A Novel 3-Hydroxysteroid Dehydrogenase That Regulates Reproductive Development and Longevity

A multidisciplinary approach identifies novel biochemical activities involved in the synthesisof C. elegans bile acid-like steroids, which act as hormones that regulate sterol metabolism and longevity.


Introduction
Small molecule metabolites and peptide hormones are well known to regulate various aspects of animal physiology, metabolism, and homeostasis. More recently they have emerged as important modulators of life span. Notably, a modest reduction in insulin/IGF-1 signaling (IIS) has been shown to increase longevity in several organisms from worms to mice and possibly humans [1]. Several naturally occurring metabolites including spermidine, trehalose, endocannabinoids, and oleic acid are among a handful of small molecules impacting longevity in model systems [2][3][4][5]. Metabolites of cholesterol, such as steroids, oxysterols, and bile acids, act through cognate nuclear hormone receptor transcription factors (NHRs) to regulate gene expression [6,7]. Of these sterol metabolites, bile acids are primarily known for their roles in dietary fat absorption, but are increasingly recognized as important signaling molecules, regulating aspects of cholesterol, glucose, and fatty acid metabolism through the control of sterolsensing NHRs, including FXRa, LXR, and G-protein coupled receptors [8][9][10].
In C. elegans, DAF-12, a homolog of FXRa and LXR, is activated by the bile acid-like dafachronic acids (DAs) and governs key events that influence longevity [11,12]. In particular, DAF-12 regulates the decision to undergo reproductive development or arrest at the dauer diapause, an alternative developmental stage characterized by stress resistance and extended longevity [12][13][14]. DAF-12 also regulates adult longevity in response to signals from the gonad. Loss of the germline through laser microsurgery or genetic manipulation leads to an extended lifespan, dependent upon DAF-12 and its ligands [1,15,16]. The discovery of endogenous ligands for DAF-12, which include D 4 -and D 7dafachronic acids and the structurally related 25-S-cholestenoic acid, provided the first evidence that bile acid-like molecules modulate animal lifespan [11,17,18].
Molecular and genetic studies indicate that an endocrine network governs dauer formation and longevity. These experiments suggest a model whereby environmental signals indicating favorable conditions are integrated via neurosensory processing, which stimulate IIS and TGF-b signaling. These pathways converge to activate DAF-12 by promoting DA biosynthesis, thus facilitating growth to reproductive maturity and a normal lifespan. Conversely, in unfavorable conditions, upstream endocrine pathways are downregulated and are thought to decrease DA levels. In the absence of its ligands, DAF-12 associates with the corepressor DIN-1/SHARP and promotes entry into the long-lived dauer stage [12,[19][20][21]. Thus, DA availability regulates DAF-12 activity, controlling the binary dauer decision and lifespan, but how DA availability is achieved is not well understood. In addition, DAF-12 and DA signaling play a conserved role in regulating nematode dauer formation across wide evolutionary distances, including in parasitic nematodes whose infective stages are analogous to dauer [22,23]. Treatment with DA promotes exit from diapause in these animals, suggesting that components of DA biosynthesis and DAF-12 signaling are potential anti-helminthic therapeutic targets.
Although DA availability controls DAF-12 activity to influence longevity, the synthesis and regulation of these molecules are poorly understood. The DAs are derived from dietary cholesterol, which is required for nematode viability and fertility. Previous studies identified DAF-36, a Rieske-like oxygenase with cholesterol 7-desaturase activity; DAF-9, a cytochrome P450 with activity similar to mammalian CYP27A1; and HSD-1, a putative 3-bhydroxysteroid dehydrogenase homolog as acting in the biosynthesis of DAs from dietary cholesterol [11,18,[24][25][26][27][28][29]. However, the nodes of regulation by upstream signaling pathways, environmental and nutritional signals, as well as the functional relationship of identified gene products, intermediary metabolites, different ligands, and the extent, structure, and biochemistry of the pathway are unclear. Moreover it is unknown whether these sterols have other physiologic functions. Such knowledge could provide novel targets for manipulation of longevity pathways or for combating parasitic diseases. Additionally, understanding the metabolism of cholesterol to bile acid synthetic pathways may yield important insights into cardiovascular disease and obesity. Here we elucidate new components of DA synthesis, including a novel conserved 3-hydroxysteroid dehydrogenase, which plays a key role in the control of reproductive development and longevity. Our findings suggest remarkable conservation of bile acid synthetic pathways, which may have implications for the physiologic role of cholesterol and bile acid homeostasis in higher organisms.

Genetic Screens Identify New Components of Dafachronic Acid Biosynthesis
Hormone biosynthetic mutants that reduce production of the dafachronic acids (DAs) have a characteristic phenotypic profile. Partial reduction of the pathway, as seen in daf-9(rh50) hypomorphs, results in gonadal migration defects (Mig), in which gonadal distal tip cells fail their scheduled turns and migrate instead into head and tail along the ventral body wall [25]. Strong reduction of the pathway, as seen in daf-9(dh6) null mutants, results in 100% penetrant constitutive dauer entry (Daf-c) at all temperatures, with the formation of ''partial dauer'' larvae, which have the characteristic dauer cuticle but show incomplete radial constriction of the pharynx and body. By contrast, daf-36(k114) null mutants display these phenotypes with partial penetrance and only in combination with additional stresses: animals are Mig upon cholesterol deprivation and Daf-c at the elevated temperature of 27uC, phenotypes that are rescued by DA supplementation [24]. These results suggest that DAF-36 may work in a branched pathway in concert with other unknown activities, which ultimately converge on DAF-9 for DA production.
To identify new activities in DA biosynthesis, we conducted genome-wide RNAi screens looking for enhancers of daf-36(k114) at the normally permissive temperature of 25uC. We identified several loci that in combination with daf-36 gave Mig and Daf-c phenotypes and whose molecular identity suggested a role in DA biosynthesis. As expected, RNAi against daf-9 enhanced daf-36 mutant phenotypes ( Figure 1A). We also identified ncr-1, a worm homolog of the Niemann-Pick C1-like proteins implicated in sterol transport that was previously noted to interact with daf-36 (unpublished data) [24,30]. Genes with potentially novel roles in DA synthesis included dhs-16, a short-chain dehydrogenase/ reductase (SDR), and emb-8, a NADPH-Cytochrome P450 oxidoreductase. SDRs comprise a large superfamily that typically carry out oxidation/reduction reactions on a variety of substrates, including sterols, xenobiotics, retinoids, and fatty acids [31]. Of the 84 SDRs present in C. elegans, the two closest relatives of DHS-16 are DHS-2 and DHS-20, with ,37% identity in protein sequences. Nematode orthologs include one in the parasitic nematode Ascaris suum, which displays ,40% identity ( Figure  S1). In humans, the closest relatives include SDR9C7/SDR-O and HSD17B6, which display 40% and 38% identity to DHS-16, respectively ( Figure S1). HSD17B6 is involved primarily in androgen metabolism, whereas SDR9C7/SDR-O is expressed almost exclusively in the liver of mice and humans and reportedly metabolizes retinoids [32,33].
Cytochrome P450 oxidoreductases are obligate co-factors for many CYP450 oxygenases, catalyzing electron transfer from NADPH/NADH to CYP450 enzymes, including those involved in sterol and bile acid synthesis. In humans, NADPH-CYP450 oxidoreductase is required for the activity of most microsomal CYP450 enzymes [34]. The single ortholog found in C. elegans, emb-8 (46% identity), is essential for early embryonic development [35]. EMB-8 is required for CYP450-mediated enzymatic activities in C. elegans [36,37] and presumably serves as a co-factor for DAF-9/CYP450 in DA production.

Author Summary
Although well known for their role in the absorption of dietary fat, bile acids have emerged as important metabolic signaling molecules that regulate cholesterol, fat, and glucose metabolism. Bile acids work through nuclear receptors, a class of transcription factors that bind to fat soluble hormones to directly control target gene expression. In the roundworm C. elegans, DAF-12 is a nuclear receptor for bile acids, called the dafachronic acids, which are known to regulate development and longevity, however the synthesis and regulation of these molecules remain unclear. Here we identify novel biochemical activities, including a conserved 3-hydroxysteroid dehydrogenase, involved in the production of the dafachronic acids, illuminating their role in cholesterol and bile acid metabolism, and longevity. The identified activities reveal remarkable evolutionary conservation to those seen in mammalian bile acid synthesis, potentially providing novel ways to manipulate animal lifespan and cholesterol homeostasis. well as daf-9(k182) hypomorphs (Table 1). Comparable enhancement was also seen with mutants of ncr-1/Niemann-Pick C1-like protein. Phenotypic enhancement of daf-36 null animals suggests that DHS-16 works in a branched rather than linear biosynthetic pathway for DA biosynthesis. Similar to dhs-16, knockdown of emb-8 by RNAi resulted in Daf-c phenotypes at 27uC and strongly enhanced the Mig defects of daf-9(k182) hypomorphs, daf-36(k114) nulls, as well as dhs-16(tm1890) deletion mutants in a daf-12dependent manner ( Figure 1E, Table S1), consistent with a role in DA production.

Genetic Epistasis Experiments Place dhs-16 at a Position Consistent with a Role in DA Production
Genetic and molecular experiments reveal that downregulation of insulin/IGF-1 and TGF-b signaling triggers dauer formation by stimulating their respective transcriptional outputs daf-16/FOXO, daf-3/SMAD, and daf-5/SNO-SKI [12]. In addition, significant crosstalk occurs between these pathways. Ultimately, IIS and TGF-b signaling converge upon DA production/DAF-12 activity and are thought to downregulate DA synthesis and promote assembly of a DAF-12/DIN-1 repressor complex that specifies dauer. Loci involved in DA biosynthesis work downstream of IIS and TGF-b signaling components but upstream of DAF-12/NHR, with respect to dauer formation [25,26]. DA can rescue the Daf-c phenotypes of daf-2/InsR, daf-7/TGF-b, and daf-9/CYP450 mutants, and all DA-related phenotypes show DAF-12 depen-dence [11]. To determine where DHS-16 acts in these pathways, we conducted epistasis experiments with dhs-16(tm1890) Daf-c mutants, constructing strains with the dauer defective (Daf-d) loci daf-12/NHR, daf-16/FOXO, and daf-5/SNO-SKI. Double mutants were then analyzed for whether Daf-c or Daf-d phenotypes prevailed. Whereas daf-12 completely suppressed dhs-16 Daf-c phenotypes, daf-16 and daf-5 did not ( Figure 1F, Table 1). Similarly, Daf-c phenotypes of emb-8 RNAi induced at 27uC were suppressed by daf-12 but not by daf-16 null mutations, placing it at a similar position in the pathway, although unexpectedly daf-5 mutants formed dauers under RNAi culture conditions ( Figure 1G, Table S1). These results place DHS-16, and possibly EMB-8, downstream of DAF-16/FOXO and DAF-5/SKI but upstream of DAF-12/NHR, consistent with a role in DA biosynthesis. Accordingly, expression levels of the let-7 related microRNA mir-241, a direct target gene responsive to DA and DAF-12 in developmental timing events, were down in dhs-16 mutants ( Figure 1H).

Sterol Supplementation Experiments Position DHS-16 in the DA Biosynthetic Pathway
The nature and intermediates of the DA biosynthetic pathway are not well understood and elements of a pathway have been assembled from fragmentary evidence. Major identified sterols metabolized from cholesterol include 7-dehydrocholesterol and lathosterol, but their function and the enzymes that produce them are unclear. Recently we showed that DAF-36/Rieske oxygenase is involved in the conversion of cholesterol to 7-dehydrocholesterol, revealing this metabolite as a precursor in DA biosynthesis [29]. Best understood is DAF-9/CYP450, with an established activity as a sterol-27-hydroxylase that successively oxidizes the side chain of the 3-keto steroids, lathosterone and 4-cholesten-3one, to the carboxylic acid moieties of D 7 -DA and D 4 -DA, respectively [11]. We used these and related sterols to a priori build a pathway model, using metabolic genetic and biochemical approaches.
To place dhs-16 at a specific step in DA biosynthesis, we first performed sterol supplementation experiments by feeding loss-offunction mutants proposed DA precursors, including cholesterol, lathosterol, lathosterone, and 4-cholesten-3-one, and looked for rescue of Daf-c phenotypes at 27uC. Compounds working downstream or parallel to the synthetic block should rescue mutant phenotypes, while those working upstream should not. Consistent with working at the last step, the daf-9(k182) hypomorphic mutant was fully rescued only by the DAs (Figure 2A), while the Daf-c ligand-insensitive daf-12(rh273) mutant was mostly unaffected by provision of DAs ( Figure S3A). In accord with a role in DA production, dhs-16 mutants were rescued by D 4 -DA and D 7 -DA ( Figure 2B).
Strikingly, dhs-16 mutants were also rescued by the 3-keto steroid lathosterone, the proposed precursor of D 7 -DA, but not by the 3b-hydroxyl steroid lathosterol, nor by cholesterol and 7dehydrocholesterol ( Figure 2B). Similarly, rescue was seen with the 3-keto steroid 4-cholesten-3-one, the proposed precursor of D 4 -DA, but not by 4-cholesten-3b-ol. These results indicate that the 3alcohol sterols (cholesterol, 7-dehydrocholesterol, lathosterol, and 4-cholesten-3b-ol) lie upstream of the block, whereas the 3-keto steroids (lathosterone, 4-cholestene-3-one, and the DAs) lie downstream. Furthermore, lathosterone rescued animals at nanomolar concentrations more efficiently than 4-cholesten-3one ( Figure S3C). This may be due to the 5-fold lower activity of D 4 -DA relative to D 7 -DA, rendering precursors of D 4 -DA less able to rescue [38]. Alternatively, DHS-16 may act with greater specificity in the D 7 -DA branch. Also consistent with a role downstream of daf-36, the Daf-c phenotype of dhs-16;daf-36 double mutants was rescued by provision of lathosterone and the dafachronic acids in a manner similar to dhs-16 single mutants ( Figure S3C-E). Given the structures of these molecules, the simplest hypothesis is that DHS-16 converts 3-hydroxy steroids to 3-keto steroids-that is, it functions as a 3b-hydroxysteroid dehydrogenase in the production of the DAs. If so, DHS-16 is surprisingly not orthologous in sequence to known 3b-hydroxysteroid dehydrogenase enzymes, thereby identifying a novel function for a conserved orphan dehydrogenase.
We also examined rescue of emb-8 RNAi Daf-c phenotypes by proposed DA precursors and found that rescue was achieved only by provision of the DAs and not by lathosterone or 4-cholesten-3one ( Figure 2C), suggesting emb-8 functions at the same step as daf-9, consistent with a role as a CYP450 reductase.

DHS-16 Has Lathosterol 3b-Dehydrogenase Activity
The feeding experiments described above indicated that DHS-16 may act in the production of lathosterone and possibly 4cholesten-3-one. To further investigate this hypothesis, synchronized wild-type N2 and dhs-16 mutant animals were cultured in liquid media en masse and harvested at the L3 stage. Lipid extracts were obtained from these animals and analyzed by LC/MS/MS for changes in proposed DA precursors. Consistent with predictions based on feeding experiments, dhs-16 mutants had 3.5-fold less lathosterone compared to wild-type animals (p,0.0001), indicating that dhs-16 is required for production of lathosterone ( Figure 2D-E).
Similar results were also obtained by growing worms on solid media and analyzing lipid extracts by GC/MS/MS: dhs-16 mutants had levels of the putative product lathosterone that were below the limit of detection, as well as significantly decreased levels of the downstream product D 7 -DA compared to wild-type (1.5 versus 4.5 pg/mg protein, p,0.05) ( Figure 2F). The residual D 7 -DA seen in dhs-16 mutants suggests that other pathways exist for its production. D 4 -DA was below the detection limit for both WT and mutant extracts (unpublished data). Surprisingly, levels of the putative precursor lathosterol did not accumulate, nor did cholesterol, 7-dehydrocholesterol, or 4-cholesten-3-one ( Figure 2F). These results suggest the pathway is not strictly linear and that lathosterol may be metabolized through other pathways. Interestingly, dhs-16 mutants also showed significantly decreased levels of 4-methyl sterols (consisting of lophenol and 4-methyl-5acholest-8(14)-en-3b-ol), suggesting a role in their production ( Figure 2F). 4-methyl derivatives are nematode-specific modifications catalyzed by STRM-1 methyltransferase. STRM-1 is proposed to irreversibly shunt sterols, such as lathosterone and 4cholestene-3-one, away from DA production and thereby regulate dauer formation [39]. Taken together, these results reveal that DHS-16 is required for the normal production of lathosterone, which subsequently impacts levels of D 7 -DA, and that earlier proposed intermediates may have complex branch points or tissue-specific changes.
In order to directly test whether DHS-16 could convert 3hydroxysteroids to 3-ketosteroids, we expressed FLAG-tagged DHS-16 in HEK293T cells and incubated microsomes isolated from these cells with putative substrates. LC/MS/MS analysis of lipid extracts from these incubations revealed a dramatic increase in lathosterone concentration when DHS-16(+) microsomes were incubated with the substrate lathosterol compared to controls (p,0.005) ( Figure 3A-B). In addition, DHS-16(+) microsomes were also capable of producing 4-cholesten-3-one when incubated with cholesterol, as levels of this compound were produced at significantly higher concentrations compared to controls (p,0.05) ( Figure 3C-D). When extracts from these incubations were fed to dhs-16 mutants, only incubations with lathosterol were able to rescue the Daf-c phenotypes of these animals at 27uC, whereas extracts from incubations with the vehicle control ethanol or cholesterol could not, consistent with the decreased ability of 4cholesten-3-one to rescue at low concentrations ( Figure 3E). In sum, these biochemical data demonstrate that DHS-16 has activity similar to 3b-hydroxysteroid dehydrogenases in the conversion of 3-alcohol to 3-keto steroids.
Although DHS-16 is capable of producing of 4-cholesten-3-one in vitro, analysis of worm extracts suggested that it is not required for this activity ( Figure 2F, Figure 3F). Another enzyme, the canonical 3b-hydroxysteroid dehydrogenase HSD-1, has been postulated to carry out this reaction [27], and therefore could work in parallel to DHS-16. Accordingly, dhs-16 mutants synergized strongly with hsd-1 (61% Daf-c, 25uC), more so than with daf-36 (30% Daf-c, 25uC), despite the fact that dhs-16 and hsd-1 single mutants had weaker phenotypes than daf-36 (Table 1). Sterol analysis of hsd-1 single mutants, however, revealed no significant reduction in levels of 4-cholesten-3-one and D 7 -DA at 20uC ( Figure 3F, Figure S4A). In addition, at the elevated temperature of 25uC, these animals displayed elevated levels of D 7 -DA relative to wild-type ( Figure S4B). Thus, while HSD-1 may act in a parallel pathway, it is not required for 4-cholesten-3-one nor D 7 -DA production. Moreover, the sterol rescue profile of dhs-16;hsd-1 double mutants resembled that of dhs-16 alone, exhibiting rescue of Daf-c phenotypes at 27uC by excess lathosterone, 4-cholesten-3one, and D 7 -DA ( Figure S3F). Also similar to dhs-16 single mutants, dhs-16;hsd-1 double mutants were deficient in lathosterone. At 20uC, double mutants showed no decrease in 4-cholesten-3-one levels, and surprisingly no change in D 7 -DA levels, suggesting that hsd-1 reverses the D 7 -DA deficiency seen in dhs-16 single mutants at this temperature. By contrast, at 25uC, D 7 -DA levels were lower in double mutants ( Figure S4B), consistent with  , and dhs-16;hsd-1 double mutants reveals that hsd-1 is not required for 4-cholesten-3-one production, as previously proposed, suggesting HSD-1 may act in an alternative parallel pathway. In addition, although the dhs-16;hsd-1 double mutants did not contain measurable lathosterone levels, D 7 -DA levels were not reduced, suggesting that lathosterone is not required for its production and that alternate pathways must exist that are independent of dhs-16 (N$6, M 6 SEM; **below detection limit, *p,0.05). doi:10.1371/journal.pbio.1001305.g003 the stronger phenotypes of these animals at elevated temperatures. Taken together these observations suggest that other pathways must exist for D 7 -DA production independent of lathosterone and dhs-16 and that additional dafachronic acid ligands or inhibitors may also exist, which may interact with complex regulation.

Hypodermal DAF-9 Expression Reflects Decreased Dafachronic Acid Levels in dhs-16 Mutants
To further elucidate the activity of DHS-16 in the DA pathway we examined its behavior in hormonal feedback loops. Previous studies demonstrated that daf-9 expression in the hypodermis is an important point of control. Hypodermal daf-9::gfp exhibits low levels of expression in wild-type animals under normal growth conditions, but is shut off in response to stressful dauer-inducing conditions during the L2 stage [40,41]. By contrast, it is upregulated in response to weak stress, including elevated temperatures (25uC), reduced food or cholesterol, and modest exposure to dauer pheromone, in animals undergoing reproductive growth. This upregulation is also seen in normal growth conditions in mutants that partly diminish IIS, TGF-b signaling, and DA biosynthesis, suggesting these pathways convey environmental stress and converge on regulation of daf-9. Hypodermal daf-9::gfp shows strict dependence on daf-12, as expression is absent in daf-12 null mutants but enhanced in ligand insensitive mutants, suggesting both positive regulation and negative feedback control by DAF-12 [40,41].
We wondered whether the environmental and genetic effects on hypodermal daf-9::gfp expression ultimately reflect DA/DAF-12 signaling. If so, then hypodermal daf-9 expression should be responsive to DA. We found that under conditions of mild thermal stress (25uC, 27uC) upregulation of hypodermal daf-9::gfp in wildtype animals was reversed by excess D 4 -or D 7 -DA ( Figure 4A, Figure S5), providing direct evidence that the effect of temperature on daf-9::gfp expression reflects DA availability. We next examined the regulatory behavior in response to perturbations in DA biosynthesis. When grown at 27uC, dhs-16 mutants either entered the dauer diapause and shut off hypodermal daf-9::gfp ( Figure 4B) or bypassed dauer and massively upregulated expression ( Figure  S5). When grown in normal 20uC growth conditions, dhs-16 mutants stimulated high expression of hypodermal daf-9::gfp relative to wild type ( Figure 4C). The high expression of daf-9::gfp is consistent with reduced DA levels, since its expression was strikingly reduced to wild-type levels upon feeding lathosterone as well as the DAs, but not lathosterol ( Figure 4D, Figure S5). By contrast, 4-cholesten-3-one failed to rescue the upregulation seen in dhs-16 mutants, again suggesting that dhs-16 function may be more crucial to the D 7 -DA branch ( Figure S5). Finally, we found that daf-9::gfp upregulation seen in daf-7/TGF-b and daf-2/InsR mutants at 15uC and 20uC was also reversed by D 7 -DA ( Figure 4E-F). Altogether these data suggest that IIS, TGF-b, and hormone biosynthesis normally promote DA signaling, which feeds back on hypodermal daf-9 (see Discussion for model).

DHS-16 Is Expressed in Multiple Endocrine Tissues and Is Regulated by Insulin/IGF-I Signaling
To gain insight into DA producing endocrine tissues and regulation, we established transgenic lines carrying dhs-16::gfp extrachromosomal arrays and analyzed the expression pattern. These C-terminal gfp fusions were functional as measured by efficient rescue of both Daf-c and Mig phenotypes of dhs-16(tm1890) mutants (Table 1). dhs-16::gfp was strongly expressed in the hypodermis, and thus partially overlaps with daf-9 expression. It was also expressed robustly in the posterior pharyngeal bulb, which is involved in feeding/pumping behavior and concentrates sterols [42], as well as a handful of unidentified head and tail neurons ( Figure 5A)-that is, in cells not overlapping with daf-9 [40]. When crossed into the background of daf-2/InsR mutants, we observed a 2-fold upregulation of dhs-16::gfp in the hypodermis of L3 stage animals as well as in dauer larvae (p,0.05) ( Figure 5B, unpublished data). By contrast, no significant changes were observed in the absence of daf-16/FOXO, daf-12/NHR, or daf-7/TGF-b. Similar upregulation was seen in the absence of daf-36/Rieske oxygenase ( Figure 5B), consistent with a role downstream of daf-36 in DA biosynthesis and responsive to perturbations in the pathway.

dhs-16 Is Partially Required for Longevity in the Absence of the Germline
In the gonadal longevity pathway, animals lacking germline stem cells live 50%-60% longer than animals with an intact gonad [15,43]. Increased lifespan is not due to sterility as animals lacking both somatic gonad and germline have normal lifespans. It is thought that gonadal signals act in an opposing manner, with the germline producing lifespan-shortening and somatic gonad producing lifespan-extending signals. The gonadal longevity pathway is dependent upon daf-16/FOXO and daf-12/NHR, as well as the hormone biosynthetic genes daf-36/Rieske oxygenase and daf-9/CYP450 [15,25,44]. Ablation of the germline precursors by laser microsurgery leads to an increased lifespan in wild-type animals, which is no longer seen in these mutant backgrounds. Recently, lifespan-extending signals from the somatic gonad were suggested to be the DAs themselves, because daf-9 is expressed in the spermatheca and the short lifespan of somatic gonad-ablated animals is restored by feeding DAs [45]. Given the importance of DAs in this process, we asked whether dhs-16 also functions in the gonadal longevity pathway. We found that the lifespan of dhs-16 null mutants after removal of the germline precursors by laser microsurgery was significantly reduced (mean = 1963 d) compared to that of wild-type (mean = 4163 d; p,0.0001) ( Figure 6A, Table S2). The failure of a complete suppression of longevity in these animals may be due to residual DA production. Interestingly, aging populations of germline-ablated dhs-16 mutant animals showed early short-lived but later longlived mortality trajectories, with an inflection point around days 10-12 seen in both independent experiments, suggesting two subpopulations or distinct temporal events.
A molecular correlate of gonadal longevity is the localization of GFP-tagged DAF-16 to intestinal nuclei, dependent upon daf-12, daf-36, and daf-9 [17,46,47]. We performed laser microsurgery on animals carrying an integrated daf-16::gfp and found that DAF-16 nuclear localization was much weaker in dhs-16 mutants compared to wild-type (p,0.05) ( Figure 6B-C). As seen in the lifespan studies, we also noted two distinct populations in DAF-16 localization, weakly versus strongly nuclear-localized. However, no correlation was found between strong localization and longevity (unpublished data). Thus, DHS-16 functions in the gonadal longevity pathway, presumably due to a requirement for maximal DA production to promote long life in the absence of the germline.
To see if dhs-16 phenotypes in the gonadal pathway stem from a deficiency in DA production, we asked whether DAF-16::GFP localization defects could be rescued by lathosterone and DA. To do this, we used a genetic model of germline ablation glp-1(e2141ts) and measured the ratio of nuclear to cytoplasmic DAF-16::GFP expression in individual intestinal cells. As seen in the laser ablation experiments, glp-1 induced DAF-16::GFP nuclear localization in the intestine during the first day of adulthood, which was significantly reduced in the dhs-16;glp-1 double mutant background (p,0.001) ( Figure 6D). Similar results were obtained upon treatment of glp-1 animals with dhs-16 RNAi ( Figure S6). Supplementation with DA or lathosterone, but not lathosterol, restored DAF-16::GFP nuclear localization in both cases. These results suggest that DHS-16 normally promotes DA signaling, which stimulates DAF-16/FOXO nuclear localization in animals lacking the germline. Thus, dhs-16 contributes to gonadal longevity by linking DA production to DAF-12 activity and nuclear localization of DAF-16/FOXO. In response to mild stress (e.g., growth at 25uC), hypodermal daf-9 is upregulated in N2 wild-type animals (WT) carrying an integrated dhIs64(daf-9::gfp) array (shown at left). This is rescued upon growth with 33 mM D 7 -DA (shown at right). Arrowheads indicate the XXX R/L neuroendocrine cells in which daf-9 expression is relatively unchanged, and arrows indicate hypodermal expression. The expression levels are displayed quantitatively to the right of each image, as the percentage of animals observed with strong (green), weak (yellow), or no (red) hypodermal GFP expression (N$3, M 6 SD; ***p,0.0001). (B) Under stressful growth conditions at 27uC N2 wild-type animals still undergo reproductive development but have high hypodermal daf-9::gfp expression (left), whereas dhs-16 mutants mostly develop as dauer larvae and shut off hypodermal daf-9::gfp expression (right) (**p,0.01). (C) Under normal growth conditions at 20uC, hypodermal daf-9 upregulation is low in wild-type (left), whereas upregulation is seen in the dhs-16(tm1890) mutant background (right), suggesting daf-9 upregulation in response to DA deficiency (***p,0.0001). (D) Upregulation of hypodermal daf-9::gfp seen in dhs-16 mutants is not rescued by provision of 33 mM of the upstream DA precursor lathosterol (left) but is rescued by the downstream product lathosterone (right) (**p,0.01). (E) Upregulation of hypodermal daf-9::gfp seen in reproductively growing daf-7(e1372)/TGF-b mutants (left) is rescued by DA (right) (**p,0.01). (F) Upregulation of hypodermal daf-9::gfp seen in reproductively growing daf-2/InsR mutants (left) is also rescued by DA (right) (***p,0.0001). doi:10.1371/journal.pbio.1001305.g004

Discussion
Endogenous small molecule metabolites that regulate animal longevity are emerging as a novel approach to influence health and life span. The dafachronic acids are an important class of bile acid metabolites that regulate C. elegans longevity through DAF-12 nuclear receptor signal transduction. DA signaling has also been implicated in the control of nematode parasitism. Here we have identified new components involved in DA biosynthesis, including DHS-16, a short chain dehydrogenase, and EMB-8, a cytochrome P450 reductase, and deduced the biochemical activity of DHS-16 to be a novel 3b-hydroxysteroid dehydrogenase. By controlling ligand availability, DHS-16 regulates DAF-12 activity and thereby directly influences the life plan and life span of C. elegans. Importantly, identified intermediates in bile acid metabolism may also provide unique ways to impact both animal longevity and nematode pathogenesis.
Several lines of evidence reveal that DHS-16 is a DA hormone biosynthetic gene. First, mutant animals show the phenotypic profile of other DA pathway mutants, including Daf-c and gonadal Mig phenotypes, perturbation in daf-9 feedback regulation, and abrogation of life span extension when the germline is absent. Second, mutants have a similar pattern of genetic epistasis, working downstream of insulin/IGF and TGF-b signaling but upstream of daf-12, and enhancing phenotypes of other mutations in DA biosynthetic genes. Third, mutant phenotypes are rescued by DA as well as specific precursors in the biosynthetic pathway. Fourth, mutants have decreased expression of a DAF-12/DA dependent target gene mir-241. Partial loss of emb-8 function also shares a similar spectrum of larval phenotypes.
Our data support a role of DHS-16 in the 3b-dehydrogenation of lathosterol to lathosterone, which is required for maximal biosynthesis of D 7 -DA. Feeding experiments predict a defect in the conversion of lathosterol to lathosterone, suggesting a function in the metabolism of 3-alcohol to 3-keto steroids. In concert with this, sterol profiles reveal that mutants are deficient in the putative product lathosterone as measured by mass spectrometric methods. Although dhs-16 mutants do not accumulate the putative precursor lathosterol, this outcome might be expected if lathosterol can be metabolized through other pathways. Also because absolute levels of lathosterol are an order of magnitude greater than lathosterone, perturbation of this step might have a negligible effect. Importantly, microsomes expressing DHS-16 are able to convert lathosterol to lathosterone, critical in vitro evidence that DHS-16 possesses or supports this biochemical activity. DHS-16 is also capable of producing 4-cholesten-3-one from cholesterol in vitro, although dhs-16 mutants still contain normal levels of this sterol. It seems likely, however, that DHS-16 has other substrates and activities. In the future, it should be interesting to determine the activities of the parasitic nematode DHS-16 orthologs, as they could be important therapeutic targets for treating such pathogens.
By elucidating metabolites and associated biochemical activities, the overall architecture of the C. elegans DA biosynthetic pathway is emerging, revealing not only key changes to previously proposed models, but also that critical biochemical aspects of metazoan bile acid metabolism are remarkably conserved ( Figure 7A). The first step in the D 7 -DA branch, the conversion of cholesterol to 7dehydrocholesterol, is carried out by the DAF-36/Rieske oxygenase [24,29]. A similar activity is proposed for the Drosophila homolog, neverland, in ecdysteroid biosynthesis [48,49]; thus, elucidation of early steps in the nematode pathway could illuminate other unsolved analogous early steps in ecdysteroid biosynthesis. Although mammals lack an obvious homolog of this Rieske oxygenase, the first step in mammalian bile acid biosynthesis involves 7-hydroxylation of cholesterol by CYP7A1 [24,50]. Speculatively, this may indicate that chemical modification at the 7-position of the sterol nucleus is important for partitioning cholesterol towards bile acid synthesis. 7-dehydrocholesterol may then be converted to lathosterol by an as yet unidentified 5a-reductase, as in mammalian pathways. Although C. elegans harbors multiple 5a-reductase homologs, RNAi knockdown did not display phenotypes typical of DA deficiency (Wollam and Antebi, unpublished), possibly because of redundancy. DHS-16 is implicated as a 3b-hydroxysteroid dehydrogenase in the conversion of lathosterol to lathosterone. A similar activity, albeit on different substrates, is ascribed to the mammalian 3b-HSDs (e.g., HSD3B7) [50], yet DHS-16 is not orthologous to these enzymes, and uniquely may not involve obligate D 4 /D 5 -isomerization for activity. We suggest that the discovery of DHS-16 activity in worms might inform mammalian biology, and we speculate that the liver-expressed SDR-O/SDR9C7 could have an analogous activity in mammalian bile acid synthetic pathways. Although dhs-16 affects D 7 -DA levels, additional dhs-16 and lathosterone-independent mechanisms for the synthesis of D 7 -DA appear to exist. Next, a critical step in bile acid synthesis entails the successive oxidation of the cholesterol side chain to the 26carboxylic acid moiety, which is carried out by DAF-9/CYP450 in a manner similar to mammalian CYP27A1. This step is likely facilitated by EMB-8/CYP450 reductase, since the Daf-c phenotypes of emb-8 RNAi map to the same step as daf-9/ CYP450 based on sterol rescue experiments. Finally, our data also suggest that HSD-1 is not required for 4-cholesten-3-one production as previously proposed, or for D 7 -DA production under some conditions, but is likely acting in another pathway. In particular, the unexpected observation that loss of hsd-1 restores D 7 -DA production in dhs-16 mutants at 20uC implies there are additional pathways and ligands with complex regulation that remain to be found. As there are numerous HSD, DHS, and CYP450 enzymes in C. elegans with potential to carry out such reactions, examination of double and triple mutants as well as enzymatic activities will be required to resolve these issues in the future.
Homologs of other genes implicated in mammalian bile acid metabolism are found in the C. elegans genome, including the Dbifunctional protein and Sterol Carrier Protein-x, which function in shortening side chains of mammalian bile acids [50]. The C. elegans counterparts, DHS-28 and DAF-22, have a different known role, promoting synthesis of dauer pheromone by shortening the long chain fatty acid side chains of ascarosides [51]. Intriguingly, dauer pheromone works in a manner opposite to DA to promote dauer formation, but it is unknown whether DAF-22 and DHS-28 also influence C. elegans bile acid metabolism or whether shorter chain bile acids are found in the worm. In conclusion, it is striking that many of the activities in the nematode pathway are biochemically analogous to those found in mammals, yet the enzymes are not always strict structural orthologs, suggesting potential convergent evolution ( Figure 7A). Speculatively, this similarity might have been exploited by parasitic nematodes to signal entry into the appropriate host.
Although daf-9 and dhs-16 expression partially overlap in the hypodermis, daf-36 is found in non-overlapping tissues including the intestine as well as epidermal seam cells in dauers [24,25]. The fact that the various gene products have distinct expression patterns suggests distributed synthesis of DA, which likely requires transport mechanisms. ABC and Niemann-Pick Type C1-like transporters, which are also implicated in dauer formation (mrp-1, ncr-1,-2), may be involved [30,52]. Distributed synthesis could be deployed to make specific DAF-12 ligands with various activities, to localize activity within particular tissues, or to develop a consensus mechanism enabling each tissue to influence the dauer decision.
Regulatory feedback circuitry is a common feature of both mammalian and C. elegans bile acid-like synthesis, dependent upon FXRa and its homolog DAF-12, respectively. In mammals, feedback regulation converges on CYP7A1, although hints of CYP27A1 regulation are evident [53]. In C. elegans, DA production is regulated primarily through hypodermal daf-9/CYP450 expression [40,41], but other control points, such as dhs-16 and daf-36, are now evident and deserve closer study. Our observations on the regulation of hypodermal daf-9 expression and its dependence on DA suggests a three-state model in which environmental cues work through dauer signaling pathways and DA biosynthesis to dictate regulation of daf-9 expression through daf-12 dependent feedback ( Figure 7B): (i) Harsh stress results in a shutdown of DA production, repression of daf-9 expression, and dauer formation; (ii) mild stress initially results in decreased DA production, provoking compensatory upregulation of daf-9 to maintain hormone levels and reproductive development; and (iii) favorable conditions result in DA excess causing negative feedback on daf-9 expression and reproductive development [40]. dhs-16 regulation is somewhat different. Contrary to predictions from genetic epistasis, dhs-16::gfp is not downregulated by decreased IIS or TGF-b, but surprisingly upregulated by reduced IIS and DA. These observations suggest that other regulatory points in the pathway, such as daf-9, take precedence, or that regulation takes place at the level of DHS-16 activity or metabolite production. Moreover it also implies that dhs-16 may function in negative feedback by IIS, DA, or its metabolites.
These complex feedback circuits impact not only dauer formation and reproductive development but also lifespan at elevated temperatures. In ttx-1 mutants, upregulation of hypodermal daf-9 at 25uC fails and animals are short lived [54]. Short lifespan resembles that seen in daf-9(rh50) hypomorphs, and in both cases lifespan is restored to normal by loss of daf-12, suggesting that sufficient DA production is required to ensure lifespan-promoting activities of DAF-12 at elevated temperatures.
Although the DAs modulate nematode lifespan, whether small molecule bile acid-like metabolites can influence mammalian lifespan remains unknown. Elevated bile acid levels are associated with the long-lived Little mouse, as well as expression of xenobiotic detoxification genes, partially dependent upon FXRa [55]. Treatment of wild-type mice with cholic acid induces a similar xenobiotic detoxification expression profile. These studies suggest that BAs may act through FXRa to induce the xenobiotic response and impact lifespan, possibly as a means of longevity assurance. In worms as well, microarray analyses indicate that genes involved in xenobiotic metabolism are altered in long-lived daf-2/InsR and daf-12(rh273) ligand-binding domain mutants [56,57]. Bile acids reduce serum triglycerides, increase insulin sensitivity, and decrease inflammation in animals fed high-fat diets [58]. Furthermore, polymorphisms in FXR may also be associated with reduced IGF levels and murine longevity [59]. In the future, it will be interesting to see if bile acid signaling plays a conserved role in impacting mammalian longevity.

Genetic Screens
RNAi screens were carried out essentially as described [61]. RNAi clones were plated in 12-well plates, two daf-36(k114) hermaphrodites placed in each well to lay eggs, and grown at 25uC. Mig and Daf phenotypes were scored in L4/young adult animals of the next generation. Screens were carried out on the whole genome. Other genes identified in the screen will be described elsewhere.

Isolation of Microsomes
Isolation of microsomes from HEK293T cells was performed as described [62]. Cells were grown in T-75 flasks and transfected with vector only or FLAG-tagged DHS-16. Expression of DHS-16 was verified by immunoblot, using anti-FLAG antibodies (Sigma). Microsomal fractions were resuspended in 0.1 M KPO 4 buffer, pH 7.4, containing 1 mM EDTA and 20% glycerol, and stored at 280uC.

Microsomal Incubations
Microsomes were thawed on ice and brought to 80 mg/mL in 0.1 M KPO 4 buffer, pH 7.4, containing 1 mM EDTA. Substrates were added at 100 mM in 0.5 mL total volume, pre-incubated for 3 min at 37uC, and reacted with 1 mM NAD for 16 h at 37uC. Reactions were processed by extracting twice with 2 mL MTBE, combining the top layers and drying under nitrogen. 0.5 mg of cholesterol-d 7 was added as an internal standard for extractions.

Rescue Assays
Microsomal extracts from three incubations were combined and resuspended in 50 mL methanol, mixed with 56 concentrated OP50, vacuum dried, resuspended in 100 mL 16 OP50, and plated on 3 cm plates containing 3 mL NGM agar. For rescue, ,100 embryos from a 4-8 h egg laying were transferred onto the bacterial lawn, and scored for dauer arrest at 27uC after 48 h. For rescue experiments with pure steroids, 10 mL compounds in ethanol (or ethanol alone) were mixed with 40 mL 56 concentrated OP50 bacteria and plated. Final concentrations include the total volume of agar (3 mL). Dauer arrest was scored after 60 h at 20uC and after 48 h at 25uC and 27uC.

Nematode Lipid Extracts
Worms were grown on twenty 10 cm NGM agar plates seeded with OP50 bacteria. Gravid adults were bleached and the resulting embryos transferred to liquid culture (S-complete medium supplemented with 1006 concentrated OP50). Two to three successive rounds of growth and lysis were performed. For the final round, worms were grown at 20uC until the L3-L4 stage, harvested, frozen in liquid nitrogen, and stored at 280uC. Thawed worms were homogenized by sonication, and total lipids (plus 1 mg cholesterol-d 7 or CDCA-d 4 /10 7 worms) were extracted with 2:1 chloroform:methanol. The resulting chloroform layer was dried under nitrogen. For GC/MS/MS analysis, growth and lysis of worms was carried out on 10 cm NGM plates.

GC/MS/MS Analysis
Dafachronic acid levels were analyzed by GC/MS/MS on a 7000A Triple Quadrupole GC/MS instrument (Agilent Technologies) equipped with an ESI source and an HP-5ms column. Briefly, lipid extracts were spiked with 5b-cholanic acid as an internal standard, derivitized with trimethylsilyldiazomethane, and analyzed in MRM mode. The following transitions were observed: 5b-cholanic acid (m/z 374.3R264.0) and D 7 -dafachronic acid (m/ z 428.3R229.1). The methods used for analysis of all other compounds will be described elsewhere.

Lifespan Analysis
Adult lifespan assays and gonadal cell ablations were performed as previously described [25]. Day 0 corresponds to the L4 stage. Exploded and egg-laying defective animals were excluded from the analysis. Statistical analyses were performed using the log-rank (Mantel-Cox) method with GraphPad Prism software.

DAF-16::GFP Nuclear Localization Experiments
Strains carrying the integrated muIs109(daf-16P::daf-16::gfp) array were used in germline ablations and placed at 20uC. After 4 d, at day 1 of adulthood, worms were scored for nuclear localization of DAF-16::GFP in intestinal cells under a dissection microscope. Worms with a clearly dotted appearance throughout the body were scored as strongly localized. In glp-1 experiments, worms were kept at the restrictive temperature of 25uC from eggs until scoring at day 1 of adulthood, and two of three experiments were performed blind. Images of the anterior intestinal cells were taken at 406 magnification and the relative expression intensity of individual nuclei versus equal areas of adjacent cytoplasm was determined using Image J software (http://rsbweb.nih.gov/ij/).

qRT-PCR
Real-time quantification for microRNAs by RT-PCR was performed with a protocol modified from a previous report [63]. Briefly, total RNA was purified from L3 stage larvae using TRIzol (Invitrogen) and the miRNeasy kit (Qiagen). TaqMan MicroRNA Reverse Transcription kit (Applied Biosystems) was used to generate cDNA with microRNA-specific primers. qRT-PCR was performed with Power SYBR Green master mix (Applied Biosystems) according to the manufacturer's instructions. Sno-RNA U18 was used as an internal control.

Phylogenetic Tree and Multiple Sequence Alignment
Sequences were retrieved using PSI-BLAST [64] with filter turned on. Reciprocal BLASTs were used to determine orthologous relationships between human, nematode, and fly proteins. HomoloGene [65] was used wherever available. Multiple sequence alignments were done with ClustalX [66] using standard parameters. For tree building, the multiple sequence alignment was trimmed using the GBlocks server [67] with relaxed settings. The tree was calculated using Phyml [68] with standard settings and 500 bootstrap steps. The resulting tree was displayed in Dendroscope [69] and prepared for publication using Adobe Illustrator. The sequence of A. suum DHS-16 was determined by BLAST against the expressed-sequence tag database for A. suum and sequencing cDNA fragments amplified from A. suum adult head cDNA (provided by the laboratory of Angela Mousley, Queen's University Belfast, Northern Ireland, U.K.). The sequence was submitted to NCBI and has received the following GenBank accession number: JF753272.