing hsd-1. However, our in vivo experiments did not identify the true or preferred substrate(s) of HSD-1. Conceivably, the substrate(s) of HSD-1 is cholesterol or its derivatives. Our experiments also did not address why 5-cholesten-3-one could rescue the hsd-1 defects. There are several possible reasons why 5-cholesten-3-one can substantially rescue hsd-1; ncr-1:(1) this Δ5-sterol may be converted toΔ 4-sterol by enzymes with ketosteroid isomerase activity,such as GSTs with a similar function to GST A3-3 in humans(Johansson and Mannervik,2001); (2) this Δ5-sterol may be processed into aΔ 5-steroid similar to 5-cholesten-3β-ol-(25S)-carboxylic acid, which has dauer-preventing activity(Held et al., 2006); (3) thisΔ 5-sterol may be modified into dauer-preventing hormones that remain unidentified. Therefore, it is possible that the in vivo metabolic pathways/networks are more complex and plastic than the proposed pathway for dafachronic acids, in both protein and sterol/steroid contents. Nevertheless,our results are consistent with the hypothesis that HSD-1 functions as a 3β-HSD in vivo.
hsd-1-mediated Δ4 steroid signaling in the XXX cells
In C. elegans, steroid signaling is thought to be present in multiple tissues. Besides the XXX cells, steroids are believed to be produced in the hypodermis, intestine and somatic gonads(Gerisch et al., 2001; Jia et al., 2002; Mak and Ruvkun; Rottiers et al.,2006). hsd-1 is the first steroidogenic enzyme that displays expression exclusive to the XXX cells. This expression pattern is not all that surprising: it is known that the XXX cells are where ncr-2is predominantly expressed in young larvae(Li et al., 2004), when the dauer decision is made. hsd-1 and ncr-2 both enhance ncr-1, resulting in dauer arrest and thus may share a similar focus in their functions, which may be to promote the generation of steroids in XXX cells. The molecular identity of HSD-1 and our sterol-feeding results suggest that HSD-1 is likely to function in Δ4-dafachronic acid biosynthesis (Fig. 4A). Thus,we speculate that the phenotypes associated with the hsd-1(mg433)null in this report represent the impairment of Δ4-signaling in the XXX cells.
hsd-1-mediated Δ4-signaling in XXX cells contributes to promoting reproductive growth, which is also indicated by the fact that hsd-1 becomes crucial when the population density reaches a certain level. Dauer pheromone is a surrogate measure of population density. The degree of hsd-1(mg433) hypersensitivity to pheromone implies that, without hsd-1, a growing worm culture would begin to form dauers at a much lower population density(Fig. 3, Table 1). The involvement of hsd-1 in promoting growth is also indicated by its synergism with several genes in preventing dauer arrest: ncr-1, daf-28 and daf-36. Together, we conclude that HSD-1 becomes crucial in the growth versus dauer decision, when population density is relatively high,cholesterol trafficking is attenuated, insulin-signaling is reduced or a parallel steroid-signaling event is compromised.
The expression of both hsd-1 and ncr-2 is limited to the head at the stage when the dauer versus growth decision is made. Given that the formation of dauers requires remodeling of multiple tissues(Cassada and Russell, 1975),including the pharynx and hypodermis, the XXX cell-expressed HSD-1 can apparently function from a distance. Consistent with this prediction, we found that the influence of hsd-1/ncr-2-mediated steroid signaling on DAF-16::GFP, which is ubiquitously expressed, appears to affect many cells(Fig. 1E,F). This is consistent with the cell non-autonomous nature of steroid hormone action. Under an epifluorescent dissecting scope, we observed that DAF-16::GFP is often found fully confined to the nucleus in the posterior of the worm, but gradually becomes more cytoplasmic towards the anterior end in both hsd-1;ncr-1 and ncr-2; ncr-1 (Fig. 1E). This implies that there is a decreasing gradient of hormones that prevent the downregulation of insulin signaling or DAF-16 nuclear localization, running in an anterior-posterior direction. Alternatively, as a recipient tissue for hormonal signals, posterior cells maybe more prone to downregulation of insulin signaling than their anterior neighbors in hsd-1; ncr-1 or ncr-2; ncr-1 animals.
Our finding that DAF-16::GFP undergoes nuclear translocation upon the reduction of steroid signaling suggests a novel mechanism for the cell non-autonomous coordination of these two endocrine pathways in directing development. It has been previously shown that a steroid/lipophilic signal induced by the loss of germline cells extends lifespan by promoting nuclear translocation of DAF-16, indicating that steroid-signaling negatively regulates the insulin pathway (Gerisch et al., 2001; Hsin and Kenyon,1999; Lin et al.,2001). Our finding suggests a comparable epistatic relationship between the two endocrine pathways in dauer regulation, except that in the latter case, the presence of steroid hormones upregulates insulin signaling. Previously, steroid signaling was thought to function downstream of or in parallel to insulin signaling in regulating development(Rottiers and Antebi, 2006). We believe that our findings are not contradictory to the previous model;instead, it reflects that these two endocrine pathways interact in a complex manner.
Redundancy of steroid signaling in C. elegans
Based on comparisons between the null mutant phenotypes, expression patterns and biochemical roles of daf-9 and hsd-1, we predict that hsd-1 functions in the daf-9 pathway, but is not the sole input to daf-9. daf-9 worms display a severe dauer arrest phenotype (Gerisch et al.,2001; Jia et al.,2002), whereas hsd-1 worms only show hypersensitivity to dauer pheromone; daf-9 is expressed in multiple tissues(Gerisch and Antebi, 2004; Mak and Ruvkun, 2004), whereas hsd-1 is expressed only in the XXX cells. DAF-9 is involved in biosynthesis of both Δ7- and Δ4-dafachronic acids (Motola et al., 2006),whereas HSD-1 is likely only to be involved in the latter. We speculate that HSD-1 provides a C-3 keto-sterol precursor for the production ofΔ 4-dafachronic acid by DAF-9 in XXX cells, whereas the role of producing C-3 keto-sterols in other pathways or tissues may be carried out by other 3β-HSDs or sterol dehydrogenases, such as HSD-2 and HSD-3. We speculate that a complex sterol metabolic network involving multiple branches or alternative pathways exists to allow versatile regulation of development in response to different environmental conditions. It is known that brassinolide signaling in plants, which regulates development in response to light,comprises such a network (Thummel and Chory, 2002).
If Δ7- and Δ4-dafachronic acids function redundantly in preventing dauer arrest, raising the level ofΔ 7-signaling should compensate for the lack of HSD-1-mediatedΔ 4-signaling. We found that two predictedΔ 7-intermediates, lathosterol, a 3β-hydroxysterol, and lathosterone, could suppress hsd-1-associated defects. Conceptually,suppression is no different from rescue. Therefore, one possible interpretation of this result is that hsd-1 functions upstream of the generation of lathosterol. However, based on protein similarity, HSD-1 is likely to produce Δ4-3-keto-sterols(Fig. 4A). Therefore, it is more likely that increased Δ7-biosynthesis accounts for the ability of lathosterol to suppress hsd-1 defects. By contrast,feeding of 7-dehydrocholesterol does not compensate for the impairment ofΔ 4-signaling (Fig. 5A,B; Fig. 6A),which could result from a rate-limiting step in the trafficking/processing of 7-dehydrocholesterol into later steroid intermediates, as discussed later.
Steroidogenesis, cholesterol intracellular trafficking and Niemann-Pick type C disease
hsd-1 is an enhancer of ncr-1, suggesting that steroidogenesis and NCR-mediated cholesterol transport are interdependent. The following are three non-mutually exclusive models that interpret why the hsd-1 mutant only displays dauer arrest when ncr-1 is deleted. (1) When both NCR transporters are present, steroid signaling other than hsd-1-mediated Δ4-signaling in XXX cells,including Δ7, may be sufficient to prevent dauer arrest;however, when NCR-1 is removed, Δ7-signaling, for example,may be attenuated due to insufficient sterol substrates, and thus HSD-1 activity becomes crucial. (2) The broadly expressed NCR-1 may be involved in sterol absorption, in addition to intracellular trafficking; thus, a lower level of cholesterol absorption in the ncr-1 mutant makes HSD-1 activity necessary in preventing dauer arrest. In mice and flies, the NPC1-related proteins NPC1L1 and dNPCb are known to be involved in dietary sterol absorption in the intestine (Altmann et al., 2004; Voght et al.,2007). In worms, although the route for initial sterol absorption remains unknown, the intestinal expression of NCR-1 would be consistent with a possible role in sterol absorption. (3) Perhaps 3β-hydroxyl steroids can prevent dauer arrest when produced in sufficient amounts, which may be permitted by the presence of both NCR transporters. This model is supported by findings showing that a 3β-hydroxyl steroid similar to dafachronic acids,5-cholesten-3β-ol-(25S)-carboxylic acid, transactivates the DAF-12 receptor in vitro (Held et al.,2006). These models are not mutually exclusive.
NPC1 is thought to mobilize cholesterol out of endo/lysosomes and redistribute it to other cellular compartments through vesicular transport(Ko et al., 2001). Mutations in NPC1 are causal to 95% cases of the fatal neurodegenerative NPC disease(Carstea et al., 1997). NPC1-related transporters potentially are involved in the transport of other sterols/lipids as well. However, little is known about substrate specificity of these transporters in general. Our sterol-feeding studies suggest that intermediates that are either beyond the point of the HSD-1 reaction, such as 4-cholesten-3-one, or the ones that do not need HSD-1 for further processing,such as lathosterol, are independent of the NCR-1 and/or NCR-2 transporters for their trafficking amongst cellular compartments/organelles. Our data further suggest that the function of the NCR transporters is likely to be limited to trafficking common dietary sterols, or those used for storage in vivo. We demonstrated that, besides cholesterol, the NCR transporters are indispensable when worms are fed with either ergosterol or β-sitosterol,which may well be the main dietary sterols available to C. elegans in the wild. To control the rate of hormone production accurately, it would be appropriate if restrictive regulation were applied to intracellular trafficking of common dietary sterols, because the intake of these sterols would most probably fluctuate in quantity. We speculate that the NPC1 pathway may serve as one such regulatory `hurdle'. This potential regulation also appears to apply to 7-dehydrocholesterol, which is known to be a major sterol component of worms (Chitwood,1999) and thus may be a storage sterol.
The mechanisms resulting in NPC neurodegeneration remain unclear. Although NPC1 was thought to have only general cell biological roles, recent studies in animal NPC models suggest that steroidogenesis could be an important functional output of NPC1-related transporters. Emerging evidence from invertebrate models of NPC disease, including this study, suggests that NCR transporters have a conserved role in supplying sterol substrates for the biosynthesis of dauer-preventing hormones in worms(Li et al., 2004; Motola et al., 2006) and of molting hormones in flies (Fluegel et al.,2006; Huang et al.,2005).
Currently, there are no effective therapeutic treatments for NPC disease. In mice, it has been demonstrated that a combination of the neurosteroid allopregnanolone and a synthetic oxysterol can promote neuronal survival and mildly suppress the lethality of NPC1-deficient mice(Griffin et al., 2004; Langmade et al., 2006). Neurosteroids have thus become a possible direction for development of drugs to treat NPC disease (Mellon et al.,2008). However, the therapeutic effects of the reported treatment in the mouse model are quite mild. As defects in the NPC1 protein impact the production of multiple known neurosteroids(Mellon et al., 2008) and potentially others awaiting identification, an effective treatment may rely on the use of steroid intermediates that are trafficked independently of NPC1 and can still be processed into a diverse range of hormones. Our results showing effective bypassing of the NCR transporters in the worm suggests that a similar steroid intermediate approach may also work in mammals.
We thank Shyam Bhansali for technical assistance; Patrick Hu, Sarah Pierce,James Priess, Leo Pallanck, Steven Voght, David Raible, Wenqing Xu, Rachel Wong and John Clark for critical reading of the manuscript; and James Thomas,Adam Antebi, Xi Wang, Ho Yi Mak, Andrew Prendergast, Thomas Bird and Robert Waterston for helpful discussions. We are grateful to Shohei Mitani (National Bioresource Project, Japan) and the Caenorhabditis Genetics Center for strains used in this study. G.R. is supported by NIH funding. We also thank the Nathan Shock Center of Excellence in the Basic Biology of Aging and the Jim and Lynne Lambright Niemann-Pick foundation for their generous support.