Intercellular signaling of reproductive development by the C. elegans DAF-9 cytochrome P450

A wide range of metazoan signaling molecules relay information between neighboring cells or throughout the entire organism. One challenge is to understand how target cells interpret and integrate multiple incoming signals and respond accordingly. The nematode C. elegans employs multiple peptide and lipophilic hormones to coordinate differentiation of various tissues upon commitment to the reproductive or the developmentally arrested state. Under favorable growth conditions, C. elegans develops through four larval stages (L1-L4) to form a reproductive adult. However, unfavorable environmental conditions cause the animal to arrest at the dauer larval L3 stage (Riddle and Albert,1997). The decision to develop via the alternative dauer stage requires the integration of multiple sensory inputs such as a small molecule dauer pheromone, food and temperature, and the execution of a remodelling program that affects the morphology and physiology of the animal. Molecular genetic analysis has identified an insulin-like pathway(Kimura et al., 1997; Pierce et al., 2001), a TGFβ-like pathway (Ren et al.,1996; Schackwitz et al.,1996) and a cGMP signaling pathway(Birnby et al., 2000), which are thought to couple sensory signals to the subsequent animal remodelling through transcriptional regulation of target genes. Downregulation of DAF-7 TGFβ ligand and DAF-2 insulin/IGF-I like receptor activity leads to activation of DAF-3 Smad protein and DAF-16 Forkhead transcription factor,respectively (Lin et al.,1997; Ogg et al.,1997; Patterson et al.,1997). In contrast to the detailed knowledge of the DAF-2 and DAF-7 signaling cascades, the mechanism by which these pathways are integrated remains largely unknown.

Genetic mosaic analysis showed that the DAF-2 insulin/IGF-I like receptor and the DAF-4 type II TGFβ receptor control reproductive development in a cell nonautonomous manner (Apfeld and Kenyon, 1998; Inoue and Thomas, 2000; Wolkow et al.,2000). A secondary signal is thought to be responsible for communication between the peptide hormone responsive tissues, such as the nervous system, and the rest of the body. Genetic analysis suggests that daf-9 functions downstream of or in parallel to daf-2 and daf-7 and upstream of daf-12(Gerisch et al., 2001; Jia et al., 2002). daf-9 encodes a cytochrome P450 enzyme, whereas daf-12encodes a nuclear receptor (Antebi et al.,2000; Gerisch et al.,2001; Jia et al.,2002). As cytochrome P450 enzymes mediate steroid hormone synthesis in mammals and Drosophila(Miller, 1988; Warren et al., 2002), and because daf-9 acts upstream of the nuclear receptor gene daf-12 (Gerisch et al.,2001; Jia et al.,2002), DAF-9 may mediate the production of a lipophilic hormone that regulates DAF-12 activity. Based on its action downstream of daf-2 and daf-7 in the genetic epistasis analysis, daf-9 expression or activity may in turn be regulated by the upstream daf-2 and daf-7 signaling pathways.

Two classes of daf-9 mutant alleles have been described. Animals carrying strong loss-of-function alleles arrest as dauers unconditionally and seldom recover (Gerisch et al.,2001; Jia et al.,2002). Ultrastructural studies revealed that daf-9 dauers display intermediate dauer morphology in selected tissues(Albert and Riddle, 1988). This suggests that parallel pathways may operate in conjunction with daf-9to complete the global remodelling in dauer animals. A second class of daf-9 alleles confers weak loss-of-function phenotypes: reversible dauer arrest and a gonadal migration defect(Antebi et al., 1998; Gerisch et al., 2001; Jia et al., 2002). Allele and temperature specific extension of adult life span has also been reported for daf-9 mutant animals (Gerisch et al., 2001; Jia et al.,2002). Furthermore, there are complex interactions between daf-9, germline and daf-2 signaling pathways in the control of adult life span, reminiscent of those reported for daf-12(Larsen et al., 1995; Gems et al., 1998; Hsin and Kenyon, 1999).

The DAF-9 protein sequence is most similar to the mammalian CYP2 family of P450 enzymes, which are responsible for degradation of steroidal and xenobiotic compounds (Nebert and Russell,2002). Nevertheless, DAF-9 is unlikely to be a functional homologue of the mammalian CYP2 enzymes. This is because daf-9expression is not induced by a range of xenobiotic compounds(Menzel et al., 2001), a feature of the mammalian CYP2 enzymes(Waxman, 1999). A biosynthetic role for DAF-9 was supported by the observations that cholesterol deprivation causes a gonadal migration defect, similar to that of hypomorphic daf-9 mutant animals (Gerisch et al., 2001). Furthermore, cholesterol withdrawal inhibits recovery from dauer arrest by daf-9 hypomorphs(Jia et al., 2002). This argues that daf-9 may participate in the modification of cholesterol in the biosynthetic pathways to steroid hormones.

In this paper, we addressed the following questions. Where is the site of action of daf-9 gene function? How does daf-9 interact with the daf-2 and daf-7 signaling pathways? Is daf-9expression regulated at a transcriptional level? We addressed these questions by expressing a functional GFP-tagged DAF-9 protein under the control of the endogenous daf-9 promoter and other well-established tissue-specific promoters. We find that daf-9 directs larval development and gonadal migration in a cell nonautonomous manner and its action is intricately linked to daf-2, daf-7 and daf-12 activities.

Strains used were as follows: wild-type N2 Bristol, daf-7(e1372)III, daf-2(e1370) III, daf-1(m40) IV, + / szT1[lon-2(e678)] I; daf-9(e1406) dpy-7(sc27) / szT1 X, dpy-7(sc27) daf-12(m20) X, daf-12(m20) X, daf-12(m583) X.

mgEx661-662: Ex[daf-9p::daf-9 genomic::GFP], mgEx663-664:Ex[dpy-7p::daf-9 cDNA::GFP; mec-7::GFP], mgEx669-670: Ex[sdf-9p::daf-9 cDNA::GFP; mec-7::GFP], mgEx665-666: Ex[che-2p::daf-9 cDNA::GFP; mec-7::GFP] and mgEx667-668: Ex[col-12p::daf-9 cDNA::GFP;mec 7::GFP].

In all tables, line 1 refers to the odd-numbered extrachromosomal array and line 2 refers to the even-numbered extrachromosomal array.

The genomic region spanning all exons of daf-9 plus 7 kb of non-coding sequence 5′ to the initiator codon of the daf-9b isoform was amplified by PCR and subcloned into the SalI/BglII sites of pPD95.69 (kindly provided by A. Fire) in two steps. The nuclear localisation signal of pPD95.69 was removed as a result. The genomic region spanning the first exon, the first intron and seven residues of the second exon together with 3 kb of non-coding sequence 5′to the initiator codon of the daf-9b isoform was amplified by PCR and subcloned into the BglII site of pPD95.75 (kindly provided by A. Fire). The ligation junctions of the above constructs were sequenced to ensure that the daf-9-coding sequence was in-frame with the GFP coding sequence. The genomic region encompassing all exons of the daf-9 plus 3 kb of non-coding sequence 5′ to the initiator codon of daf-9b isoform was amplified by PCR and subcloned into the BglII/AgeI sites of pPD95.75. Intron 1 of the daf-9b isoform was removed from the last construct by recombinant PCR and the product subcloned into the BglII/AgeI sites of pPD95.75. The nuclear receptor consensus half site in intron 1 of the daf-9b isoform was mutated into a LexA binding site by recombinant PCR and the product subcloned into the BglII/AgeI sites of pPD95.75. For the last three constructs, all exons and introns of daf-9 were fully sequenced.

The above constructs were injected into N2 wild-type animals at 10 to 30 ng/μl. pBluescript was used to normalise the total concentration of injection mix to 100 ng/μl. Extrachromosomal arrays which gave robust daf-9::GFP expression were introduced into + / szT1[lon-2(e678)]; daf-9(e1406) dpy-7(sc27) / szT1 animals by genetic crosses.

Tissue specific daf-9::GFP transgenic constructs were generated by assembling three PCR products by a modified recombinant PCR method(Hobert, 2002). The following tissue specific promoters were amplified from N2 genomic DNA: sdf-9(3.7 kb), che-2 (2.5 kb), dpy-7 (0.4 kb) and col-12(1 kb) (Johnstone and Barry,1996; Fujiwara et al.,1999; Ohkura et al.,2003). The primer sequences defining the 5′ end of the promoter in these transgenes are as follows:

• sdf-9, 5′-tcaaaaatacattatggcgactc-3′;

• che-2, 5′-gtcacacatgaatgagtctcgcc-3′;

• dpy-7, 5′-tcattccacgatttctcgcaac-3′; and

• col-12, 5′-gaaagttcagaactggcatggag-3′.

Each promoter fragment encompasses sequence immediately 5′ to the start codon of the respective gene. GFP-coding sequence and unc-543′ UTR sequence were amplified by PCR using pPD95.75 as template. The coding sequence of the daf-9b isoform was amplified by PCR using a daf-9 cDNA clone as template (kindly provided by Y. Kohara). Purified PCR products were injected into N2 wild-type animals at 5 ng/μl in the presence of mec-7::GFP plasmid at 30 ng/μl. pBluescript was used to normalise the total concentration of injection mix to 100 ng/μl. Extrachromosomal arrays which gave robust daf-9::GFP expression were introduced into + / szT1 [lon-2(e678)]; daf-9(e1406)dpy-7(sc27) / szT1 animals by genetic crosses.

Adults were allowed to lay eggs on nematode-growth plates for 3 hours at room temperature, and progeny were incubated at 20°C for 68 and 92 hours or 25°C for 55 hours. Dauers were distinguished by a radially constricted body, dauer alae and a constricted pharynx. Dauer assays for each strain were repeated at least three times.

Adult lifespan of various strains was determined at 25°C, with agar plates containing 0.1 g/ml FUDR to prevent growth of progeny. Synchronous populations of worms from 3 hour egglays on nematode-growth medium plates were allowed to develop at 15°C until young adult stage, before being transferred to FUDR plates and shifted to 25°C. Worms were monitored every 2-4 days and were scored as dead when they no longer responded to gentle prodding with a platinum wire. Lifespan is defined as the time elapsed from the day when worms were put on FUDR plates (day 0) to when they were scored as dead. Worms that crawled off the plates were excluded from calculations. Lifespan assays were repeated at least twice.

We generated a daf-9::GFP fusion gene that bears the daf-9 5′ regulatory region, the entire daf-9-coding sequence, including introns, and fuses GFP to the last amino acid residue of the DAF-9 protein. This translational fusion gene rescues the strong loss-of-function daf-9(e1406) mutation(Table 1), suggesting that visualization of the fluorescent fusion protein should reveal where the functional gene product acts. daf-9(e1406) animals carrying the fusion gene developed into reproductive adults, in contrast to their non-transgenic siblings that arrest as dauer larvae irreversibly. Consistent with previous reports, daf-9::GFP is expressed in a pair of cells in the anterior ganglion in L1 larvae, which persists in all larval stages and in adults, in the hypodermis from the mid-L2 stage to the end of L4 stage, and in the spermatheca of adult hermaphrodites(Gerisch et al., 2001; Jia et al., 2002). The daf-9::GFP-expressing head cells have been identified as XXXL/R,which are thought to be embryonic hypodermal cells(Ohkura et al., 2003). Nevertheless, we favour the assignment of XXXL/R cells as neuron-like because they possess axon-like projections, and daf-9::GFP in these cells are refractory to feeding RNAi against daf-9 in larvae and adult animals(data not shown), two characteristics of neuronal cells(Fraser et al., 2000; Tavernarakis et al., 2000; Timmons et al., 2001).

daf-9 encodes a cytochrome P450 enzyme and is proposed to synthesize a lipophilic hormone. Given its spatially restricted expression pattern, DAF-9 may fulfill its role in preventing dauer arrest by mediating the maturation or destruction of a hormonal signal that acts in a cell-nonautonomous manner. To address this, we generated a series of promoter fusion genes where daf-9 cDNA expression was driven by tissue-specific heterologous promoters, and tested whether they rescue the dauer constitutive phenotype of daf-9(e1406) mutant animals.

We used the collagen gene dpy-7 promoter to direct hypodermal expression of daf-9 during larval stages(Gilleard et al., 1997). A GFP tag was appended to the DAF-9 C terminus so that tissue specific DAF-9 expression could be confirmed by visualization of the GFP signal. Restoration of daf-9 activity in the hypodermis alone was sufficient to prevent dauer arrest of daf-9(e1406) mutant animals(Table 1); these transgenic animals developed into reproductive adults with the same growth rate as animals carrying a daf-9 transgene driven by its own promoter. Next,we expressed daf-9 exclusively in XXXL/R cells under the control of the sdf-9 gene promoter (Ohkura et al., 2003). At 20°C, daf-9 expression in the two XXXL/R cells was sufficient to prevent dauer arrest of daf-9(e1406)mutant animals (Table 1). Nevertheless, 5-20% of daf-9(e1406) animals that expressed daf-9 in the XXXL/R cells arrested as dauers at 25°C(mgEx667, mgEx668; n=898). This was not due to downregulation of the sdf-9p::daf-9::GFP transgene at 25°C as no dramatic diminution of GFP fluorescence was observed in the XXXL/R cells. By contrast, dpy-7 promoter-driven, daf-9expression in the hypodermis led to complete suppression of dauer arrest of daf-9(e1406) animals at 25°C (mgEx663, mgEx664; n=300). Therefore, hypodermal daf-9 expression or a general increase in daf-9 expression level may be crucial in promoting reproductive development at elevated, dauer-inducing temperatures. Taken together, daf-9 activity in the hypodermis is sufficient to orchestrate reproductive development in an otherwise daf-9 deficient animal. Hence, our observations support the hypothesis that daf-9inhibits dauer arrest in a cell nonautonomous manner.

In contrast to the strong loss-of-function allele daf-9(e1406),daf-9 weak loss of function alleles confer a gonadal migration defect,where the distal tip cells fail to migrate dorsally at the third larval stage(Gerisch et al., 2001; Jia et al., 2002). This indicates that daf-9 also plays a role in gonadal migration during reproductive development. Transgenic expression of daf-9 under the control of its endogenous promoter fully rescued daf-9(e1406) mutant animals and no gonadal migration defect was observed at the L4 or adult stage(Table 2). Similarly, normal gonadal migration was observed in daf-9(e1406) mutant animals when daf-9 was expressed exclusively in the hypodermis or the XXXL/R cells under the control of dpy-7 or sdf-9 promoters, respectively(Fig. 1A and Table 2). Taken together, our results suggest that daf-9 acts in a cell nonautonomous manner to direct the movement of the distal tip cells that drive proper gonadal migration, and that expression either from the broadly distributed hypodermal cells or the anterior XXXL/R cells supplies sufficient signal to do so.

To test whether DAF-9 expression can confer endocrine function to other tissues, we expressed daf-9 in 56 ciliated neurons using the che-2 gene promoter that is inactive in XXXL/R cells(Fujiwara et al., 1999; Ohkura et al., 2003). Ectopic daf-9 activity in the ciliated sensory neurons was sufficient to prevent dauer arrest of the majority of daf-9(e1406) mutant animals(Table 1). Only a small fraction of transgenic animals arrested as partial dauers. This suggests that DAF-9 can indeed function in ciliated sensory neurons and allow production of the putative lipophilic hormone that promotes reproductive development.

Even though expression of daf-9 in the ciliated neurons using the che-2 promoter rescued daf-9(e1406) dauer arrest, more than 70% of the reproductive adults displayed a gonadal migration defect(Fig. 1B and Table 2). Such defect is also observed in animals bearing daf-9 weak loss-of-function alleles. One possibility is that the daf-9-expressing cells and the target distal tip cells are too distant from each other. Notably, of the 56 neurons where the che-2 promoter is active, 49 of them are located in the head,five in the tail and none in the vicinity of the distal tip cells(Fujiwara et al., 1999). However, this is ruled out by the complete suppression of gonadal migration phenotype of daf-9 mutant animals when daf-9 activity was restored only in the two XXXL/R cells in the head(Table 2). We therefore attribute the gonadal migration defect to a suboptimal level of lipophilic hormone that is produced by DAF-9 in ciliated neurons. Alternatively, the complement of enzymes in ciliated sensory neurons may only be able to synthesize a hormone that promote non-dauer fate but not proper gonadal migration, unlike the ones in the native XXXL/R cells or hypodermis.

Genetic analysis suggests that daf-9 functions either downstream of or in parallel to daf-16 and daf-3 in the dauer pathway(Gerisch et al., 2001; Jia et al., 2002). One attractive model is that daf-9 expression is regulated at a transcriptional level by DAF-16 and DAF-3 in response to the daf-2insulin-like and daf-7 TGFβ like signaling pathways,respectively. Modulation of daf-9 gene expression would in turn alter the level of a probable lipophilic secondary hormonal signal that promotes reproductive development. If this is the case, constitutive expression of DAF-9 may substitute for a loss of daf-7 or daf-2 signaling and direct reproductive development in daf-7(–) and daf-2(–) mutant animals, which normally arrest as dauers.

To uncouple daf-9 expression from any potential transcriptional control originating from the daf-7 and daf-2 pathways, we introduced transgenes that direct daf-9 expression in the hypodermis or the XXXL/R cells under the control of dpy-7 or sdf-9promoters, respectively, into daf-7(e1372), daf-1(m40) and daf-2(e1370) mutant animals. The same transgenes were fully functional in rescuing the dauer phenotype and gonadal migration defect of daf-9(e1406) mutant animals.

The daf-7 and daf-1 genes encode a TGFβ like ligand and a type I TGFβ receptor, respectively(Georgi et al., 1990; Ren et al., 1996). Constitutive hypodermal expression of daf-9 suppressed the dauer arrest phenotype of daf-7(e1372) mutant animals(Table 3). At 25°C, none of the transgenic animals arrested at the dauer stage, and the majority of them became gravid adults while all non-transgenic daf-7(e1372) mutant animals arrested as dauers. Similar results were obtained when the same transgenes were introduced into daf-1(m40) animals. By contrast, daf-9 expression in the XXXL/R cells, verified by GFP fluorescence of the DAF-9::GFP fusion protein, was unable to prevent dauer arrest of daf-7(e1372) mutant animals at 25°C(Table 3), although a fraction of the dauers did recover spontaneously upon prolonged incubation. Our results demonstrate that constitutive daf-9 hypodermal, but not XXXL/R,expression can substitute for the loss of daf-7 neuroendocrine signal and bypass the block in TGFβ signaling in target tissues. This argues a major role for daf-9 in transducing the daf-7 signal,perhaps through production of a lipophilic hormone in the hypodermis. Alternatively, daf-9 may act in parallel of daf-7 in promoting reproductive development.

daf-2(e1370) animals carry a missense mutation in the kinase domain of the C. elegans insulin receptor homologue(Kimura et al., 1997). All daf-2(e1370) animals arrest at the dauer stage when grown at 25°C. Expression of daf-9 exclusively in the XXXL/R cells was unable to suppress the dauer arrest phenotype (n=278). However,constitutive expression of daf-9 in the hypodermis, under the control of the dpy-7 promoter, partially suppressed the dauer arrest phenotype of daf-2(e1370) mutant animals(Fig. 2). Although no transgenic animals became reproductive adults, they displayed non-dauer characteristics in a tissue-specific manner. At 25°C, daf-2(e1370) animals complete molting into dauers at ∼52 hours post egg laying. At this time, sporadic pharyngeal pumping persisted in transgenic animals expressing DAF-9 in the hypodermis, indicating that they were not completely transformed into arrested dauers. After 90 hours of development, individual tissues of these transgenic animals and their non-transgenic siblings were examined under high magnification using Nomarski optics. Non-transgenic daf-2(e1370) dauer had a constricted pharynx and a highly refractile intestine because of accumulation of lipids. Moreover,vulval and germline development were arrested and dauer alae were visible on the cuticle. By contrast, daf-2(e1370) transgenic animals overexpressing DAF-9 in the hypodermis, the pharynx was not constricted and no dauer alae were observed. However, the intestine was highly refractile and resembled that of a dauer animal. It was striking to note that although the vulva of some of the transgenic animals displayed L4 morphology (36%, n=76), germline development was arrested and the distal tip cells failed to reflex. Taken together, these data show that daf-9expression in the hypodermis could compensate for the loss of daf-2activity and promote reproductive development of the hypodermis, pharynx and vulva. This implies that daf-9 may act downstream of or in parallel to the daf-2 signaling pathway and specify non-dauer fate in a subset of tissues. However, the lipophilic hormone processed by DAF-9 in the hypodermis is unlikely to overcome the lack of daf-2 signaling in the intestine and germline. Our results are consistent with mosaic analysis which suggests a cell-autonomous role of daf-2 in the reproductive development of P₁-derived germline and gonad(Apfeld and Kenyon, 1998).

In addition to the dauer arrest phenotype, daf-2(e1370) animals have a life span twice as long as wild-type animals(Kenyon et al., 1993; Larsen et al., 1995). We tested the hypothesis that daf-9 may act in the daf-2signaling pathway to control adult life span. Expression of daf-9 in the hypodermis was driven by col-12 or dpy-7 promoters in daf-2(e1370) animals. The col-12 promoter is active in larval stages and in the first 2 days of adulthood, whereas the dpy-7promoter is active only in larval stages(Johnstone and Barry, 1996). Transgenic and non-transgenic daf-2(e1370) animals were allowed to develop at the permissive temperature (15°C) until the late L4 stage, and then shifted to the restrictive temperature (25°C) for determination of adult life span. In four independent trials, daf-2(e1370) animals expressing daf-9 in the hypodermis from the col-12 or dpy-7 promoters (n=351) had a similar adult life span as their non-transgenic siblings (n=204)(Fig. 3 and data not shown). In conclusion, we found no evidence that daf-9 acts downstream of daf-2 in the control of adult life span.

We monitored daf-9 expression in XXXL/R cells, hypodermis and spermatheca, using a functional daf-9::GFP fusion gene that contains∼7 kb of the endogenous daf-9 promoter as well as all exons and introns of daf-9 (construct i, Fig. 4E). This transgene rescued the strong loss of function mutation daf-9(e1406), suggesting that it contains all the sequence elements for proper expression of daf-9. When this transgene was introduced into daf-2(e1370)and daf-7(e1372) mutant animals grown at the non-permissive temperature (25°C), we found that hypodermal expression of daf-9::GFP was absent in dauers, even though the daf-9::GFP expression in XXXL/R persisted. The lack of hypodermal daf-9::GFP expression was also noted in wild-type dauer animals derived from starvation. The daf-9::GFP transgene was unable to suppress dauer arrest of daf-2(e1370) and daf-7(e1372) animals (n>100) at the restrictive temperature (25°C) and enhanced dauer arrest of these animals at the permissive temperature (15°C). At 15°C, daf-2(e1370) and daf-7(e1372) transgenic animals displaying hypodermal daf-9::GFP expression entered the reproductive program without delay, whereas animals lacking such expression arrested as dauer for prolonged periods (>7 days). We speculate that daf-2 and daf-7 signaling pathways may play a role in controlling the hypodermal expression of daf-9, which is crucial in the commitment of reproductive development. In daf-2(e1370) and daf-7(e1372)animals, daf-9 expression driven by its endogenous promoter may be compromised and therefore insufficient to initiate the reproductive program.

daf-9 was placed upstream of daf-12 by genetic epistasis analysis (Gerisch et al.,2001; Jia et al.,2002). As feedback regulation of cytochrome P450 genes by nuclear receptor is well documented (Waxman,1999; Chawla et al.,2001), we wondered if daf-9 could be under the transcriptional control of DAF-12 nuclear receptor. We monitored daf-9 expression using a functional daf-9::GFP fusion gene(construct i, Fig. 4E). When this transgene was introduced into animals carrying the strong loss of function allele daf-12(m20)(Antebi et al., 1998), a dramatic reduction of hypodermal expression of daf-9::GFP was observed at 15°C and 25°C, while expression in the XXXL/R cells and spermatheca persisted (Fig. 4A-D, Table 4, and data not shown). Similar results were obtained in daf-12(m583) mutant animals (data not shown). Hence, DAF-12 activates daf-9 gene expression in the hypodermis. Furthermore, daf-9 and daf-12appear to form a feedback regulatory loop in which DAF-12 is downregulated by an increase in antagonistic ligand production by DAF-9.

What is the response element in the daf-9 promoter at which DAF-12 exerts its transcriptional control? As DAF-12 positively regulates daf-9 expression in the hypodermis, deletion of such an element should result in diminution of daf-9 hypodermal expression in wild-type animals. To this end, a series of transcriptional and translational daf-9::GFP transgenes were constructed(Fig. 4E). The spatial and temporal expression pattern of daf-9 could be fully recapitulated by a transgene containing 3 kb of promoter sequence plus intron 1 of the b isoform of daf-9. However, deletion of intron 1 resulted in a dramatic reduction of hypodermal daf-9::GFP expression, reminiscent of that observed in daf-12(–) animals. We were unable to quantitate the signal as the GFP fluorescence was readily bleached upon excitation. Interestingly, the removal of intron 1 also led to the elimination of daf-9::GFP expression in the spermatheca. We noticed a nuclear receptor consensus half site (TGTTCT) within intron 1 that is also evident in an analogous location in the C. briggsae daf-9 gene. To test whether this sequence represents a DAF-12-binding site, we mutated it from GTGTTCTGT into a LexA-binding site (GTACTGTAT) in the context of a rescuing daf-9::GFP fusion gene that is normally expressed in XXXL/R,hypodermis and spermatheca (Fig. 4E, construct v). Surprisingly, elimination of the nuclear receptor half site led to a loss of daf-9::GFP expression in the spermatheca alone (three extrachromosomal arrays, 25°C, hypodermal expression=90% n=375; spermathecal expression=2%n=155). This suggests that the sequence TGTTCT is critical for spermathecal daf-9 expression. However, daf-9 hypodermal expression is unlikely to be controlled by DAF-12 through binding to the same sequence. Taken together, the 181bp of intron 1 sequence plays a major role in directing daf-9 expression in the hypodermis and spermatheca. Although daf-12, in part, appears to control the hypodermal expression, an unknown factor is likely to bind to a TGTTCT element in daf-9 intron 1 and govern the spermathecal expression.

Genetic analysis shows that daf-9 acts downstream of or in parallel to daf-2 insulin-like and daf-7 TGFβ like signaling pathways to control a developmental decision between reproductive development versus arrest at the dauer diapause stage, as well as metabolism and life span in C. elegans(Gerisch et al., 2001; Jia et al., 2002). Here, we have demonstrated that daf-9 regulates dauer arrest and gonadal migration in a cell nonautonomous manner. This was achieved by overexpressing a DAF-9::GFP fusion protein in specific tissues of genetically daf-9(–) animals. Our results support the hypothesis that DAF-9 cytochrome P450 enzyme acts in the synthesis pathway of a lipophilic ligand. Such a lipophilic ligand may well serve as a secondary signal which mediates,at least in part, daf-2 and daf-7 signaling pathways as overexpression of daf-9 suppresses the dauer arrest phenotype of daf-7(–) animals and allows reproductive development of a subset of tissues in daf-2(–) animals. Finally, we provided evidence that DAF-12 nuclear receptor is engaged in feedback regulation with DAF-9 cytochrome P450 through transcriptional activation of daf-9expression in the hypodermis.

daf-9 acts upstream of daf-12 and antagonizes its activity in the dauer pathway (Gerisch et al., 2001; Jia et al.,2002). Two simple models consistent with the genetic analysis are that daf-9 participates in the synthesis of a daf-12antagonist, or that daf-9 degrades a daf-12 agonist. We favour the former model. First, expression of daf-9 in single tissues is sufficient to direct reproductive development of other tissues, suggesting that daf-9 mediates the synthesis of an endocrine signal. Second, daf-9 expression is restricted to three tissues, while a ubiquitous expression pattern was reported for daf-12(Antebi et al., 2000; Gerisch et al., 2001; Jia et al., 2002). It is conceivable that once the daf-12 ligand is generated in XXXL/R,hypodermis or spermatheca by DAF-9, it would diffuse to target tissues via the pseudocoelom. At this stage, we cannot exclude the possibility that the DAF-12 ligand may be degraded by DAF-9. However, this necessitates transport of the DAF-12 ligand throughout the body to the three daf-9-expressing tissues for destruction. Another model involves co-expression of daf-9, and other P450 enzyme(s) that are genuinely involved in daf-12 ligand synthesis, in the same tissue. The level of agonist produced, and DAF-12 transcriptional potential, would then depend on the relative activity of the competing P450 enzymes. This is similar to cases in Drosophila and mammals, where hormone availability is modulated locally by enzymes, such as cytochrome P450, in target tissues(Luu-The, 2001; Gilbert et al., 2002). The validity of this model awaits identification of additional enzymes that participate in daf-12 ligand metabolism.

The daf-9-expressing head cells have been identified as XXXL/R where the che-2 promoter is inactive(Ohkura et al., 2003). This is intriguing because expression of daf-9 in ciliated sensory neurons using the che-2 promoter was clearly able to suppress dauer arrest of daf-9(–) animals and hence substitute for the loss of daf-9 activity in XXXL/R and other tissues. We obtained similar results when daf-9 expression in ciliated sensory neurons was driven by the osm-6 promoter (H.Y.M. and G.R., unpublished). Nevertheless,ectopic daf-9 expression in mechanosensory neurons using the mec-7 promoter did not suppress the dauer arrest phenotype of daf-9(–) animals (Gerisch and Antebi, 2004). To reconcile the different observations, we propose the following models. As XXXL/R cells are adjacent to the che-2-expressing, but not the mec-7-expressing, neurons(e.g. IL1s), one can imagine intercellular shuttling of lipophilic intermediates of hormone synthesis over a short distance. Alternatively,cytochrome P450 enzymes that normally act upstream and downstream of DAF-9 may be present in the ciliated neurons and given the cholesterol derived substrate is available, daf-9 expression in these neurons may be sufficient to produce the bona fide hormone.

daf-9(+) activity in ciliated sensory neurons was sufficient to rescue the dauer arrest phenotype of daf-9-deficient animals, even though these animals display a gonadal migration defect. This phenocopies animals bearing daf-9 weak loss-of-function alleles(Gerisch et al., 2001; Jia et al., 2002). Perhaps DAF-9 in ciliated sensory neurons can only produce a suboptimal dose of its cognate lipophilic hormone that is nevertheless sufficient to prevent dauer arrest. However, it is known that particular head chemosensory neurons emit key signals to control dauer arrest(Bargmann and Horvitz, 1991; Ren et al., 1996; Schackwitz et al., 1996), and they may be the target cells that respond to a paracrine signal generated by DAF-9. It is conceivable that daf-9 may be involved in the synthesis of an endocrine signal for gonadal migration and a second paracrine signal for reproductive development. Accordingly, DAF-9 may be proximal in a hormone synthesis pathway in which the product of DAF-9 can be further modified by multiple downstream P450 enzymes to yield different lipophilic signaling molecules. This model predicts that the ciliated sensory neurons are unable to produce the endocrine signal that direct proper gonadal migration despite ectopic expression of daf-9, because of a lack of its downstream partners.

daf-9 may act in the pathway for the synthesis of a secondary signal that mediates the cell nonautonomous action of the daf-2insulin/IGF-I receptor and daf-4 TGFβ type II receptor(Apfeld and Kenyon, 1998; Inoue and Thomas, 2000; Wolkow et al., 2000; Gerisch et al., 2001; Jia et al., 2002). In support of this hypothesis, we found that constitutive expression of daf-9 in the hypodermis, under the control of a heterologous promoter, was sufficient to completely suppress the dauer arrest phenotype of daf-7(–)and daf-1(–) animals. The ability to bypass daf-7/TGFβ signaling deficiency by a gain-of-function daf-9 transgene strongly suggests that daf-9 is a major transducer of the daf-7 reproductive signal.

Constitutive expression of daf-9 in the hypodermis allowed partial suppression of the dauer arrest phenotype of daf-2(–) animals. It appeared that the dauer program was never initiated in a subset of tissues,while others were resistant to the excess hormonal signal generated by DAF-9. We postulate that daf-9 may indeed mediate daf-2 signaling in the hypodermis, pharynx and vulva. By contrast, it is unlikely to mediate daf-2(+) activity in the germline and intestine. One possibility is that daf-2 may exert cell-autonomous action on the latter set of tissues and does not normally employ daf-9 to relay its activity. Notably, mosaic analysis suggested that reproductive development of the germline may require additional daf-2(+) activity that acts in a cell-autonomous manner (Apfeld and Kenyon,1998). The partial dauer arrest phenotype displayed by daf-2(e1370) mutant animals overexpressing daf-9 is similar to those observed in daf-2(e1370); daf-12(m20) and daf-2(e1370); daf-12(m583) animals(Larsen et al., 1995). The similarity in the effects of daf-9 overexpression and daf-12severe loss of function, with respect to the highly tissue-specific phenotype of partial daf-2 dauers, again highlights the close functional link between daf-9 and daf-12. Nevertheless, the observation that daf-2(e1370); daf-12(m20) animals lives twice as long as daf-2(e1370) animals provides an exception to this notion(Larsen et al., 1995), as the life span of daf-2(e1370) animals could not be altered by constitutive expression of daf-9 in the hypodermis. It has been reported that the daf-2 pathway acts in adult animals (up to 4 days old) to specify life span (Dillin et al.,2002); however, the activity of our hypodermal daf-9transgene ceases in 2-day-old adults. It may also be possible that daf-2 and daf-9 function independently to control adult life span.

We initially attempted to suppress dauer arrest of daf-2(–)and daf-7(–) animals by expressing daf-9 under the control of its endogenous promoter. Unlike the hypodermis-specific dpy-7 promoter, daf-9 expressed from its own promoter failed to rescue daf-2(–) or daf-7(–) animals at the restrictive temperature. We note that the daf-9 promoter is unable to support daf-9 hypodermal expression in daf-2(–), daf-7(–) or natural dauers derived from starvation, even though daf-9 expression persists in XXXL/R. This correlates well with our observations that sdf-9 driven daf-9 expression in XXXL/R alone is not sufficient to prevent dauer arrest of daf-2(–) and daf-7(–) animals. Taken together, we propose that favorable growth conditions, transduced in part by the daf-2 and daf-7signaling pathways may trigger daf-9 hypodermal expression at mid-L2 stage. Given the mass and coverage of the hypodermis across the entire animal, daf-9 expression in this tissue may induce a dramatic increase in lipophilic hormone production that may be crucial in the initiation or reinforcement of the reproductive program. This is demonstrated by our results that constitutive expression of daf-9 in the hypodermis is sufficient to promote reproductive development in dauer constitutive (Daf-c) mutant animals.

The execution of the dauer program is crucially dependent on daf-12 (Antebi et al.,2000). Therefore, it is not surprising that its activity should be tightly regulated to prevent entry into diapause under favorable growth conditions. This is unlikely to be achieved through modulation of daf-12 expression as it is expressed at high levels at the L2 stage in the hypodermis (Antebi et al.,2000). Instead, we propose that daf-12 activity is regulated by the synthesis of a DAF-12 antagonist by DAF-9. According to this model, high level of daf-12 expression and transcriptional activity will be counteracted by an increase in hypodermal daf-9 expression that is daf-12 dependent. However, attenuation or prevention of daf-9 hypodermal expression may be a prerequisite for the dauer program, as seen in daf-2(–) and daf-7(–)animals. In this case, DAF-12 may disengage from the daf-9 promoter. Alternatively, DAF-12 may repress daf-9 hypodermal expression through recruitment of co-repressor complexes in response to dauer inducing signals.

A DAF-12 response element is located within intron 1 of the daf-9b isoform. This 181 bp sequence also appears to specify daf-9spermathecal expression independent of DAF-12. A nuclear receptor consensus half site (TGTTCT) is found within this sequence, which is also evident in an analogous location in the C. briggsae daf-9 gene. Surprisingly,mutation of the nuclear receptor consensus binding site eliminated daf-9 spermathecal expression without affecting the hypodermal expression. We propose that an unidentified nuclear receptor may bind to the TGTTCT element and regulate daf-9 expression in the spermatheca. As the consensus binding site for DAF-12 is not known at present, we cannot exclude the possibility that DAF-12 may bind directly to other parts of the 181 bp intron 1. Alternatively, DAF-12 may be part of a transcriptional cascade and indirectly control daf-9 hypodermal expression through other transcription factors.

Our results show that daf-9 signals at a pivotal position in the dauer pathway to integrate daf-2 insulin-like and daf-7TGFβ-like signaling pathways. The next challenge will be to identify the DAF-9 substrate that should shed light on the nature of the hormonal signal.

We thank Patrick Hu and Weiqing Li for critical reading of the manuscript;Snjezana Joksimovic and Xue Li for technical assistance; members of the Ruvkun laboratory for helpful discussions; Theresa Stiernagle at the Caenorhabditis Genetics Center for providing strains; Andrew Fire for GFP vectors; Yuji Kohara for cDNA clones; and Adam Antebi and Birgit Gerisch for communication of unpublished results. This work was supported by a long term fellowship from the Human Frontier Science Program to H.Y.M. and by grants from the NIH to G.R.