The C. elegans developmental timing protein LIN-42 regulates diapause in response to environmental cues

Animals monitor a variety of environmental cues, including seasonal variation in day length as well as more unpredictable and stress-inducing fluctuations in temperature and nutrient availability, and they respond to these cues by making adaptive changes in metabolism, physiology and behavior. The stress responses of juveniles are further complicated relative to those of adults, as they must integrate stress-induced changes into developmental programs (Rougvie, 2005; Edgar, 2006). Under sufficient stress, animals can stall temporal progression of development and delay sexual maturation until conditions improve, a complex process requiring synchronous changes in multiple developmental pathways.

The influence of environmental conditions on developmental timing has become an important area of investigation in humans, where, for example, studies have correlated adolescent obesity with earlier onset of puberty (Kaplowitz, 2008). To understand the impact stress-related cues have on developmental programs, several groups have applied genetic methods to model organisms. Such studies have revealed that multiple conserved signaling pathways (e.g. insulin, TGFβ, TOR and nuclear receptor signaling pathways) are used to coordinate developmental progression with external stimuli through modulation of body size, nutrient storage and the timing of developmental events (for reviews, see Edgar, 2006; Hu, 2007).

One of the best-understood developmental responses to environmental stress occurs in the nematode C. elegans. Under optimal growth conditions, C. elegans develops through four larval stages (L1-L4) prior to achieving reproductive competence as an adult. Development through these larval stages is coordinated temporally by the genes of the heterochronic pathway, which ensure that developmental events occur at the correct time (Resnick et al., 2010; Moss, 2007). If L1 larvae encounter an unfavorable environment (e.g. food shortage, overpopulation, high temperature), inputs from insulin, TGFβ and hormonal signaling pathways (Fig. 1A-C) suspend reproductive development after the L2 stage, and the animal forms a dauer larva, an alternative third larval stage specialized for long-term survival and dispersal (Cassada and Russell, 1975; Hu, 2007). These signaling pathways monitor environmental status and relay that information to mediate the decision to enter the dauer pathway. Once the decision to form a dauer is made, not only must the genetic pathways required for the L2-to-dauer transition be activated, but the timing mechanisms that promote continued developmental progression through the L2-to-L3 transition must be suspended. Therefore, the onset of dauer formation provides a unique opportunity to study the interaction between stress-responsive pathways and developmental timing mechanisms. Although a number of proteins have been identified that regulate one program or the other, the nuclear receptor (NR) DAF-12 provides key inputs to both (Antebi et al., 1998; Antebi et al., 2000).

DAF-12 acts as a molecular switch during development, indirectly sensing environmental conditions in order to make appropriate life-stage decisions (Fig. 1C) (Magner and Antebi, 2008). Under normal growth conditions, dietary cholesterol is converted into steroid hormones through a series of enzymatic steps that include modifications by the Rieske-like oxygenase DAF-36 and the cytochrome P450 DAF-9 (Gerisch et al., 2001; Jia et al., 2002; Motola et al., 2006; Rottiers and Antebi, 2006). The resulting hormones (known as dafachronic acids) bind the ligand-binding domain (LBD) of DAF-12 (Motola et al., 2006), and the complex then promotes the L2-to-L3 transition and inhibits dauer formation. Under poor growth conditions, hormone levels drop and ligand-free DAF-12 interacts with the SHARP co-repressor homolog DIN-1 (Ludewig et al., 2004; Motola et al., 2006), yielding a complex with the opposite activities: DAF-12–DIN-1 promotes dauer formation and blocks reproductive development through the L3 stage. In addition to their dauer phenotypes, daf-12 mutants display timing defects in several tissues during continuous development (Antebi et al., 1998). Although there are multiple phenotypic classes of daf-12 mutants, a general theme is that L2 stage-specific events are reiterated during L3, resulting in a retarded phenotype.

Here we report our discovery that the heterochronic gene lin-42 acts with daf-12 at the critical juncture where stress-responsive and developmental timing pathways intersect. In dauer formation, LIN-42 acts in opposition to ligand-free DAF-12 by promoting continuous development through the activity of the ligand-bound form of DAF-12. LIN-42 activity is essential for reproductive development under conditions of mild stress; its activity provides robustness to the system, ensuring that development continues without interruption. By contrast, when environmental conditions mandate dauer formation for survival, the LIN-42 temporal expression pattern is significantly altered to allow for dauer entry.

lin-42 and daf-12 are also uniquely linked during continuous development as they time a similarly broad set of events, including gonad morphogenesis (Antebi et al., 1998; Tennessen et al., 2006), a process in which other heterochronic genes apparently have no role. In contrast to daf-12, loss of lin-42 activity causes precocious execution of later-stage events (Jeon et al., 1999; Tennessen et al., 2006).

It is noteworthy that lin-42 is the C. elegans homolog of the circadian rhythm gene period (Jeon et al., 1999) because metabolic homeostasis and stress responses are coordinated with the circadian clock in other organisms (Hardin et al., 1990; Yang et al., 2006; Kohsaka et al., 2007). Our work reveals an intimate relationship between LIN-42 and DAF-12 activities during dauer formation, and provides important insight into how Period-like proteins might control metabolic homeostasis by regulating the activity of nuclear receptors.

Nematode growth was carried out as described previously (Sulston and Hodgkin, 1988) using nematode growth media (NGM) supplemented with 5 μg/ml cholesterol except where noted. All strains containing daf-7 or daf-2 alleles were maintained at 15°C.

Strains were constructed using standard methods; the more complex manipulations are outlined here. lin-42(lf); daf-12(lbd)/+ strains were constructed by crossing lin-42/+ males to daf-12 hermaphrodites and identifying F2 progeny that were semi-Dpy [indicating homozygosity of lin-42(ve11)] and that segregated animals with the daf-12 gonad migration (Mig) phenotype. lin-42(lf); daf-12(lbd)/+ animals segregated one quarter lin-42; daf-12 progeny that had a synthetic constitutive dauer formation (SynDaf) phenotype and were sterile as adults. To test lin-42(+)-mediated rescue of the SynDaf phenotype, lin-42(lf); daf-12(lbd)/+ animals were injected with the constructs indicated, and transgenic double homozygotes were selected by picking fertile GFP⁺ Mig animals.

The lin-42(ve11); din-1(dh149); daf-12(rh285) triple mutant was constructed by crossing din-1/+ males to lin-42(ve11); daf-12(rh285)/+ hermaphrodites. An F1 animal segregating all three genotypes was identified, and its homozygous daf-12(rh285) progeny were singly picked based on their Mig phenotype. F3 animals homozygous for din-1(dh149) were then selected based on their non-Mig phenotype, and their progeny were screened for Dpy animals, indicating the presence of lin-42(ve11). The resulting triple-mutant strain was confirmed by sequencing.

To construct strains containing daf-16(mu86), the deletion allele was followed via PCR using the primers MF396 and MF425 (5′-TGTCTCTCAATCGGCCACCA-3′ and 5′-TGCAAGAAGTGGATTCTGAGCA-3′). daf-5(e1386) was followed via its StyI polymorphism (da Graca et al., 2004) after amplification with primers JMT131 and JMT159 (5′-GGTGTCTCCATAGCGGGTCTCTTCG-3′ and 5′-CAGTTCCGAGCAGTCATGG-3′).

C. elegans transformations were performed as described previously (Mello et al., 1991) using either sur-5::gfp (Yochem et al., 1998) or str-1::gfp (Troemel et al., 1997) as co-injection markers. The transgenes used for rescuing lin-42(lf) were described previously (Tennessen et al., 2006). Pdpy-7::lin-42C::gfp::unc-54 (pJT97) contains 1271 bp of the dpy-7 promoter driving expression of a lin-42C cDNA fused to gfp, and was injected at 2.5 ng/μl. Pdpy-7::lin-42C::gfp::unc-54 causes some larval lethality, the penetrance of which varies between transgenic lines. These animals were not counted in our assays.

LIN-42 immunofluorescence was performed as described previously (Tennessen et al., 2006) using MH27 staining as a positive control. Imaging and quantification of daf-9::gfp were performed as described previously (Gerisch et al., 2001).

Larvae were synchronized using the hypochlorite method of egg selection and RNA was prepared using Trizol Reagent (Invitrogen) with the inclusion of ∼500 μm baked glass beads (G-8772; Sigma) to aid cuticle disruption. Contaminant DNA was removed using the DNA-free Kit (Ambion), and cDNA was generated using random hexamers and Transcriptor Reverse Transcriptase following the manufacturer's protocols (Roche). Real-time (RT) PCR was carried out using FastStart SYBR Green (Roche) and a Realplex 2 (Eppendorf). lin-42 message levels (primers AR205 and AR262: 5′-CCACTGACCCGAGAAGCAC-3′ and 5′-GAGTTGGTGCCACTTGTCGG-3′) were quantitated relative to those of ama-1 (primers AD78 and AD79: 5′-AGGAGATTAAACGCATGTCAG-3′ and 5′-CATGTCATGCATCTTCCAC-3′). Analysis was carried out using the comparative C_T method, comparing each time point to the mean value for the wild-type mid-L1 stage (Schmittgen and Livak, 2008). Four independent time courses were carried out for daf-7 and two for wild-type populations. At each time point for a given time course, three populations of worms were harvested independently and processed to make three separate cDNA pools. Each cDNA pool was then analyzed in triplicate by quantitative (q) RT-PCR to determine lin-42 and ama-1 expression levels. Not all time points were assayed in each data set.

Dauer formation was quantified by allowing individual hermaphrodites to lay eggs for 12-16 hours at the indicated temperature on plates seeded with E. coli OP50. To analyze mutants that do not form a functional vulva, gravid adults were placed in M9, cut with a hypodermic needle, and the released eggs were transferred to plates. Two days later, each brood was scored for dauer formation. Progeny of at least ten hermaphrodites (split between at least two trials performed on different days) were scored for each strain. For cholesterol deprivation experiments, hermaphrodites were placed on sterol-depleted media (NGM made with agarose instead of agar, and lacking peptone and added cholesterol), seeded with 150 μl of OP50 bacteria that had been washed three times with M9 buffer and concentrated 20:1 from the initial culture volume (Yochem et al., 1999). For RNAi experiments, percent dauer formation was calculated as the average of at least three different sets of double-stranded (ds) RNA injections, with a minimum of ten hermaphrodites per injection.

Animals homozygous for either lin-42(n1089) or lin-42(ve11) in combination with the daf-12 LBD mutations (rh193, rh284, rh285 and rh61) are sterile. To extrapolate the SynDaf phenotype of double-homozygous strains, the progeny of lin-42; daf-12/+ hermaphrodites were scored for dauer formation. The background rate for each daf-12 strain was subtracted from the rate of dauer formation for each heterozygous brood and multiplied by four. A SynDaf phenotype was not observed when daf-12(rh193) was assayed in combination with lin-28(ga54), lin-28(RNAi), hbl-1(ve18) or lin-41(RNAi) (0% in each case; n≥360).

dsRNA synthesis and microinjections were conducted as described previously (Kamath et al., 2001) using pPD129.36. lin-42 dsRNA was synthesized from pJT27, which contains cDNA spanning exons 6-9 (Tennessen et al., 2006). lin-41 dsRNA was synthesized as described previously (Lin et al., 2003), and lin-28 dsRNA was made from an AgeI/SacI fragment (pJT101). For each strain analyzed, the progeny of ten injected hermaphrodites were scored for dauer formation, omitting from analysis embryos laid in the first 24 hours post-injection. For statistical purposes, each injection set was scored as one value and at least three sets were scored for each strain.

daf-12 and lin-42 cDNAs were cloned into the yeast two-hybrid vectors pGBKT7 and pGADT7, and assays were performed using the Clontech Matchmaker GAL4 Two Hybrid System 3 (Clontech). To test for interactions, constructs were transformed pairwise into yeast strain AH109. Transformants were grown overnight in minimal media lacking leucine and tryptophan (SD-L-T), and interactions were tested for by growth on SD-L-T-histidine-adenine+X-α-gal plates, assessing activation of three reporters: ADE2, HIS3 and MEL1. Murine p53 and SV40 large T-antigen were used as a positive control, and T-antigen and Lamin C were used as negative controls. None of the test clones caused appreciable activation of the reporters when tested alone or versus empty vectors. Western analyses confirmed accumulation of fusion proteins in all strains.

lin-42 and daf-12 act antagonistically during continuous reproductive development (Tennessen et al., 2006). If this relationship were to hold true for the regulation of dauer formation, lin-42 mutants should have a phenotype opposite to the dauer defective (Daf-d) phenotype of daf-12(lf) mutants and should be hypersensitive to forming dauers. Although lin-42 mutants do not form dauers inappropriately under normal growth conditions, they are sensitized toward dauer formation. When grown in the presence of food and at low population density, but under conditions of temperature stress (27°C), 71% of lin-42(ve11) and 21% of lin-42(n1089) animals became dauer larvae compared with 2% of wild-type larvae (Fig. 1D). This characteristic of revealing a dauer phenotype when exposed to temperature stress is shared by mutations in several genes that alter the outputs of dauer signaling pathways (Ailion and Thomas, 2000; Ailion and Thomas, 2003; Rottiers et al., 2006).

The observation that ve11 exhibits a more penetrant phenotype than does n1089 parallels previous studies of their heterochronic phenotypes (Tennessen et al., 2006) and suggests that the more 3′ exons of lin-42 are important in the control of dauer formation. Therefore, studies reported here focus on the lin-42(ve11) allele and elimination of lin-42 activity via RNAi. Transgenic expression of lin-42D, the downstream isoform that rescues lin-42(lf) heterochronic defects (Tennessen et al., 2006), also rescued the lin-42(ve11) dauer phenotype (Fig. 1D,E), demonstrating that one function of wild-type LIN-42 is to inhibit dauer formation.

The lin-42(lf) high-temperature-induced dauer (Hid) phenotype could be a non-specific consequence of the precocious phenotype exhibited by these mutants. For example, if an inhibitor of dauer formation is expressed in a stage-specific manner, the temporal shift in tissue identities associated with lin-42(lf) might indirectly produce a Hid phenotype by changing the way dauer development is regulated. In this case, other precocious mutants should exhibit a similar Hid phenotype. However, when the precocious heterochronic mutants hbl-1(ve18), lin-28(ga54) and lin-41(ma104) (Moss et al., 1997; Slack et al., 2000; Abrahante et al., 2003; Lin et al., 2003) were raised at 27°C and scored for dauer formation, the animals did not form dauers at a significantly higher frequency than did the wild type (Fig. 1D). Furthermore, all of the lin-42 heterochronic phenotypes are dependent on the zinc-finger transcription factor LIN-29 (Rougvie and Ambros, 1995; Tennessen et al., 2006), but the lin-42(lf) Hid phenotype is not lin-29 dependent. The majority of lin-42(ve11) lin-29(n836) double mutants raised at 27°C formed dauers, demonstrating that suppression of the lin-42(lf) heterochronic phenotypes has no effect on the dauer phenotype (Fig. 1D). lin-29 single mutants also exhibited a mild Hid phenotype, an observation consistent with previous studies implicating lin-29 in inhibition of dauer formation (Liu and Ambros, 1989; Liu et al., 2004).

Because LIN-42 acts to inhibit dauer formation, the LIN-42 accumulation pattern might differ between L2 stage animals undergoing continuous development and those in the pre-dauer state (L2d) that are destined to form dauers. To test this hypothesis, LIN-42 accumulation was analyzed in daf-7(e1372) and daf-2(e1370) temperature-sensitive mutants, which are dauer constitutive (Daf-c) when raised at 25°C (Riddle et al., 1981). During continuous development in wild-type animals, LIN-42 levels cycle during each larval stage, peaking in the intermolt and becoming undetectable in the molt (Tennessen et al., 2006). This pattern was observed during the L1 stage in wild-type, daf-2 and daf-7 mutants; LIN-42 accumulation was spatially and temporally indistinguishable among these strains (Fig. 2; data not shown). However, the pattern differed significantly between wild-type and Daf-c mutants after the L1 molt. LIN-42 was detected in wild-type L2 stage animals soon after the first molt and remained detectable (Fig. 2C) until the L2 molt, after which it disappeared, returning following the molt in the L3 stage (Fig. 2D). By contrast, LIN-42 was absent from daf-7 and daf-2 mutants in the early L2 stage and remained undetectable in the intermolt through 26 hours of larval development (Fig. 2G,K; data not shown), ∼7-8 hours after the temperature-sensitive period for dauer formation ends in each of these mutants (Swanson and Riddle, 1981). This suggests that the presence of LIN-42 could be inhibitory to dauer formation. A pulse of LIN-42 protein accumulation was observed during the L2d molt in Daf-c mutants, centered around 32 hours (Fig. 2N; data not shown). At this time, daf-7 mutants are molting into the dauer stage and undergoing dauer remodeling, which includes constriction of the pharynx and body and synthesis of the dauer cuticle (Swanson and Riddle, 1981) (K.J.O. and A.E.R., unpublished), suggesting that LIN-42 might contribute to these processes.

The response of lin-42 expression to dauer entry is manifest at the RNA level. We performed a series of qRT-PCR experiments to examine lin-42 mRNA levels in daf-7 mutants grown at 25°C; a summary time course is shown in Fig. 2M and additional information is shown in Fig. S1 in the supplementary material. qRT-PCR confirmed the previously reported cyclical pattern of lin-42 mRNA accumulation in wild-type animals (Jeon et al., 1999), with levels peaking during each intermolt and declining during the molts. A similar pattern was observed during the L1 stage in daf-7 mutants grown at 25°C (Fig. 2M). However, in contrast to the wild type, lin-42 mRNA levels did not rise again in the early L2; they remained low or absent in daf-7 mutants until the L2d molt, an additional 12-15 hours. By this time, wild-type animals had completed the L2 and L3 stages.

Many genes are downregulated as animals initiate dauer development and transit through the dauer stage (Wang and Kim, 2003; Liu et al., 2004). Thus, an important question is whether the temporal alteration of LIN-42 accumulation during the L2d intermolt is functionally significant for dauer regulation or is instead a passive consequence of the physiological changes associated with preparing for dauer entry. To address this question, dauer formation was examined in daf-7 and daf-2 mutant animals engineered to express lin-42C::gfp from a dpy-7 promoter (Fig. 3; see Table S1 in the supplementary material; data not shown). In contrast to the absence of endogenous LIN-42 during the early second larval stage of daf-7 mutants, LIN-42C::GFP was detected in the L2 stage hypodermis of daf-7(e1372); veEx[Pdpy-7::lin-42C::gfp] animals raised at 25°C (Fig. 3A). Although these animals appeared to develop through an L2d-like stage, as did daf-7 mutants raised at 15°C (as judged by their dark intestine and the extended length of the larval stage), 83% of these animals developed directly into L3 larvae rather than forming dauers. By contrast, their nontransgenic siblings instead arrested as dauer larvae, as did control transgenic animals expressing gfp alone from the dpy-7 promoter (Fig. 3B-E). Similarly, 79% of daf-2(e1368); veEx[Pdpy-7::lin-42C::gfp] animals raised at 25°C did not form dauers (Fig. 3B). These results indicate that forced expression of LIN-42 during L2d can interfere with dauer formation. Interestingly, when a stronger daf-2 allele, e1370, was used, 34% of the transgenic animals bypassed dauer formation, but they failed to complete development and instead arrested as L3 stage larvae (Fig. 3F,G), a phenotype reminiscent of daf-2(lf); daf-12(lf) double mutants or daf-2(lf) mutants treated with exogenous DAF-12 ligand (Vowels and Thomas, 1992; Motola et al., 2006).

Dauer formation is controlled by multiple inputs, including signaling through conserved insulin/IGF and TGFβ pathways that converge upon the NR DAF-12. In order to position lin-42 with respect to these well-established dauer formation pathways, the lin-42(lf) Hid phenotype was examined in combination with dauer defective (Daf-d) mutations in daf-16 and daf-5, which encode the most downstream components of the insulin and TGFβ signaling pathways, respectively (Fig. 1A,B) (Ogg et al., 1997; da Graca et al., 2004). daf-16(lf) and daf-5(lf) mutations failed to suppress the lin-42(lf) Hid phenotype (Fig. 4A), placing lin-42 downstream of, or in parallel to, these pathways. By contrast, lin-42(ve11); daf-12(rh61rh411) double mutants failed to form dauers when raised at 27°C (Fig. 4A), indicating that the Hid phenotype requires daf-12 activity and that lin-42 acts through (or in parallel to) daf-12.

Previous studies have shown that Daf-c mutations in components of the DAF-12-based steroid hormone signaling pathway result in the inappropriate formation of partial dauers with normal dauer alae, but they fail to radially constrict their pharynxes (Antebi et al., 2000; Gerisch et al., 2001; Jia et al., 2002; Ohkura et al., 2003; Li et al., 2004; Rottiers et al., 2006). lin-42 mutants raised at 27°C similarly formed partial dauers with robust dauer alae and unconstricted pharynxes (Fig. 5A-D), suggesting that LIN-42 and DAF-12 regulate dauer development through a common mechanism.

To further examine the relationship between lin-42 and DAF-12-based hormone signaling, lin-42 activity was depleted in animals weakly compromised for function of the P450 DAF-9, which is involved in ligand synthesis (Fig. 1C) (Gerisch et al., 2001; Jia et al., 2002; Motola et al., 2006). Animals bearing lin-42(lf) mutations or the weak daf-9(lf) allele rh50 rarely formed dauers when raised at 25°C. By contrast, 33% of lin-42(ve11); daf-9(rh50) and 69% of lin-42(RNAi); daf-9(rh50) animals produced a synthetic constitutive dauer formation phenotype (SynDaf) (Fig. 4Bi). Similarly, loss of function of SDF-9, a phosphatase that appears to act upstream of DAF-9 (Ohkura et al., 2003), did not induce dauer formation in animals raised at 25°C, but 36% of sdf-9(ut163); lin-42(RNAi) animals grown in parallel were SynDaf (Fig. 4Bi). The lin-42 SynDaf phenotype appears to be specific to components of the dafachronic acid pathway, as lin-42(ve11) failed to enhance the dauer formation phenotypes of either daf-2(e1370) or daf-7(e1372) mutants when raised at 20°C, and lin-42(RNAi) did not produce a dauer phenotype in combination with other mutations tested for which SynDaf phenotypes have been described [akt-1(ok525), unc-3(e151) or unc-31(e169) mutants raised at 25°C (see Table S2 in the supplementary material)] (Ailion and Thomas, 2000; Ailion and Thomas, 2003). The observed genetic interactions are consistent with the idea that loss of lin-42 activity makes animals hypersensitive to changes in DAF-12-based hormonal signaling.

Mutations in genes such as daf-9 and sdf-9 that lower dafachronic acid signaling cause animals to form dauers when raised on plates containing food but lacking cholesterol (Gerisch et al., 2001; Jia et al., 2002; Ohkura et al., 2003; Rottiers et al., 2006). lin-42 mutants behaved in a similar fashion; they formed dauers when grown on cholesterol-depleted media (Fig. 4Bii), again linking lin-42 function with dafachronic acid signaling.

If lin-42(lf) makes animals hypersensitive to changes in DAF-12 ligand production, then mutations that weaken the interaction between DAF-12 and its ligand might also result in SynDaf phenotypes in combination with loss of lin-42 function. Indeed, the daf-12 LBD mutations rh193, rh284 and rh285 (Fig. 4C) (Antebi et al., 2000), which alone are non-Daf, produced a strong SynDaf interaction when lin-42 activity was also disrupted via RNAi or ve11 (Fig. 4Di, Fig. 5E,F), and this phenotype was rescued by transgenic expression of wild-type lin-42 (Fig. 4Di; see Table S1 in the supplementary material). The SynDaf phenotype is dependent on increased amounts of ligand-free DAF-12, rather than a general disruption of daf-12 activity, because lin-42(RNAi) performed in the presence of daf-12(rh286), a mis-sense allele that does not disrupt the LBD (Antebi et al., 2000), was not SynDaf at 20°C or 25°C (Fig. 4C,Di; data not shown).

To determine whether the SynDaf phenotype of lin-42; daf-12(LBD) mutants requires insulin or TGFβ signaling, lin-42 RNAi was conducted in daf-16(lf); daf-12(rh193) and daf-5(lf); daf-12(rh193) double mutants. Similar to the lin-42(lf) Hid phenotype, dauer formation in daf-12(rh193); lin-42(RNAi) double mutants was not suppressed by the absence of either the FoxO gene daf-16 or the Sno gene daf-5 (Fig. 4Dii) (Ogg et al., 1997; da Graca et al., 2004), indicating that LIN-42 does not act in these two pathways.

Finally, the SynDaf phenotype is not simply a general interaction between daf-12(LBD) alleles and precocious heterochronic mutants, because depletion of lin-28, lin-41 or hbl-1 in a daf-12(rh193) background failed to elicit this response (see Materials and methods). Rather, the SynDaf interaction between lin-42 and daf-12 is unique among the known heterochronic genes.

LIN-42 function in dauer formation is also independent of dafachronic acid signaling. daf-9(m540) and daf-12(rh273) mutants have strong Daf-c phenotypes, albeit for different reasons: daf-9(m540) mutants form dauers because they do not produce dafachronic acid (Jia et al., 2002), whereas the daf-12(rh273) mutation renders the DAF-12 LBD unable to bind dafachronic acid (Antebi et al., 2000; Motola et al., 2006). Expression of LIN-42 from the Pdpy-7::lin-42C::gfp transgene suppresses the Daf-c phenotypes of these mutants, indicating that LIN-42 can block dauer formation in the absence of dafachronic acid signaling (Fig. 4Diii; see Table S1 in the supplementary material).

These genetic experiments suggest that LIN-42 antagonizes the activity of ligand-free DAF-12 downstream of the known dauer signaling pathways. The only other factor known to act at this point in dauer signaling is DIN-1, the SHARP co-repressor that binds ligand-free DAF-12, forming a dauer-promoting complex (Ludewig et al., 2004). din-1(lf) mutations cause a Daf-d phenotype and, previous to this study, were the only reported suppressors of the daf-12(LBD) Daf-c phenotype. We found that the SynDaf phenotype of lin-42(lf); daf-12(lbd) mutants is dependent on formation of a DAF-12–DIN-1 complex. A lin-42(ve11); din-1(dh149); daf-12(rh285) triple mutant did not form dauers (Fig. 4Di), consistent with a model in which LIN-42 antagonizes the DAF-12–DIN-1 complex. As an additional means of testing this model, we examined the interaction between lin-42(ve11) and rh61, an unusual allele of daf-12. The daf-12(rh61) lesion results in a premature stop early in the LBD that renders the DAF-12 protein insensitive to its ligand but also decreases the binding affinity between the DAF-12 LBD and DIN-1 (Fig. 4C) (Antebi et al., 2000; Motola et al., 2006). As a result of decreased DIN-1 binding, rh61 mutants are Daf-d even though they are unable to bind dafachronic acid. If LIN-42 inhibits dauer formation by antagonizing the DAF-12–DIN-1 complex, removal of LIN-42 in a rh61 background might allow this mutant to form dauers even with decreased complex formation. Consistent with this prediction, 12% of lin-42(ve11); daf-12(rh61) animals formed partial dauers (Fig. 4Diii). Interestingly, many of these animals formed dauer alae only over the anterior seam cells, whereas the posterior half appeared similar to that of L3 larvae (Fig. 5G), suggesting that positional information influences the dauer decision.

Finally, we monitored expression of a daf-9::gfp reporter (Gerisch and Antebi, 2004; Mak and Ruvkun, 2004) in lin-42(ve11) mutants. As previously reported, daf-9::gfp expression is upregulated under conditions with abnormally low levels of the ligand-bound DAF-12, such as in animals raised on cholesterol-deficient media or at high temperature (Fig. 5H,I) or in those carrying the LBD mutation daf-12(rh273) (Gerisch and Antebi, 2004; Mak and Ruvkun, 2004). Hypodermal DAF-9::GFP levels appeared to be unchanged in lin-42(ve11) animals raised at 20°C (Fig. 5I,J), indicating that lin-42(lf) neither appreciably decreases the production of the DAF-12 ligand nor alters the ability of DAF-12 to bind its ligand.

The genetic experiments described above indicate that LIN-42 and DAF-12 might interact in a transcriptional complex that regulates dauer formation. To test this hypothesis, LIN-42 isoforms were assessed for their ability to interact with DAF-12 peptides in a yeast two-hybrid system. In pairwise tests, LIN-42B interacted with DAF-12 in both bait and prey conformations, whereas LIN-42C interacted with DAF-12 only when expressed as bait (Fig. 6). These LIN-42 isoforms also interacted with DAF-12 peptides that represent the rh61 and rh285 truncations, and the interactions were noticeably stronger than those between LIN-42B/C and the longer DAF-12 peptide containing an intact LBD (Fig. 6). These results indicate that the C-terminus of DAF-12 can influence its ability to interact with LIN-42, perhaps reflecting a role for the AF2 homology domain, which regulates NR transcriptional activity via ligand-dependent conformational changes (reviewed by Wärnmark et al., 2003). Results from the smallest DAF-12 peptide tested, which contains the LBD but lacks the hinge domain and C-terminus, indicated that sequences within the LBD contribute to the interaction detected. The two shorter LIN-42 isoforms revealed little, if any, interaction with DAF-12 (data not shown), suggesting that the role of LIN-42D in dauer formation involves a weaker interaction with DAF-12 or that it regulates dauer formation through additional binding partners.

We report the identification of the developmental timing protein LIN-42 as a stress-responsive negative regulator of dauer formation. Our work positions LIN-42, a Period (Per) family member, at the nodal point where nutrient sensing interfaces with DAF-12 to mediate the choice between continuous reproductive development and dauer formation. Several lines of evidence converge to place LIN-42 function at the level of DAF-12: (1) when lin-42 mutants are stressed and inappropriately enter the dauer pathway, they form partial dauers similar to those produced by mutants with defects in dafachronic acid signaling, suggesting misregulation of DAF-12 activity; (2) ectopic LIN-42 expression can suppress the Daf-c phenotype of daf-9(m540) and daf-12(rh273), indicating that LIN-42 can act independently of dafachronic acid signaling; and (3) overexpression of lin-42 in a strong daf-2 loss-of-function mutant background causes animals to arrest as L3 larvae, a synthetic phenotype described previously for daf-2(lf); daf-12(lf) double mutants (Vowels and Thomas, 1992), or daf-2(lf) mutants treated with dafachronic acid (Motola et al., 2006), suggesting that lin-42 and daf-12 act at a similar position. Furthermore, consistent with a model of interplay between LIN-42 and DAF-12, the precocious defects associated with lin-42(lf) during continuous development are partially suppressed in lin-42(lf); daf-12(LBD)/+ animals (see Table S3 in the supplementary material), even though daf-12(LBD)/+ animals do not have retarded defects (Antebi et al., 1998). This finding indicates that even a modest increase in ligand-free DAF-12 can compensate for a lack of LIN-42.

Our results indicate that lin-42 acts antagonistically to daf-12 to regulate developmental timing and dauer formation, and together these two genes ensure proper development in a changing environment. As depicted in the model in Fig. 7 and as proposed by Antebi and colleagues (for a review, see Magner and Antebi, 2008), the balance between ligand-free and ligand-bound DAF-12 gates entry to dauer formation. Under favorable growth conditions of mild temperature, low population density and abundant food, dafachronic acids are synthesized and shift the balance of power to ligand-bound DAF-12, thereby disrupting the L3-inhibitory and dauer formation signals of ligand-free DAF-12, and continuous development ensues (Fig. 7A). Under these conditions, the presence of LIN-42 protein is required to prevent inappropriately early initiation of L3 fates (Jeon et al., 1999). Based on the findings reported here, we propose that when animals are stressed and dafachronic acid signaling is lowered, the presence or absence of LIN-42 is a crucial factor in the decision of whether to become a dauer. If animals encounter mild stress that does not necessitate dauer development, dafachronic acid signaling decreases, but the presence of LIN-42 prevents ligand-free DAF-12 from initiating dauer formation (Fig. 7B). By contrast, when harsh environmental growth conditions trigger dauer formation, both LIN-42 and ligand levels drop. The decreased concentrations of these components shift the equilibrium to the ligand-free DAF-12–DIN-1 complex, which, in the absence of LIN-42, now promotes dauer entry through the L2d predauer stage and inhibits L3 development (Fig. 7C). Thus, the interplay of LIN-42 and DAF-12 activities provides the worm with an additional level of modulation of the response to environmental conditions, increasing the robustness of the system and allowing reproductive development in cases of mild stress.

Important challenges for the future are to determine how environmental conditions and/or nutritional status regulate LIN-42 accumulation, and how changes in LIN-42 levels in turn modulate hormonal signaling via DAF-12. An interesting possibility suggested by our yeast two-hybrid assays is that LIN-42 regulates the activity of a DAF-12 transcriptional complex, perhaps with relevance to the recently described role for DAF-12 in modulating levels of heterochronic miRNAs (Bethke et al., 2009; Hammell et al., 2009). LIN-42, however, appears to be properly localized in a daf-12(0) mutant, as judged by anti-LIN-42 staining (data not shown), and thus is not an obligate DAF-12 partner for either stability or localization.

Links between heterochronic genes and dauer formation have long been known. For example, lin-14 activity prevents precocious dauer formation at the L1 molt (Liu and Ambros, 1989), and the phenotypes of most heterochronic mutants are suppressed by development through the dauer pathway (Liu and Ambros, 1989; Euling and Ambros, 1996). However, none of these other heterochronic genes appears to participate in the relationship between lin-42 and daf-12; the role of lin-42 in modulating the stress response is unique. The LIN-42 PAS domain could be a key component of this aspect of LIN-42 function by contributing to the sensing of growth conditions, as PAS domains are deployed as important environmental sensors in organisms ranging from archaea to humans (Taylor and Zhulin, 1999). Conceivably, PAS domain-containing LIN-42 isoforms could be regulated by metabolic cues. However, transgenic expression of LIN-42D, which lacks the PAS domain, is capable of rescuing the dauer phenotypes, so the PAS domain is not an obligate component of LIN-42-mediated inhibition of dauer entry.

This study demonstrates a role for a Per family protein in the mediation of a response to deteriorating environmental conditions and suggests an ancient role for Per proteins in the regulation of stress responses. Similar to the role of lin-42 in regulating dauer formation, the expression of both mouse and fly Per genes is sensitive to changes in external and intrinsic stressors. Per1 is rapidly upregulated when mice are exposed to physical stress or hormone fluctuations, suggesting that there is an increased demand for PER1 under stressful conditions (Balsalobre et al., 2000; Takahashi et al., 2001). Similarly, expression of Drosophila period is sensitive to a combination of insulin signaling and oxidative stress (Zheng et al., 2007). The dynamic role for LIN-42 in the control of dauer formation demonstrates a stress-responsive role for a Per family member in the nematode, and suggests that stress-induced changes in Per expression might prove to be functionally significant in other organisms.

LIN-42 modulation of ligand-free DAF-12 signaling provides an important insight into the relationship between Per proteins and NR signaling. Using a mechanism that is reminiscent of the genetic hierarchy between daf-12 and lin-42, circadian rhythms in mice are controlled by an elegant feedback loop between Per family members and the NRs REV-ERBα (Nr1d1 – Mouse Genome Informatics) and RORα (Preitner et al., 2002; Sato et al., 2004). As both REV-ERBα and RORα respond to metabolic changes, the interaction between Per proteins and these NRs could potentially coordinate physiology with circadian rhythms (for a review, see Ramakrishnan and Muscat, 2006). Indeed, PER2 was recently found to directly interact with a number of NRs including REV-ERBα, and this interaction with REV-ERBα appears to play a role in the regulation of glycogen storage (Schmutz et al., 2010).

Per proteins are also indirectly associated with other steroid hormone signaling pathways. For example, the levels of cortisol in humans and other mammals fluctuate with a 24-hour circadian cycle (Hellman et al., 1970), and the release of ecdysone in insects is often coordinated with circadian rhythms (Nijhout, 1994). Together with our work, these studies suggest that the inter-relationship between Per proteins and environmentally sensitive NR signaling represents an ancient genetic framework by which biological clocks can adapt to changes in external growth conditions. Future analysis of lin-42 should allow a mechanistic dissection of how a per family member coordinates environmental conditions with intrinsic timing mechanisms.

We thank Sarah Malmquist for performing preliminary Y2H experiments; Robert Herman, David Greenstein and Jeff Simon for comments on the manuscript; Meg Titus for insightful discussions; and Adam Antebi and the Caenorhabditis Genetics Center for strains. This work was supported by grants from the National Institutes of Health (RO1 GM50227 to A.E.R. and Predoctoral Training Grant HD007480 to J.M.T.) and by a University of Minnesota Doctoral Dissertation Fellowship to J.M.T. Deposited in PMC for release after 12 months.