Repression of a Potassium Channel by Nuclear Hormone Receptor and TGF-β Signaling Modulates Insulin Signaling in Caenorhabditis elegans

Transforming growth factor β (TGF-β) signaling acts through Smad proteins to play fundamental roles in cell proliferation, differentiation, apoptosis, and metabolism. The Receptor associated Smads (R-Smads) interact with DNA and other nuclear proteins to regulate target gene transcription. Here, we demonstrate that the Caenorhabditis elegans R-Smad DAF-8 partners with the nuclear hormone receptor NHR-69, a C. elegans ortholog of mammalian hepatocyte nuclear factor 4α HNF4α), to repress the exp-2 potassium channel gene and increase insulin secretion. We find that NHR-69 associates with DAF-8 both in vivo and in vitro. Functionally, daf-8 nhr-69 double mutants show defects in neuropeptide secretion and phenotypes consistent with reduced insulin signaling such as increased expression of the sod-3 and gst-10 genes and a longer life span. Expression of the exp-2 gene, encoding a voltage-gated potassium channel, is synergistically increased in daf-8 nhr-69 mutants compared to single mutants and wild-type worms. In turn, exp-2 acts selectively in the ASI neurons to repress the secretion of the insulin-like peptide DAF-28. Importantly, exp-2 mutation shortens the long life span of daf-8 nhr-69 double mutants, demonstrating that exp-2 is required downstream of DAF-8 and NHR-69. Finally, animals over-expressing NHR-69 specifically in DAF-28–secreting ASI neurons exhibit a lethargic, hypoglycemic phenotype that is rescued by exogenous glucose. We propose a model whereby DAF-8/R-Smad and NHR-69 negatively regulate the transcription of exp-2 to promote neuronal DAF-28 secretion, thus demonstrating a physiological crosstalk between TGF-β and HNF4α-like signaling in C. elegans. NHR-69 and DAF-8 dependent regulation of exp-2 and DAF-28 also provides a novel molecular mechanism that contributes to the previously recognized link between insulin and TGF-β signaling in C. elegans.


Introduction
Transforming growth factor b (TGF-b) signaling plays fundamental roles in cell proliferation, differentiation and apoptosis [1]. TGF-b binding induces the formation of a heterotetrameric type I and type II transmembrane receptor complex, and the type I receptor then phosphorylates the C-terminal Ser-X-Ser motif of receptor-associated Smad (R-Smad) transcription factors [2]. Phosphorylated R-Smads form homodimers and heterotrimers with a Co-Smad and translocate to the nucleus where they regulate the transcription of target genes [1].
TGF-b signaling is evolutionarily conserved; for example, in the lower metazoan Caenorhabditis elegans, two partially overlapping TGFb signaling pathways have been described. One controls body size and male reproductive development whereas the other controls the formation of a specialized larval stage, the dauer [3][4][5]. The dauer larva is a developmentally arrested, long-lived and stress resistant diapause stage, formation of which is induced by overcrowding, starvation, and high temperatures [5]. Screening for mutations that result either in constitutive (Daf-c) or defective (Daf-d) dauer development [5] revealed multiple signaling circuits that control the dauer developmental switch, including a classical TGF-b signaling pathway. The TGF-b ligand DAF-7 promotes larval growth and inhibits dauer arrest; accordingly, transcription of daf-7, which is expressed in the amphid single ciliated neuron I (ASI) chemosensory neurons, is repressed by the dauer-inducing pheromone [6]. Downstream of daf-7, daf-1 and daf-4 encode the nematode type I and type II TGF-b receptor kinases, respectively, and both genes are required for non-dauer larval development. DAF-1 and DAF-4 phosphorylate DAF-8, an R-Smad that antagonizes the Co-Smad DAF-3 to promote larval growth [7]. In line with these roles, daf-1, daf-4, daf-7, and daf-8 mutants all exhibit Daf-c phenotypes. In an interesting divergence from mammalian TGF-b signaling, the C. elegans Co-Smad DAF-3 represses the transcription of the antagonistic upstream genes daf-7 and daf-8 by directly binding their promoters, thus creating a positive feedback loop [7].
Dauer formation is also influenced by other sensing and signaling pathways, such as cyclic guanosine monophosphate (cGMP) signaling, steroid receptor action and insulin/insulin-like growth factor 1 (IGF-1) signaling (IIS) [5]. Interestingly, several recent reports indicate that IIS and TGF-b signaling interact in C. elegans. In the regulation of fasting/re-feeding induced quiescence, cGMP, insulin/IGF-1, and TGF-b signaling act as parallel activation signals for the protein kinase G (PKG) EGL-4 [8]. All three pathways also affect insulin gene expression, and the gene expression profiles of TGF-b and IIS pathway mutants exhibit a rather broad overlap [9][10][11]. Intriguingly, daf-7 mutants exhibit increased nuclear localization of the transcription factor DAF-16 [12], a key target of IIS that is also nuclear in worms carrying mutations in the insulin receptor gene daf-2. Moreover, the phosphatase PDP-1, which affects TGF-b signaling downstream of daf-7, and likely at the level of daf-8 and daf-14, also promotes DAF-16 nuclear localization and insulin gene expression [13]. Thus, PDP-1 and TGF-b signaling may directly promote the expression of insulin peptides, providing one possible mechanistic link between TGF-b signaling and IIS. Accordingly, both the Daf-c and the longevity phenotypes of TGF-b signaling mutants are suppressed at least partially in a daf-16 mutant [14] [9]. These studies support the notion that TGF-b signaling may in part act upstream of IIS to regulate dauer formation and longevity. However, a simple linear relationship between TGF-b and IIS is unlikely even in the regulation of life span, because the longevity of TGF-b pathway mutants is very modest compared to IIS pathway mutants [9]. In any case, the mechanisms that couple IIS and TGFb signaling remain poorly defined, and an apparent lack of recognizable Smad-binding sites amongst the genes co-regulated in IIS and TGF-b pathway mutants [9] suggests that these signaling pathways integrate upstream of transcriptional regulation.
Another mechanism that acts downstream of TGF-b signaling in the dauer formation cascade centers on the Nuclear Hormone Receptor (NHR) DAF-12 [15]. DAF-12 is one of 284 C. elegans NHRs that are involved in a variety of processes, including embryonic development, dauer formation, regulation of life span, molting, lipid metabolism, xenobiotic metabolism and excretory duct organogenesis [16] [17]. Of the 284 NHRs, most are related to a Hepatocyte Nuclear Factor 4 alpha (HNF4a) related ancestor. Most of these remain poorly characterized, although several HNF4a orthologs regulate lipid metabolism and/or life span [18][19][20][21][22]. To date however, DAF-12 remains the only NHR with a defined role in the regulation of dauer formation.
To better define the regulatory network and the transcriptional complexes associated with C. elegans DAF-8/R-Smad, we set out to identify its physical interaction partners. We describe a novel physical and functional interaction between DAF-8 and NHR-69 that affects dauer formation, longevity, and the expression of the potassium channel gene exp-2, which in turn affects the secretion of the insulinlike peptide DAF-28. Our data indicate that TGF-b signaling and HNF4a-like signaling exhibit regulatory crosstalk to influence the secretion of an insulin-like peptide in C. elegans, providing a novel mechanistic link between TGF-b signaling and IIS.

NHR-69 and DAF-8 physically interact in vitro and in vivo
To identify proteins that bind DAF-8, we used anti-FLAG immunoprecipitation to purify DAF-8 and associated peptides from a worm strain stably expressing a DAF-8::FLAG transgene [7]. Peptide identities were determined by mass spectrometry and we identified several peptides representing NHR-69 (data not shown). NHR-69 is one of the two closest C. elegans orthologs of human HNF4a with 39% amino acid identity and 60% similarity (as determined by reciprocal Blast analysis; see also [23]). To validate the interaction between these two transcription factors we generated anti-DAF-8 antibodies ( Figure S1) and transgenic animals expressing a translational NHR-69::GFP fusion protein (nhr-69p::nhr-69::gfp; see Materials and Methods). Co-immunoprecipitation (Co-IP) showed that NHR-69::GFP associates with DAF-8 in vivo ( Figure 1A). We also performed in vitro binding assays using GST-fusions and in vitro transcribed/translated proteins to test for direct interaction between NHR-69 and the functionally distinct SMAD proteins DAF-8, DAF-14, and DAF-3. We found that GST::NHR-69, but not GST alone, associated specifically with DAF-8, but not with DAF-14/R-Smad or DAF-3/Co-Smad ( Figure 1B). Together, these data suggest that NHR-69 and DAF-8 specifically interact in vitro and in vivo.
As DAF-8 and NHR-69 interact physically, we would expect their in vivo expression patterns to at least partially overlap. Our previous work showed that DAF-8 is expressed in numerous tissues, including the ASI neurons and the intestine [7]. A NHR-69::GFP fusion protein driven by the nhr-69 promoter was detected in the nucleus of the E8 intestinal precursor cells in developing embryos ( Figure 1D, 1F), and strong intestinal expression persisted throughout larval development until adulthood ( Figure 1L). In adults, we also detected expression in the ASI neurons ( Figure 1H, Figure S2), in the hypodermis, and in tail neurons ( Figure 1J, 1L and 1N). In all of these tissues, NHR-69::GFP was localized exclusively to the nucleus, as expected for a transcription factor. Taken together with our previous findings, NHR-69 and DAF-8 [7] expression overlaps in the ASI neurons and in the intestine, suggesting that these two factors may interact functionally in these two tissues.

Author Summary
All animals must ensure metabolic homeostasis; if they fail to do so, diseases such as obesity and diabetes can develop. To maintain glucose balance, insulin is secreted upon glucose intake in a highly regulated and coordinated process. Previous studies suggested that the transforming growth factor beta (TGF-b) signaling pathway regulates insulin secretion in mammals. In the genetically tractable roundworm Caenorhabditis elegans, TGF-b and insulin signaling modulate larval development and aging, although the molecular link between insulin and TGF-b signaling remains poorly understood. In this study, we show that the TGF-b signaling component DAF-8 partners with NHR-69, a nuclear hormone receptor, to control the expression of the potassium channel exp-2, which in turn modulates the secretion of an insulin-like peptide. A lossof-function exp-2 mutant exhibits increased insulin secretion and a shortened life span, whereas a gain-of-function mutant exhibits decreased insulin secretion. We also show that tissue-specific expression of nhr-69 in a pair of neurons that secrete neuropeptides causes reduced glucose content, increased insulin-like peptide levels and a lethargic phenotype. Because insulin and TGF-b signaling are linked to numerous diseases, our data may provide novel insights into the mechanisms contributing to pathophysiological changes.
not cause any obvious phenotypes in wild-type worms [22], we utilized a sensitized genetic background to test whether nhr-69 contributes to dauer formation. This worm strain, eri-1(mg366); sdf-9(m708), is hypersensitive to RNAi due to the eri-1 mutation [24], and sensitized to dauer formation due a mutation in the tyrosine phosphatase gene sdf-9, which acts parallel to IIS in C. elegans [25]. In this background, nhr-69 RNAi significantly enhanced dauer formation ( Figure S3).
To corroborate the RNAi data we obtained a strain carrying a deletion in the nhr-69 gene, nhr-69(ok1926) (see Materials and Methods). The deletion spans the 59 end of the gene and likely represents a null mutation. In agreement with the RNAi data, the nhr-69(ok1926) mutation enhanced the dauer formation of temperature-sensitive daf-7(e1372), daf-1(m40), daf-8(m85), and daf-14(m77) mutants at the semi-restrictive temperature (22.5uC Table 1; note that even null mutants in the TGF-b pathway like daf-8 are temperature-sensitive because dauer formation is temperature-dependent [26]).
Given this regulation and given that NHR-69::GFP is expressed in multiple tissues, we next addressed whether insulin or TGF-b signaling affect nhr-69 expression in any particular tissue. To this end, we generated strains that express the NHR-69::GFP fusion protein in the daf-2(e1370) and daf-7(e1372) mutant backgrounds. We observed normal NHR-69::GFP expression in the ASI neurons of daf-7(e1372) mutants (9663% of L2 larvae showed GFP fluorescence in both neurons), but NHR-69::GFP was barely detectable in the ASI neurons of daf-2(e1370) mutants (0.360.04% of L2 larvae showed GFP fluorescence in both neurons, p,0.001; Figure 2B c and d). Together, these data suggest that insulin

Insulin signaling is impaired in a daf-8 nhr-69 double mutant
The daf-8(m85) nhr-69(ok1926) double mutants exhibited defects in dauer entry and exit, processes regulated by insulin signaling [27]; thus, we next asked whether daf-8 and nhr-69 affect insulin signaling in C. elegans. First, we examined the expression of sod-3 and gst-10, two downstream targets of IIS, which are regulated by the transcription factors DAF-16 and SKN-1, respectively [30] [31]. SKN-1 is a transcription factor of the bZIP family that acts downstream of daf-2 in C. elegans. qPCR analysis showed that both transcripts were upregulated in the daf-8(m85) single mutant; for sod-3, this is consistent with previous findings that TGF-b signaling regulates daf-16 target genes [9]. Furthermore, sod-3 and gst-10 induction was enhanced in the daf-8(m85) nhr-69(ok1926) double mutant, suggesting that insulin signaling was further reduced in the double mutant ( Figure 3A, 3B).
Because adult life span is extended by reduced insulin signaling [28], we next tested whether nhr-69 and daf-8 affect longevity. Figure 3C shows that nhr-69(ok1926) single mutants exhibited a slightly extended life span relative to wild-type worms (mean life span 17.461.3 days, p = 0.027), whereas the daf-8(m85) mutant life span was similar to that of wild type (mean life span 14.660.8 days, p.0.05; Figure 3C). Notably, the life span of daf-8(m85) nhr-69(ok1926) double mutants was significantly longer than that of wild type worms (mean life span 21.561.7 days, p,0.001; Figure 3C), consistent with reduced insulin signaling. By contrast, animals over-expressing NHR-69::GFP exhibited a shortened life span (mean life span 11.860.6 days, p,0.001) ( Figure 3C and Table S1). Together, these data suggest that daf-8 and nhr-69 cooperatively affect insulin signaling in C. elegans.
Defective neuropeptide secretion in daf-8 nhr-69 mutants Next, we wished to define the mechanism by which nhr-69 and daf-8 affect insulin signaling. Given that both genes are expressed in the neuropeptide-secreting ASI neurons, we hypothesized that they might control the release of neuropeptides such as the insulinlike peptide DAF-28, which is secreted by the ASI neurons [32]. To examine this possibility, we first attempted to introduce the DAF-28::GFP transgene into the daf-8(m85) mutant background. For unknown reasons, these worms exhibited a non-recoverable Daf-c phenotype, making it impossible to study DAF-28::GFP secretion in this background (data not shown). As an alternative approach, we used worms expressing an aex-3 promoter-driven ANF::GFP (human Atrial Natriuretic Factor-GFP fusion protein), which can be used as a surrogate to study neuropeptide secretion in C. elegans [33]. aex-3p::ANF::GFP is expressed in a pan-neuronal fashion, secreted into the inter-organ space (pseudocoelom), and then endocytosed by the coelomocytes, scavenger-like cells that endocytose secreted material [33]. The accumulation of ANF::GFP in coelomocytes was similar in wild-type worms and daf-8(m85) and nhr-69(ok1926) single mutants, but daf-8(m85) nhr-69(ok1926) double mutants accumulated less ANF::GFP ( Figure 3D d, Figure S4A). The overall neuronal ANF::GFP gene expression was comparable in all worm strains ( Figure S4C).
The differential ANF::GFP accumulation in coelomocytes could result either from differences in secretion or endocytosis. Thus, we tested whether the daf-8(m85) nhr-69(ok1926) mutants are generally defective for endocytosis. Specifically, we examined a signal sequence GFP (ssGFP) driven by the myo-3 promoter, which is expressed in muscle cells, secreted into the pseudocoelom and endocytosed by coelomocytes [34]. The accumulation of ssGFP in coelomocytes was similar in wild-type worms and in daf-8(m85) nhr-69(ok1926) double mutants ( Figure 3D f and h, Figure S4B), suggesting that the daf-8(m85) nhr-69(ok1926) mutants are defective specifically for neuropeptide secretion and not for endocytosis.
The Y1H data suggested that NHR-69 can directly bind the exp-2 promoter. Given the similarity of NHR-69 to mammalian HNF4a, we examined the 2.5 kb region 59 of the exp-2 start codon for HNF4a-like consensus binding sites [37] and identified four putative elements ( Figure 4B). To test whether NHR-69 directly binds any of these sites, we performed chromatin immunoprecipitation (ChIP) on mixed-stage nhr-69p::nhr-69::gfp animals using an  anti-GFP antibody. NHR-69 was found to directly associate with at least two of these sites ( Figure 4C). Intriguingly, NHR-69 binding was reduced, albeit not entirely lost, in daf-8(m85) mutants, suggesting that DAF-8 is required for optimal NHR-69 binding to the exp-2 promoter ( Figure 4C).
To confirm that NHR-69 directly regulates the exp-2 promoter we developed a mammalian reporter assay. We previously showed that the nuclear localization of DAF-8 requires the TGF-b Type 1 receptor DAF-1 [7], and others have reported that mutating two C-terminal serines (Ser465 and Ser467) to glutamates results in phosphomimetic, constitutively active human Smad2 [38]. Generating a phosphomimetic form of DAF-8 (pmDAF-8) by changing two orthologous C-terminal serines (S543 and S544) to glutamate followed by immunohistochemistry showed that pmDAF-8 localizes to the nucleus in transiently transfected HEK293 cells ( Figure 5A; resembling the nuclear localization of a DAF-8::GFP fusion protein in C. elegans [7]). NHR-69::GFP was also localized to the nucleus when transiently transfected into HEK293 cells ( Figure 5A). Importantly, NHR-69::GFP repressed the activity of a construct in which the wild-type exp-2 promoter is fused to a luciferase reporter ( Figure 5B). Moreover, this repression was enhanced by co-transfection of pmDAF-8 ( Figure 5B), resembling the results obtained in C. elegans.
EXP-2 functionally contributes to insulin signaling downstream of NHR-69 and DAF-8 The above data indicate that exp-2 is a direct target of nhr-69 and daf-8; thus, we next wished to test whether exp-2 was functionally required downstream of daf-8 and nhr-69. EXP-2 is a potassium channel that regulates the action potential in pharyngeal muscle [36]. Worms carrying a heterozygous exp-2 gain-of-function allele (exp-2(sa26), encoding an EXP-2 protein with a C480Y point mutation) exhibit defective feeding behavior, hyperactive head movements, and egg-laying defects, whereas homozygous worms arrest terminally after hatching [36] [39]. Worms carrying null mutations in exp-2, such as the exp-2(sa26ad1426) allele carrying a T457I mutation, exhibit no defects in defecation or head movement, but have long-lasting pharyngeal muscle contractions [36]. Together, these mutants provide an opportunity to address the functional requirement of exp-2 in insulin signaling.
To study the effect of exp-2 loss, we generated a daf-8(m85) nhr-69(ok1926) exp-2(sa26ad1426) triple mutant and found that the lifespan extension due to loss of daf-8 and nhr-69 was abolished by concomitant loss of exp-2 ( Figure 6F, Table S1). Taken together, our results suggest that NHR-69 and DAF-8 cooperate to directly regulate exp-2, and that exp-2 is a critical downstream target of these two transcription factors.
If EXP-2 regulated DAF-28 secretion, it would be expected to affect downstream insulin target genes and phenotypes such as longevity. Indeed, the mRNA levels of sod-3 and gst-10 were upregulated in gain-of-function exp-2(sa26) mutants ( Figure 6B, 6C), and we also observed increased nuclear localization of DAF-16::GFP in the exp-2(sa26) background ( Figure S6), consistent with decreased insulin signaling [43]. Moreover, exp-2(sa26ad1426) lossof-function mutants exhibited a significantly shortened life span compared to that of wild-type worms (mean life span: 9.460.6, p = 0.0116; we were unable to assess adult life span of exp-2(sa26) mutants due to lethality). In contrast, the life span of the exp-2(sa26ad1426); daf-2(e1370) double mutants was comparable to that of the daf-2(e1370) mutant itself ( Figure 6E, p = 0.7606), suggesting that exp-2 acts upstream of daf-2 to regulate life span, consistent with its role in DAF-28 secretion.
Because exp-2 acts selectively in ASI neurons to control DAF-28 secretion ( Figure 6A), we next addressed whether life span regulation by exp-2 also originates from the ASI neurons. Expression of wild-type exp-2 specifically in the ASI neurons (gpa-4p::exp-2) rescued the short life span phenotype of exp-2(sa26ad1426) mutant worms (14.3 days, p = 0.0139), whereas pharyngeal-specific expression did not (8.7 days, p = 0.8326; Figure 6G). This suggests that exp-2 plays an important role in ASI-neurons to regulate adult life span, possibly through its effect on DAF-28 secretion. Taken together, these data provide evidence that life span is inversely related to EXP-2 activity.

Overexpression of NHR-69 in the ASI neurons confers a hypoglycemic phenotype
As NHR-69::GFP is expressed in ASI neurons and DAF-28 is produced and secreted in these cells [32], we asked whether NHR-69 activity in the ASI neurons directly affects insulin signaling. First, we wished to ensure that an nhr-69::gfp translational fusion was biologically functional; thus, we expressed nhr-69::gfp from its own promoter (mIs39[nhr-69p::nhr-69::gfp]). Figure 3C shows that worms expressing this construct in the daf-8(m85) nhr-69(ok1926) mutant background exhibited a shortened life span compared to wild-type and a normal dauer recovery phenotype, suggesting that functional NHR-69 was overexpressed in these worms ( Figure 3C and Table 2). Having assured that nhr-69 rescue can work in principle, we generated two transgenic lines that expressed NHR-69::GFP under the control of the ASI-specific gpa-4 promoter (gpa-4p::nhr-69::gfp) [41]. In these worms, we observed a higher accumulation of DAF-28::GFP in coelomocytes (compared to the wild-type background; Figure 7A). Moreover, the ASI specific expression of nhr-69::gfp resulted in a lethargic phenotype with slow, sluggish movement ( Figure 7B) reminiscent of hypoglycemia in mammals [44]. Indeed, the endogenous glucose content was reduced in gpa-4p::nhr-69::gfp expressing animals compared to that of wild type ( Figure 7C). In agreement with lower glucose levels in gpa-4p::nhr-69::gfp animals, the lethargic phenotype was almost completely rescued by exogenous glucose, but not by nonmetabolizable 2-deoxyglucose, suggesting that glucose metabolism, not uptake is impaired in these animals. We conclude that NHR-69 overexpression in the ASI neurons results in hypoglycemia manifested as lethargic behavior, and that NHR-69 acts in the ASI neurons to regulate DAF-28 secretion.

Discussion
The combinatorial interactions of Smad proteins with other DNA-binding transcription factors provide a critical selectivity required to regulate specific subsets of target genes. In this regard, identifying the transcriptional regulators that cooperate with Smad proteins will help delineate the mechanisms underlying signal integration. Here, we describe a molecular crosstalk in C. elegans whereby the R-Smad DAF-8 partners with NHR-69 to regulate insulin signaling. We find that NHR-69 and DAF-8 regulate the exp-2 potassium channel gene, which in turn modulates the secretion of the insulin-like peptide DAF-28. We further demonstrate that NHR-69 affects DAF-28 secretion and glucose homeostasis through its actions in the neuropeptide-secreting ASI neurons that functionally resemble the mammalian pancreas. These data not only implicate the HNF4a-like NHR-69 in insulin signaling, but also suggest a novel mechanism for the previously suggested link between IIS and TGF-b signaling. Intriguingly, Narasimhan et al. recently showed that the phosphatase PDP-1 modulates IIS by acting at the level of the Smad proteins DAF-8 and DAF-14 [13]. Taken together with our findings, this strengthens the notion that there is significantly more crosstalk between IIS and TGF-b signaling than previously appreciated.
In line with this notion, we observed an extended life span and upregulation of the insulin-regulated genes sod-3 and gst-10 in the daf-8 nhr-69 double mutant. We note that the observed longevity likely reflects altered expression of genes other than just sod-3 and gst-10, akin to the observations made in long-lived daf-2 mutants, wherein numerous genes are thought to contribute to longevity [9]. Lastly, we observed a reduced accumulation of ANF::GFP in the coelomocytes of daf-8 nhr-69 double mutants, indicating that neuropeptide secretion is impaired in daf-8 nhr-69 double mutants, representing the likely cause for impaired insulin signaling.
The relationship between TGF-b signaling and IIS may be conserved in evolution. Members of the TGF-b superfamily affect the development and physiology of mammalian pancreatic b-cells, although the results are somewhat conflicting. In one study, Smad3 repressed the insulin gene promoter and Smad3 knockdown enhanced glucose stimulated insulin secretion (GSIS) [45], suggesting a repressive role for TGF-b signaling. However, in isolated rat fetus pancreatic b-cells, TGF-b1 stimulates long-term (48 hours) insulin secretion [46], perhaps through transcriptional changes. Similarly, TGF-b1 and Activin A, another TGF-b family member, stimulate short-term (five minutes) insulin secretion in mouse insulinoma cells by increasing cytoplasmic Ca 2+ concentration [47] [48]. In addition, conditional expression of the inhibitory Smad7 in b-cells resulted in reduced pancreatic insulin levels and in hypoinsulinemia. Lastly, expression of the TGF-b related bone morphogenetic protein 4 (BMP4) and its receptor BMPR1A regulate glucose metabolic genes to enhance GSIS and promote insulin signaling [49]. Our results are more consistent with the latter reports as the daf-8 nhr-69 double mutant exhibited decreased overall insulin signaling.

NHR-69 and DAF-8 control exp-2 expression to regulate DAF-28 secretion
Mechanistically, several lines of evidence suggest that NHR-69 and DAF-8 influence DAF-28 secretion by directly regulating the expression of exp-2, which encodes a voltage-activated (Kv-type) potassium channel. Using ChIP, we found that NHR-69 associated with at least two elements in the exp-2 promoter region that resemble consensus HNF4a binding sites; moreover, this binding was reduced in daf-8 mutants. daf-8 nhr-69 double mutants synergistically induce exp-2 expression, suggesting that exp-2 is repressed by these two factors. Interestingly, using expression profiling, Shaw et al. previously found exp-2 to be induced in several TGF-b Daf-c mutants, including daf-8 mutants [9], although they did not report the extent of the upregulation (we observed a mild induction in daf-8 mutants that was not statistically significant; see Figure 4A). Lastly, we used a heterologous cell culture system to confirm that NHR-69 and DAF-8 collaboratively repress exp-2, with three distinct NHR-69 binding sites being required for full repression.
Together, our results imply that NHR-69 and DAF-8 assemble a complex on the exp-2 promoter that represses exp-2 transcription. Due to the divergent nature of the DNA-binding domain of DAF-8 (only 28% similar to the MH1 domain of human Smad1, and 30% similar to the MH1 domain of Drosophila melanogaster Mad1; Park & Taubert, unpublished), it is difficult to predict whether DAF-8 would bind to the consensus Smad-binding-element (SBE) [50]. Further experiments are required to assess whether DAF-8 directly binds the exp-2 promoter or whether it is tethered to this region through an interaction with NHR-69.
Our data further indicate that exp-2 is critical for insulin signaling downstream of DAF-8 and NHR-69 (but upstream of DAF-2). Firstly, the exp-2(sa26ad1426) mutation, which shortens the life span of wild-type worms, did not shorten the life span of daf-2 mutants. Secondly, exp-2(sa26ad1426) and exp-2(sa26) mutants altered DAF-28 secretion. Thirdly, exp-2(sa26) mutants exhibited increased nuclear localization of DAF-16/FoxO, a typical consequence of reduced DAF-2 activity. Lastly, the exp-2(sa26) mutant showed increased expression of two downstream DAF-2 target genes. Thus, we find evidence for a novel regulatory relationship between insulin secretion and TGF-b signaling. As nhr-69 transcription is itself induced by insulin signaling, a positive Synchronous day-1 adults were video recorded for two minutes individually and worms that stopped and resumed moving more than four times were judged to be lethargic (N = 50). (C) Bars show endogenous glucose content in wild-type and mIs40[gpa-4p::nhr-69::gfp] worms. Synchronous day-1 adults were assayed in four independent biological replicates. Error bars indicate SEM. doi:10.1371/journal.pgen.1002519.g007 regulatory loop to promote larval development and prevent dauer formation is initiated, but only when environmental stimuli activate both TGF-b and insulin signaling.
EXP-2 is a member of the six-transmembrane, voltage-activated potassium channel family, which also controls muscle contraction in the C. elegans pharynx. EXP-2 produces an outward current at the plasma membrane, which terminates the action potential by repolarizing the membrane [36]. As EXP-2 exhibits homology to multiple mammalian Kv-type channels it is difficult to provide a direct parallel to the functions of one mammalian channel. However, several Kv-type channels are known to affect insulin signaling and/or glucose balance [51]. Most prominently, pharmacological inhibition or genetic disruption of Kv2.1 causes increased glucose-stimulated insulin secretion [52] [53]. In the future, it will be interesting to test whether NHR action and TGFb signaling coregulate these mammalian channels.

Neuronal NHR-69 overexpression affects glucose homeostasis and DAF-28 secretion
The expression of nhr-69 overlaps with that of daf-8 in ASI neurons and in the intestine [7], suggesting that the molecular crosstalk between DAF-8 and NHR-69 may occur in these two tissues. Cell ablation analyses demonstrated that the ASI neurons inhibit dauer formation [54], and are required for the expression of the TGF-b ligand DAF-7 and of the insulin-like peptide DAF-28, both of which promote reproductive growth [55] [32]. Because of the latter property, the ASI neurons have been proposed to functionally resemble the insulin-producing pancreatic b-cells [40]. To test whether NHR-69 was functionally important in the ASI neurons, we generated a strain that overexpresses nhr-69 specifically in these neurons. These worms exhibited greater accumulation of DAF-28::GFP, a lethargic phenotype, and reduced glucose levels, reminiscent of a hyperinsulinemic/hypoglycemic phenotype [44]. NHR-69 may achieve these effects by regulating other neuropeptides in addition to DAF-28; in any case our results suggest that the regulation of NHR-69 levels and activity in the ASI neurons is critical to achieve glucose homeostasis in C. elegans. They also support the notion that ASI neurons exhibit characteristics of mammalian pancreatic b-cells [40]. We note, however, that NHR-69 and DAF-28 are also co-expressed in the intestine, a tissue that plays a key role in metabolic regulation in C. elegans and may also perform pancreas related functions.
Our overall model (Figure 8) integrates the observations described above: in favorable conditions, DAF-8 and NHR-69 cooperate to repress exp-2 transcription in the ASI neurons, thus causing sustained DAF-28 secretion, which in turn activates IIS and promotes reproductive growth. This mechanism would act in parallel to the one proposed by Narasimhan et al., whereby DAF-8 promotes IIS and reproductive growth by directly inducing the transcription of agonistic insulin peptide genes such as ins-4 [13]. In harsh conditions, exp-2 transcription is de-repressed as a consequence of reduced DAF-8 phosphorylation and lower nhr-69 expression; increased EXP-2 activity would presumably lead to a rapid reduction in the overall levels of agonistic insulins, thus providing a rapid signal for dauer formation (Figure 8).

NHR-69 functionally resembles pancreatic HNF4a
The C. elegans genome encodes 284 NHRs, most of which derive from an HNF4a-like ancestor. The human HNF4A gene encoding HNF4a is mutated in a subtype of MODY (Maturity Onset Diabetes of the Young) and HNF4a is a central regulator of hepatic and pancreatic gene transcription [37]. MODY patients fail to properly secrete insulin in response to glucose challenge [56] and b-cell-specific knockout of HNF4a in mice results in reduced GSIS [57] [58]. Intriguingly, akin to our discovery of physical and functional DAF-8:NHR-69 interaction, HNF4a and Smad3 interact physically and functionally in transfected human cells [59], although the physiological impact of this interaction remains unclear. Our study suggests that such HNF4a:Smad complexes may affect glucose homeostasis by regulating potassium channel genes in pancreatic b-cells.
The functions of most C. elegans HNF4a-related NHRs remain unknown and no C. elegans NHR has yet been implicated in glucose and/or insulin metabolism. We provide several lines of evidence that NHR-69 performs these particular functions, somewhat resembling pancreatic HNF4a. Most compellingly, we observe hypoglycemia in animals that overexpress NHR-69 specifically in ASI neurons [32] [40]. However, we note that, compared to HNF4a, the defects in nhr-69 single mutants (and in nhr-69 heterozygotes) are relatively mild indicating that possibly other NHRs such as the closely related NHR-64 act redundantly or can compensate for the loss of nhr-69. Given the vast number of HNF4a-related NHRs in C. elegans, other NHRs may also control dauer formation, especially in a sensitized genetic background. As many C. elegans NHRs form heterodimers [60], we speculate that NHR-69 may form heterodimers to implement its regulatory roles.
Human HNF4a targets hundreds of genes in the liver and pancreas to modulate metabolic processes such as lipid biosynthesis, glucose homeostasis and insulin secretion [37]. However, the molecular and physiological functions of individual HNF4a target genes and their combinatorial regulation remains poorly understood. Identifying the functions and the target genes of C. elegans HNF4a-like NHRs can provide insights into the mechanism of related mammalian NHRs. We recently proposed that the functions collectively executed by mammalian HNF4a might be distributed amongst individual C. elegans HNF4a orthologs [17]. For example, NHR-31 regulates the expression of a vacuolar ATPase in the excretory cell in C. elegans [61], possibly hinting at a similar regulatory role of mammalian HNF4a in kidney physiology. Our results suggest that NHR-69 may be a partial functional ortholog of pancreatic HNF4a. Thus, NHR-69 may provide a useful tool to dissect pancreatic HNF4a function, to define in vivo regulators of HNF4a and to delimit additional downstream targets that participate in the regulation of glucose homeostasis.

Transgene construction and transformation
A genomic fragment containing the entire coding region of nhr-69 plus 3 kb upstream of the ATG was amplified. The gfp gene was amplified from pPD95.75 (kindly provided by A. Fire) and used for PCR fusion (see [63]) to generate nhr-69p::nhr-69::gfp, a translational fusion of gfp at the end of nhr-69 exon 9. The gpa-4p::nhr-69::gfp fusion construct was generated by the same method, with a 2.8 kb 59 gpa-4 promoter region [41] being used. The fusions were introduced by microinjection into the adult germ line (45 ng/ml) along with the injection marker pRF4 [rol-6 (su1006)] (9 ng/ml).

Life span analysis and measurement of glucose content
Eggs were prepared by alkaline hypochlorite treatment and synchronized by hatching in M9 buffer. After growth on OP50 at 15uC, animals at mid-L4 stage were transferred to NG agar plates containing 40 mM of 5-fluoro-29-deoxyuridine (FUdR) to prevent progeny production [64]. Life span experiments were performed at 25uC as previously described [65]. We averaged mean life spans from three independent biological replicates. Representative survival curves are shown. We used one-way analysis of variance (ANOVA) for statistical analysis of mean life spans ANOVA [66].
We measured total worm glucose levels with a Glucose Assay Kit (Biovision K606-100). Worms grown on NG agar plates were collected and washed five times in M9 buffer. The worm pellet was resuspended in 500 ml of glucose assay buffer, frozen at 280uC, thawed on ice, and then sonicated (Branson Sonifier 450) five times (output level 3, time 10 seconds) to obtain extracts. After centrifugation at 13,000 rpm at 4uC, the supernatant was used for the glucose assay according to the manufacturer's instructions. We normalized glucose content by protein content in each sample. Statistics were performed using GraphPad Prism 5.

Lethargy assays
Worms were grown synchronously at 20uC from L1 to L4 stage, and then transferred to 25uC. Individual day-1 old adults were elegans. Based on our data, we propose that DAF-2 signaling positively regulates transcription of nhr-69, whereas the TGF-b receptor is known to phosphorylate DAF-8 to activate its function [7]. NHR-69 and DAF-8 cooperatively repress the transcription of the exp-2 Kv channel gene in ASI neurons, which in turn causes sustained secretion of the insulin-like peptide DAF-28, hence activating a feed-forward loop on DAF-2 in favorable conditions. TGF-b and Insulin/IGF-1 signaling crosstalk at the exp-2 promoter through the DAF-8 and NHR-69 transcription factors. Upregulation of exp-2 by the reduction of DAF-8 and NHR-69 activity would attenuate DAF-28 secretion, providing a fast decision for the dauer formation. doi:10.1371/journal.pgen.1002519.g008 video recorded for two minutes, and were judged to be lethargic if they stopped and resumed moving more than four times (N = 50). The plates were tapped manually ten seconds prior to video recording to stimulate movement. For assaying suppression of lethargy by glucose, worms were grown as above and transferred to NG-agar plates supplemented with either 2 mM of D-(+)-Glucose (Sigma, G8270) or 2 mM of 2-deoxy-D-glucose (Sigma D8375).

RNAi and dauer formation and recovery assays
Percent dauer formation was scored visually at 2506 magnification for each genotype grown for two days from hatching at 20uC, 22.5uC, and 25.5uC. To assay dauer recovery, the dauer larvae were induced by growth at 25.5uC for each genotype. 50 dauer larvae were put on NG agar plates seeded with OP50 and incubated at 15uC. Dauer recovery was scored visually monitoring pharyngeal pumping and body size 24 and 48 hrs after transfer. RNAi feeding was performed using clones from the Ahringer library [67] on 1 mM Isopropyl b-D-1-thiogalactopyranoside (IPTG) as described [68].

Chromatin immunoprecipitation (ChIP) and quantitative PCR
For ChIP, mixed-stage populations were subjected to a 1.5% formaldehyde solution for 30 minutes at room temperature to cross-link DNA and proteins [69]. We performed immunoprecipitation by incubating the lysate with either anti-GFP antibody (1:250, Abcam) or IgG at 4uC for 16-18 hours. The precipitates were washed, the crosslinks reversed, and the DNA eluted. 5 ml of each eluate were used for quantitative PCR (qPCR).
For qPCR, 1 mg total RNA from each sample was used for reverse transcription with Superscript II reverse transcriptase (Invitrogen) and oligo dT priming. The resulting first-strand cDNA was analyzed by qPCR on an Applied Biosystems 7500 with BioRad iTaq SYBR green supermix using ROX as a normalization dye. The primers used for qPCR are listed in Table  S2.

Antibodies, immunohistochemistry, and promoter analysis in HEK293 cells
The phosphomimetic pmDAF-8 construct was amplified by PCR with primers dh_295_L and dh_296_R (Table S2). The resulting cDNA was cloned into the pcDNA3.1 (+) Neo vector using BamHI and EcoRI sites. The cDNA for nhr-69::gfp was amplified from mIs39 animals by using dh_312_L and dh_313_R (Table S2). The resulting PCR product was cloned in the pcDNA3.1 (+) Hygro vector using NheI and ApaI sites. The wild-type 2.5 kb exp-2 promoter was PCR amplified using dh_299_L/dh_300_R (Table S2) and subcloned in pTK-GLuc vector (New England BioLabs N8084) using KpnI and BamHI sites to generate pTK-Gluc/exp-2. The mutant exp-2 promoter constructs were generated using a site directed mutagenesis kit (Invitrogen) with pTK-Gluc/exp-2 as a template. The constructs were delivered to HEK293 cells by transient transfection using Lipofectamine (Invitrogen) according to the manufacturer's manual. The luciferase assays were performed using a BioLux Gaussia Luciferase Flex Assay Kit (New England BioLabs E3308L), and luminescence was measured in a Tecan M200 plate reader.

Microscopy
We used a Zeiss Axioscope equipped with a QImaging camera (RETIGA 2000R) for differential interference contrast (DIC) and fluorescence microscopy. ImageJ software was used to estimate the intensity of the GFP signal in coelomocytes. For DiI staining, cultures were synchronized by hatching purified eggs into M9 buffer, grown on NG agar plates until the L2 stage, washed in M9 buffer, and DiI stained, as described [70]. Fluorescent signals generated by immunostaining were visualized and captured using an Olympus FLUOVIEW FV10i confocal laser-scanning microscope system. Images from different color channels were colorized and merged using ImageJ software. Figure S1 Anti-DAF-8 antibody specifically detects endogenous DAF-8. Mixed-stage worms were washed three times with M9 buffer and SDS sample buffer was added to the worm pellet. Total protein extracts were obtained by boiling. 40 mg of total protein was loaded into each well. The blot was probed with anti-DAF-8 antibody (1:1,000). (TIF) Figure S2 nhr-69p::nhr-69::gfp and daf-8p::daf-8::gfp are expressed in the ASI neurons. Dorsal view of expression in ASI neuron (arrow in lower panel). The same animals were stained with DiI for ciliated neurons in the head (arrows in middle panel). (TIF) Figure S3 RNAi against nhr-69 enhances dauer formation. The bar graphs show the percentage of wild type or eri-1(mg366); sdf-9(m708) worms that formed dauer larvae when grown on HT115 bacteria harboring control RNAi plasmid (No), or gfp, akt-1, or nhr-69 RNAi plasmids. Worms were initially grown on OP50 at 15uC, transferred to lay eggs on plates with bacteria harboring the indicated RNAi construct, and then removed. The F1 progeny were grown to the L4 stage at 15uC; then, three F1 L4 larvae were transferred to fresh RNAi plates at 25uC, and removed as young adults after they had laid 30-40 eggs. Dauer formation in these F2 worms was then scored visually. akt-1 RNAi was used as the positive control because we previously found that it promotes dauer formation in the eri-1; sdf-9 background. (TIF) Figure S4 Accumulation of neuropeptide-GFP fusion proteins in coelomocytes (CC) and quantification of ANF::GFP expression. (A) Quantification of ANF::GFP in coelomocytes of wild type and daf-8(m85) nhr-69(ok1926) worms (black, p,0.001) indicates that the mutant strain is deficient in neuropeptide secretion. The daf-8 (grey) and nhr-69 single mutants (hatched) showed 10.2% and 15.4% reduction in the GFP intensity, respectively. (B) Quantification of coelomocyte ssGFP in wild type and daf-8(m85) nhr-69(ok1926) worms (p.0.05) shows that the difference seen in (A) is not due to a deficiency in coelomocyte uptake. The intensity in either daf-8 (grey) or nhr-69 (hatched) single mutant was comparable to that of wild type (p.0.05). (C) qPCR for gfp mRNA from animals expressing aex-3p::ANF::GFP in wild type and daf-8 nhr-69, daf-8 and nhr-69 (p.0.05) shows that the expression of intestinal peptides is similar in all strains. Bars indicate SEM for four independent experiments. (TIF) Figure S5 Coelomocyte DAF-28::GFP accumulation and daf-28 expression in exp-2 mutants. (A) Quantification of coelomocyte DAF-28::GFP in wild type worms, loss-of-function exp-2(sa26ad1426) and gain-of-function exp-2(sa26) mutants (p,0.001), and worms expressing exp-2(sa26) in ASI-specific (red bar, p.0.05 compared to exp-2(sa26) mutant) or pharyngeal-specific fashion in the exp-2(sa26ad1426) mutant background (green bar, p = 0.7966 compared to exp-2(sa26ad1426) mutant). (B) qPCR quantification of daf-28 mRNA levels in wild-type, loss-of-function exp-2(sa26ad1426), and gain-of-function exp-2(sa26) worms (p.0.05). Red and green bars represent mRNA levels in worms expressing exp-2 in ASI-specific or pharyngeal-specific fashion, respectively. Error bars indicate SEM. (TIF)