The P-type ATPase CATP-1 is a novel regulator of C. elegansdevelopmental timing that acts independently of its predicted pump function

During embryonic and postembryonic stages, metazoans have to synchronize the development of a large number of cells, tissues and organs. As an example,the symmetrical organs of Bilateria usually grow and develop at the same speed. Recent light has been shed on the molecular mechanisms responsible for such regulation, with the finding that retinoic acid is essential in vertebrates to synchronize the somitogenesis clock between the left and right developing mesoderm (reviewed by Brent,2005). However, the mechanisms responsible for synchronization during most steps of metazoan development are still unknown(Rougvie, 2005).

The nematode Caenorhabditis elegans provides a genetically tractable model system to understand how animals synchronize postembryonic developmental events. After hatching, the C. elegans larva proceeds to adulthood through a discontinuous development with four larval stages (L1 to L4), each terminated by a molt. This molt cycle is tightly coordinated with postembryonic cellular divisions. Because the cell lineage is invariant in C. elegans, it is possible to monitor developmental events at a single cell resolution (Sulston and Horvitz, 1977). However, the mechanisms that control and synchronize the timing of cell divisions(Kipreos, 2005) and molts(Frand et al., 2005) are still poorly understood. We recently described a novel paradigm to study developmental synchronization in C. elegans by using the nicotinic agonist DMPP (dimethylphenylpiperazinium)(Ruaud and Bessereau, 2006). We demonstrated that illegitimate activation of nicotinic acetylcholine receptors (nAChRs) by DMPP during the second larval stage induced a lethal heterochronic phenotype by disconnecting developmental speed from the molting timer, hence resulting in deadly exposure of a defective cuticle to the surrounding environment at the subsequent molt. Environmental conditions that delay the second to third larval molt suppress DMPP-induced lethality by enabling resynchronization of the delayed development with the molting time. The primary target of DMPP is likely neuronal as loss of expression of the nAChR subunit UNC-63 in neurons partially protects the animals from DMPP toxicity. Using a forward genetic screen, we further demonstrated that the nuclear hormone receptor DAF-12 is required to implement the developmental effects of DMPP. These results uncovered two independent pathways: one controlling the timing of C. elegans molt and one regulating the timing of other developmental events. Moreover, they likely define a previously undescribed neuroendocrine pathway that is able to modulate the timing of developmental events in response to environmental parameters.

Molecular players involved in this novel pathway as well as environmental stimuli modulating DMPP sensitivity overlap with key components of the dauer pathway, a genetic network controlling entry into a facultative diapause stage. Under favorable conditions, C. elegans goes from hatching to reproductive adulthood in 3 days. If L1 larvae are exposed to adverse conditions including limited food, high temperature and high population density, animals can enter a facultative L3 diapause stage called the dauer larva (Cassada and Russell,1975). Dauer larvae survive for several months without feeding and are able to resume development to fertile adults when conditions are favorable again. A complex neuroendocrine network involving a TGFβ, an insulin-like and a nuclear hormone receptor (NHR) pathway controls dauer entry (reviewed by Beckstead and Thummel, 2006; Riddle and Albert, 1997). Insulin/IGF and TGFβ peptides are synthesized and released from sensory neurons in response to favorable stimuli(Li et al., 2003; Ren et al., 1996). Under these conditions, DAF-7/TGFβ signals through its receptor to inactivate DAF-3/SMAD and DAF-5/SNO (Patterson and Padgett, 2000), and insulin-like agonists stimulate the DAF-2-insulin-like receptor, initiating a cascade that inactivates DAF-16/FOXO by cytoplasmic segregation (Henderson and Johnson, 2001; Kimura et al.,1997; Ogg et al.,1997), thus allowing reproductive development. Under adverse conditions, a DAF-3-DAF-5 complex and nuclear DAF-16 specify diapause. In addition to their direct effect on specific transcription factors, the DAF-7/TGFβ and DAF-2/InsR pathways also act through lipophilic hormone signaling, mostly by regulating the expression of DAF-9, a cytochrome P450 involved in dafachronic acid synthesis(Gerisch and Antebi, 2004; Gerisch et al., 2001; Jia et al., 2002). Dafachronic acids are steroid hormones functioning as ligands for the nuclear receptor DAF-12 (Antebi et al., 2000; Motola et al., 2006). Under favorable conditions, DAF-7/TGFβ and DAF-2/InsR signaling pathways are active. This promotes DAF-9/CYP expression and subsequent production of steroid hormones. The DAF-12 NHR is mostly liganded and promotes reproductive growth. Under adverse conditions, DAF-7/TGFβ and DAF-2/InsR signaling are poorly active and DAF-9 expression is low. Consequently, dafachronic acid levels are low, and unliganded DAF-12 triggers dauer diapause entry. We previously showed that the liganded form of DAF-12 is specifically required to implement nicotinic agonist toxicity during C. elegans larval development (Ruaud and Bessereau,2006).

In this study, we conducted a forward genetic screen to identify additional molecular players involved in synchronizing the molt cycle with other developmental events during the C. elegans L2 stage. We identified the cation-transporting ATPase CATP-1 as being a novel factor required for DMPP sensitivity. catp-1 functions in parallel with unc-63and daf-12 for larval developmental timing and interacts with the insulin/IGF and Ras-MAPK pathways to control several postembryonic developmental events. Interestingly, we demonstrated that CATP-1 mostly functions independently of its predicted pump function to control C. elegans postembryonic developmental timing by using ATPase-dead mutants. These results are in line with previous data suggesting that Na⁺/K⁺-ATPases could act as molecular scaffolds to modulate signaling pathways in mammalian cells and provide the first in-vivo demonstration for a transport-independent function of a P-type ATPase.

Unless noted otherwise, worms were cultured at 20°C on NG agar plates inoculated with Escherichia coli OP50(Sulston and Hodgkin, 1988). Synchronous developing populations were obtained by allowing young gravid adults to lay eggs for 30 minutes. Eggs were then collected and transferred onto new plates. The wild strain N2, the following mutant alleles and transgenic markers were used: age-1(mg44) I, catp-1(kr17) I, daf-2(e1368,e1370, e1391, m41, m577, m596) III, daf-7(e1372) III, daf-9(dh6) X,daf-11(m47) V, daf-12(rh61rh411) X, daf-16(mgDf50) I, ksr-1(n2526) X,let-60(n1046) IV, lin-15(n765ts) X, mek-2(n1989) I, mpk-1(ku1) I, sid-1(qt2)V, unc-63(kr13) I, krEx36 and krEx49 [Pcatp-1::catp-1(C. briggsae);sur-5-GFP], krEx45 and krEx46 [Pcatp-1(C. briggsae)::catp-1(C. elegans); sur-5-GFP], krEx50 and krEx51 [Pcatp-1(C. briggsae)::GFP;lin-15(+)], krEx251 and krEx252 [Pcatp-1(C. briggsae)::catp-1D409E;Pmyo-3::GFP], krEx253 and krEx254 [Pcatp-1(C. briggsae)::catp-1R669Q;Pmyo-3::GFP], krEx264 and krEx265 [Pdpy-7::catp-1(dsRNA); Prab-3::GFP;Pmyo-3::GFP], ccIs4251 [Pmyo-3::Ngfp-lacZ; Pmyo-3::Mtgfp].

Mutants of MAPK pathways that tested DMPP sensitive were: jkk-1(km2) X,jnk-1(gk7) IV, mek-1(ks54) X, nsy-1(ag3) II, sek-1(ag1, km4) X.

1,1-Dimethyl-4-phenylpiperazinium (DMPP) (Sigma) was dissolved in water and added to 55°C-equilibrated NG agar at a concentration of 0.75 mM, unless noted otherwise, just before plates were poured. Gravid adult worms were allowed to lay eggs for several hours on standard plates. Eggs were then carefully transferred onto DMPP-containing plates and counted. Surviving L4,adults and dauer larvae were scored after 3 days of development (20°C). Dauer pheromone was purified as described(Golden and Riddle, 1984) and added to streptomycin-containing (Sigma, 50 μg/mL) plates when mentioned.

N2 worms were mutagenized by germline mobilization of the Drosophila transposon Mos1(Williams et al., 2005). Young-adult F1 worms were transferred onto 0.75 mM DMPP plates and allowed to lay eggs for 1 day. Three days later, plates were screened for healthy living adult animals. In EN17 catp-1(kr17::Mos1), a mutagenic Mos1insertion localizes at position 14,438,562 of chromosome I by inverse PCR(WormBase web site, http://www.wormbase.org,release WS160, date 07/2006).

An 8.1 kb genomic fragment corresponding to C. elegans Y105E8A.12 was PCR amplified from C. briggsae AF16 genomic DNA using Expand Long Range PCR (Roche) (primers 5′-CACATCATCGCATCATCGTC-3′ and 5′-GATGAGTCGTCTTAGTAGTG-3′).

A 2.9 kb C. briggsae catp-1 promoter fragment was PCR amplified from C. briggsae genomic DNA using a Taq/Pfu mix and primers 5′-GCCTGCAGCACATCATCGCATCATCGTC-3′ and 5′-GCGGATCCCTTTCAGTGTATTCTTCTGTTTC-3′. This PCR fragment was cloned in pPD115.62 using restriction sites PstI and BamHI.

catp-1 cDNA was cloned by RT-PCR using primers 5′-GCGGTACCACCGGTTTACTGTATGACTCGGAAACC-3′ and 5′-GCGAATTCACCATTTGATAAGGCGAACA-3′ and inserted downstream of the catp-1 C. briggsae promoter in pAF29.

pAF31 was mutagenized using the QuikChange Site-Directed Mutagenesis kit(Stratagene) using primer 5′-CCACCGTAATCTGCGCAGAGAAATCAGGCACTC-3′. A 1 kb-long region containing the mutagenized site was sequenced and a pool of four independently mutagenized plasmids was injected.

A 1637 bp EcoRV-EcoRI fragment from pAF31 was cloned in pBS SK and mutagenized using primer 5′-GCTCGGCAACGAGGGTCGACAAGTGATCGCCTTTGC-3′. The mutagenized region was sequenced and cloned back into pAF31.

An 894 bp PstI-KpnI dpy-7 promoter fragment(Gilleard et al., 1997) was subcloned into pPD115.62 (PstI-BamHI) to generate Pdpy-7::GFP. The 5′ fragment of catp-1 cDNA(5′-GCGGTACCACCGGTTTACTGTATGACTCGGAAACC-3′ to 5′-GACCTGGCAGCGACGGATTGA-3′) was then inserted into Pdpy-7::GFP using AgeI and EcoRI. A 5′ fragment(5′-GAATTCGCCCTTAAAGACTTCGTTCGTCGA-3′ to 5′-TATCGGTGTTCGACGCGTGGATCCCCCGGG-3′) of the cDNA was finally cloned backward into the previous plasmid to create a 620 bp hairpin.

Transformation was performed by microinjection of plasmid DNA into the gonad (Mello et al., 1991). catp-1(kr17) worms were injected with a DNA mixture containing a C. briggsae genomic fragment and pTG96 (sur-5-GFP) (100 ng/μL) or pAF31 (10 ng/μL) and pTG96 (90 ng/μL) for rescue experiments. The pump-dead mutant plasmids pAF88 and pAF90 were injected in catp-1(kr17) at 10 ng/μL with pPD115.62 (Pmyo-3::GFP) (5 ng/μL) as a co-injection marker and 1 kb+ DNA ladder (INVITROGEN) (85 ng/μL). For tissue-specific RNAi, pAF92 was injected in ccIs4251;sid-1(qt2) hermaphrodites at 0.1 ng/μL together with pHU4(Prab-3::GFP, 20 ng/μL), pPD115.62 (5 ng/μL) and 1 kb+ DNA ladder (75 ng/μL). pAF29 was injected in lin-15(n765ts) at 20 ng/μL with EKL15 (lin-15(+)) (80 ng/μL) as a co-injection marker.

Light microscopy was performed as previously described(Ruaud and Bessereau,2006).

Stimulation of nicotinic receptors by the agonist DMPP during the second larval stage of C. elegans development causes a lethal heterochronic phenotype by slowing cellular developmental events but not molting. Consequently, animals die at the L2/L3 molt. To identify the molecules required to implement the effect of DMPP, we performed a forward genetic screen for mutants that can develop on the drug. An insertional mutagenesis technique based on the mobilization of the drosophila transposon Mos1in the C. elegans germ line(Bessereau et al., 2001; Williams et al., 2005) was used to facilitate rapid identification of the mutated genes. Among seven resistant mutants (Ruaud and Bessereau,2006), we identified one strain containing the mutant allele kr17 that was strongly resistant to DMPP compared with wild type(Fig. 1A). kr17homozygous mutants are healthy and do not display any gross behavioral nor morphological abnormalities by DIC inspection.

To identify the gene carrying the kr17 mutation, we performed inverse PCR on genomic DNA of the mutant strain and detected a Mos1insertion in the predicted open reading frame Y105E8A.12(Fig. 1B). This gene was named catp-1 because it codes for a cation-transporting ATPase of the P-type family (see below). The identification of catp-1 as a novel DMPP resistance gene was confirmed by three sets of experiments. First, the kr17 mutation genetically mapped to the right end of chromosome I where Y105E8A.12 is located. Second, RNAi against catp-1 phenocopied the DMPP resistance of kr17 mutants (data not shown). Third, catp-1(kr17::Mos1) was rescued by expression of the CATP-1 protein. Due to the presence of multiple repeated sequences in the Y105E8A.12 genomic locus, we could not PCR amplify a rescuing genomic DNA fragment from the wild-type N2 strain. However, in the closely related nematode species C. briggsae, the genomic region containing the catp-1 ortholog has a much simpler organization(Fig. 1B). Based on the WABA software for cross-species alignment (Kent and Zahler, 2000), the protein encoded by the C. briggsae catp-1 ortholog was predicted to share 88% identity and 92% similarity with C. elegans CATP-1 (see Fig. S1A in the supplementary material). As cross-species rescue between C. elegans and C. briggsaehas already been achieved for several Caenorhabditis genes(Aronoff et al., 2001; Maduro and Pilgrim, 1996; Thacker et al., 1999), we PCR amplified C. briggsae genomic DNA to obtain a fragment spanning the entire catp-1 region. This fragment rescued the DMPP resistance of catp-1(kr17) (Fig. 1C). Next, we cloned C. elegans catp-1 cDNAs by RT-PCR and identified 16 exons, which encode a 1121-amino acid-long protein (CATP-1a)with a start ATG codon at position 14,451,796 on chromosome I(Fig. 1B). We also detected a short transcript lacking the 3′ part of exon 16 (CATP-1b). It encodes a truncated product compared with other proteins of the same family and was not used for further analysis (see Fig. S1B in the supplementary material). The C. elegans catp-1 cDNA expressed under the control of the C. briggsae catp-1 promoter rescued the DMPP resistance of catp-1(kr17::Mos1) (Fig. 1C). Taken together, these data demonstrate that catp-1is required to implement DMPP toxicity during C. elegans larval development.

Based on sequence homology, catp-1 was predicted to encode a P-type ATPase (Fig. 2)(Kuhlbrandt, 2004; Okamura et al., 2003). P-type ATPases form a large family of diverse membrane proteins that actively transport charged substrates such as cations and phospholipids across biological membranes. P-type ATPases possess ten hydrophobic membrane-spanning helices (M1-M10), and highly conserved cytoplasmic domains inserted between helices M2 and M3 and between M4 and M5, an organization found in CATP-1(Fig. 2B, see Fig. S2 in the supplementary material). They are biochemically characterized by the presence of an acid-stable phosphorylated aspartate residue that forms during the pumping cycle. This phosphorylable residue is easily identified in CATP-1 at position 409 (Fig. 2C). In catp-1(kr17::Mos1) mutants, the transposon insertion introduces a premature STOP codon before helix M8 (see Fig. S1B in the supplementary material). As this mutation is fully recessive and is phenocopied by catp-1 RNAi, catp-1 truncation likely causes a loss of function of the protein. Therefore kr17 is probably a null allele for CATP-1 enzymatic activity. However, we cannot exclude that the protein is still being made and retains the ability to interact with cellular partners.

Based on sequence similarity, the P-type ATPase family is divided into five branches, referred to as types I-V, among which up to ten different subtypes or classes can be distinguished, each subtype being specific for a particular substrate ion (reviewed by Kuhlbrandt,2004). Based on sequence analysis, CATP-1 can be assigned to theα-subunits of the type IIC ATPase subgroup, which contains the Na⁺/K⁺- and H⁺/K⁺-ATPases(Fig. 2A). CATP-1 shares 32%identity and 49% similarity with the human Na⁺/K⁺-ATPase NK1 and is more distantly related to the human sarcoplasmic reticulum Ca²⁺-ATPase (25% identity and 43% similarity). Na⁺/K⁺ and H⁺/K⁺ P-type ATPases have been extensively characterized in vertebrates(Jorgensen et al., 2003; Kaplan, 2002). Both vertebrate Na⁺/K⁺- and H⁺/K⁺-ATPases form functional heterodimers consisting of a larger α-subunit (110 kDa) and a highly glycosylated β-subunit (35 kDa core molecular mass)(Moller et al., 1996). Theα-subunits of both enzymes probably incorporate all of the structural features required for enzymatic activity, whereas the β-subunits are necessary to ensure both the structural integrity of the dimeric protein complex as well as its proper delivery to the plasma membrane(Geering, 1990; Gottardi and Caplan, 1993). Type IIC ATPases have been subdivided into three subclasses based on their ionic specificity, pharmacological properties and expression profile: (1) the Na⁺/K⁺-ATPases, which are expressed in virtually all cells where they extrude Na⁺ in exchange for K⁺ ions and are sensitive to the antagonist ouabain, (2) the gastric H⁺/K⁺-ATPases, which are expressed on the apical side of stomach cells where they exchange H⁺ for K⁺ ions and are ouabain-insensitive, and (3) the so-called nongastric H⁺/K⁺-ATPases can exchange both H⁺ and Na⁺ against K⁺ and are ouabain-sensititive(Caplan, 1997; Jaisser and Beggah, 1999; Kuhlbrandt, 2004)(Fig. 2). The C. elegans genome encodes five α and three β subunits of the type IIC ATPase family (Okamura et al.,2003). The α-subunit EAT-6 clearly falls into the Na⁺/K⁺-ATPases group by sequence homology and functional properties (Davis et al.,1995). Other C. elegans class IIC α-subunits(C01G12.8, C09H5.2, C02E7.1 and CATP-1) do not fall into any of the vertebrate classes (Fig. 2). Out of these,C09H5.2, C02E7.1 and CATP-1 lack the ouabain-binding site and the motifs correlated with α/β assembly, and show little conservation in the amino acids associated with ion specificity (see Fig. S2 supplementary material) (Okamura et al.,2003). These characteristics may reflect an ancestral form of the Na⁺/K⁺, H⁺/K⁺-ATPase, and preclude a more precise prediction of CATP-1 transport specificity.

To get further insights into catp-1 function, we analyzed its tissue expression pattern. GFP was placed under the control of the promoter sequence previously used to express CATP-1 for mutant rescue. GFP was detected in epidermal cells including the head epidermal cells hyp1 to hyp5, the hyp7 syncytium, the tail epidermal cells hyp8 to hyp11, and the ventral Pn.p cells. GFP expression was also detected in the excretory duct cell(Fig. 1D-F). To confirm that catp-1 was required in the epidermal cells, we expressed CATP-1 under the control of the epidermal promoter Pdpy-7(Gilleard et al., 1997). Unfortunately, this construct was extremely toxic in vivo, maybe due to improper expression during development or overexpression levels, and was unable to rescue catp-1 mutants. Therefore, we used RNA interference to selectively inhibit CATP-1 synthesis in epidermal cells by expressing a hairpin catp-1 dsRNA driven by the Pdpy-7 promoter. Experiments were performed in a sid-1 mutant background to prevent systemic spreading of RNAi (Winston et al., 2002). Epidermal repression of catp-1 conferred partial resistance to DMPP (Fig. 1H). Altogether, these data suggest that the P-type ATPase CATP-1 functions in the epidermis to implement DMPP toxicity.

Apart from regulating ion gradients through their pump activity, recent reports suggest that Na⁺/K⁺-ATPases might function to modulate intracellular signaling pathways. Specifically, reducing Na⁺/K⁺-ATPase activity alters signaling through the Src,FAK and MAPK pathways by direct protein-protein interactions between the pump and the signaling kinase Src (Tian et al.,2006) (reviewed by Xie and Askari, 2002). Signaling properties of the Na⁺/K⁺-ATPase are largely independent from its catalytic function as a pump-dead mutant can rescue the signaling activity of Na⁺/K⁺-ATPase-depleted cells(Liang et al., 2006). To test if CATP-1 might also function independently of its pump activity during C. elegans postembryonic development, we engineered two different mutations that had been demonstrated to abolish the activity of the sodium pump. The D409E mutation removes the conserved aspartate which is phosphorylated during the catalytic cycle (Fig. 2C)(Liang et al., 2006; Ohtsubo et al., 1990). The R669Q mutation disrupts ATP binding and, consequently, the pump activity(Fig. 2D)(Jacobsen et al., 2002). CATP-1 engineered point mutants were expressed in catp-1(kr17). Each of these pump-dead proteins was able to partially rescue DMPP resistance of the mutant animals (Fig. 2E). These results indicate that CATP-1 functions, at least in part, independently of its ATPase function during C. elegans postembryonic development.

Chronic exposure to DMPP is lethal by slowing development at the second larval stage without affecting the molt timing(Ruaud and Bessereau, 2006). Insensitivity to DMPP was achieved (1) when DMPP was unable to induce a lineage delay, as demonstrated in the daf-12(0) mutant, and (2) when development was slowed down at the L2 larval stage, thus enabling resynchronization of molting with a slowed development, as for animals entering a L2 diapause. To identify which of these mechanisms might account for the DMPP resistance of catp-1(kr17) mutants, we monitored the molt cycle and L2 development. First, we observed that the duration of the L2 stage was extended in mutants to 145% of wild type(Fig. 3A) and L2 cell divisions were accordingly postponed (Fig. 3B). This developmental delay was specific for the L2 stage, as the duration of the other larval stages was not altered. Second, exposure of catp-1(kr17) mutants to DMPP did not cause any further developmental delays (Fig. 3B). Together,these results suggest that catp-1 functions to tune both the developmental and molting timers during wild-type L2 development.

In a previous study, we showed that the nAChR subunit UNC-63(Culetto et al., 2004) and the nuclear hormone receptor DAF-12 (Antebi et al., 1998; Antebi et al.,2000) are required to implement DMPP toxicity. UNC-63 might be part of a DMPP receptor, whereas DAF-12 is thought to provide a permissive activity to implement DMPP signaling. These two genes interact in a non-linear pathway (Ruaud and Bessereau,2006). To further understand the function of catp-1during the developmental response triggered by exposure to DMPP, we tested genetic interactions between catp-1(kr17::Mos1) and null mutations of unc-63 and daf-12 that affect sensitivity to DMPP. As both catp-1(kr17) and daf-12 null mutants show a strong DMPP resistance (Ruaud and Bessereau,2006) (Fig. 4), we used a high drug concentration (1 mM) that kills a fraction of catp-1and daf-12 mutants in order to unmask possible synergistic effects. In these conditions, we found that both double mutants [unc-63(kr13)catp-1(kr17) and catp-1(kr17); daf-12(rh61rh411)] were more resistant than any of the single mutants(Fig. 4), suggesting that catp-1 acts in parallel to both daf-12 and unc-63.

Previously identified DMPP-resistant mutants such as daf-9 and daf-12 do not properly take the decision to enter dauer, a facultative L3 diapause stage chosen under adverse environmental conditions(Antebi et al., 1998; Gerisch et al., 2001; Jia et al., 2002). Does this property also apply to catp-1 mutants? Under replete culture conditions, catp-1(kr17) animals never form dauer larvae and are therefore not dauer formation constitutive (Daf-c). We tested whether catp-1 mutants were dauer defective (Daf-d) using partially purified dauer pheromone (Golden and Riddle,1984). catp-1(kr17) mutants do not form morphologically typical dauer larvae on high pheromone. Instead, we observed a high proportion of short dauers that had not molted from their L2 cuticle(Fig. 5A). These animals were SDS-resistant, suggesting that occlusion of the buccal cavity typical of dauer morphogenesis had been completed. Interestingly, these abnormal dauer larvae unable to molt are highly reminiscent of worms grown on lophenol, a methylated derivative of cholesterol (Matyash et al.,2004). In these conditions, it is hypothesized that hormones required for molting and for non-dauer development, derived from non-methylated cholesterol, are not synthesized. Our result could implicate CATP-1 in the control of the L2/L3d molt.

The decision to enter dauer diapause is controlled by a complex genetic network. Schematically, signals from the DAF-2/InsR (insulin receptor) and the DAF-7/TGFβ pathways are integrated at the level of DAF-12/NHR via the production of lipophilic hormones (Fig. 5B). To place catp1 in this network, we performed epistasis experiments between catp-1(kr17) and dauer constitutive mutants of the different pathways. Among all mutant combinations tested, we specifically detected genetic interactions between catp-1 and daf-2 (Fig. 5C). catp-1(kr17) fully suppressed the Daf-c phenotype of daf-2(m41) mutants and caused abnormal dauer morphogenesis of daf-2(m596) and daf-2(e1391) mutants in epidermal tissue(Fig. 5D and data not shown):the dauer alae were abnormal and animals did not elongate properly. Because the pharynx was constricted as in normal daf-2 dauers, it appeared squeezed in the head. These results identify catp-1 as a novel allele-specific suppressor of the daf-2/InsR dauer constitutive phenotype and suggest that catp-1 could function downstream or in parallel with daf-2/InsR to control dauer formation and morphogenesis.

To further investigate the interaction between catp-1 and daf-2/InsR signaling, we examined the DMPP resistance of mutants of daf-2/InsR and its main downstream effector, the transcription factor daf-16/FOXO (Lin et al.,1997; Ogg et al.,1997). daf-2(e1370) mutants are strongly DMPP resistant and this phenotype is fully suppressed by a daf-16(mgDf50) mutant(Fig. 6). By contrast, a daf-16(mgDf50) catp-1(kr17) double mutant is as resistant as catp-1(kr17) alone (Fig. 6), suggesting that catp-1 functions downstream or in parallel of daf-16 to mediate DMPP sensitivity and probably dauer formation. Several genetic differences distinguish CATP-1 and DAF-16: (1) daf-16(0) suppresses constitutive dauer formation and increased life span of all daf-2(lf) alleles(Kenyon et al., 1993; Riddle et al., 1981) whereas catp-1(kr17) does not (Fig. 5C and see Fig. S3 in the supplementary material); (2) daf-16(0) suppresses age-1/PI3K whereas catp-1(kr17) does not (Fig. 5C); and (3) daf-16(0) is DMPP sensitive whereas catp-1(kr17) is DMPP resistant(Fig. 6). These data are not in favor of catp-1 acting downstream of daf-16 but rather support a model where catp-1 and daf-16 would act in parallel to differentially modulate signaling through activated DAF-2/InsR.

Analysis of dauer formation and aging in C. elegans has defined a linear DAF-2/InsR signaling pathway regulating DAF-16/FOXO transcriptional activity. However, two recent studies have unmasked functions of the Ras-MAPK pathway in DAF-2-dependent regulation of development and aging in C. elegans (Hopper, 2006; Nanji et al., 2005). An activated Ras mutation, let-60(n1046gf), which affects the GTPase domain of Ras (Han and Sternberg,1990) and causes an extended life-time of LET-60 in its active GTP-bound conformation (Barbacid,1987; Beitel et al.,1990; Polakis and McCormick,1993) was demonstrated to weakly suppress the constitutive dauer formation of some daf-2 mutants. However, the interaction between daf-2(m41) and let-60(n1046gf) had not been analyzed previously. We observed that dauer entry of daf-2(m41) was partially suppressed by let-60(n1046gf), as opposed to daf-2(e1370)which was unaffected by the activation of Ras(Fig. 7A). These results further support the role of the Ras pathway in DAF-2/InsR signaling during larval development.

Because catp-1 appeared to play a role in insulin signaling using a branch parallel to daf-16, we speculated that it might function by modulating a DAF-2-dependent Ras-MAPK pathway. To test this hypothesis, we first assayed DMPP resistance of mutants in the Ras-MAPK pathway. A loss-of-function mutant of mek-2, the MAPKK of the ERK-MAPK pathway(Wu et al., 1995), was partially DMPP-resistant (Fig. 7B). We could not test stronger mutants of the pathway because of their rod-like L1 lethality (Sundaram,2006). The weak viable mutant of mpk-1(Lackner et al., 1994), which encodes the C. elegans homolog of the ERK-MAPK, was not DMPP resistant. However, this allele has been previously shown to induce weaker phenotypes than the weakly resistant mek-2 allele(Nicholas and Hodgkin, 2004). In addition, a null mutant of the adaptor protein-encoding gene ksr-1(Kornfeld et al., 1995; Sundaram and Han, 1995) was also resistant to DMPP (Fig. 7B). The very weak DMPP resistance of ksr-1(n2526) could result from redundancy between KSR-1 and a second KSR protein, KSR-2, as previously demonstrated for a subset of MAPK-controlled developmental decisions (Ohmachi et al.,2002). By contrast with the ERK-MAPK pathway, null mutants of the p38 and JNK MAPK pathways (Sakaguchi et al., 2004) were sensitive to DMPP (data not shown). Altogether,these results suggested that the C. elegans ERK-MAPK pathway is required to implement DMPP toxicity.

To test if CATP-1 was interacting with the MAPK pathway, we used the activated Ras mutation. let-60(n1046gf) individuals were DMPP sensitive. The activated Ras partially suppressed catp-1(kr17) DMPP resistance (Fig. 7B), hence suggesting that catp-1 and let-60 might act in the same pathway to implement DMPP toxicity. To test if Ras signaling is indeed involved in DAF-2-dependent sensitivity to DMPP, we introduced the let-60(n1046gf) in daf-2 mutants. Interestingly, activation of Ras only weakly suppresses the DMPP resistance of the daf-2(e1370)mutants whereas it fully suppresses the resistance of daf-2(m41)mutants (Fig. 7B). Altogether,these results suggest that both Ras-dependent and Ras-independent pathways are involved in DAF-2/InsR signaling during development modulation by DMPP-stimulation of AchRs. Moreover, CATP-1 would mostly interact with DAF-2/InsR signaling by modulating a Ras-dependent pathway.

In a forward genetic screen for C. elegans mutants able to develop in the presence of the nicotinic agonist DMPP, we identified catp-1,a gene coding for an epidermal cation-transporting ATPase. catp-1 is involved in the control of postembryonic developmental timing and in the decision to enter the dauer diapause stage. These features likely result from interactions with daf-2/InsR signaling, most probably through the interaction between CATP-1 and the Ras-MAPK pathway. By rescuing catp-1(kr17) mutants with ATPase-dead versions of CATP-1, we could demonstrate that the developmental functions of CATP-1 are largely independent of its putative cation transport function but might rely on scaffolding properties provided by this transmembrane protein, as recently proposed for other members of the P-type ATPase family.

We previously demonstrated that illegitimate activation of nAChRs by DMPP during the second larval stage induced a lethal heterochronic phenotype by slowing developmental speed without affecting the molting timer, hence resulting in deadly exposure of a defective cuticle to the surrounding environment at the subsequent molt (Ruaud and Bessereau, 2006). Most probably, the defective cuticle exposed at the L2/L3 molt does not fulfill its diffusion barrier function and animals dissolve rapidly in a way reminiscent of an osmotic shock. Different parameters can be modified by environmental conditions or genetic mutations to cause DMPP resistance. Some mutants, such as daf-12(0), do not slow development in the presence of DMPP and thus do not desynchronize. Alternatively, animals reared on restricted amounts of food or mutants such as eat-6(lf) possess extended intermolt periods, hence enabling the compensation of developmental delay. Finally, some osmotic-stress-resistant mutants like osm-7 (Wheeler and Thomas, 2006) might survive even with a defective cuticle (A. F. Ruaud and J. L. Bessereau, unpublished). At first glance, catp-1might function in osmoregulation, osmotic stress sensing or response to osmotic stress as CATP-1 is expressed in the epidermis and has a predicted ion transport function. However, we do not favor this hypothesis. First, catp-1 mutants are as sensitive as wild-type animals to high osmolarity on 800 mM sodium acetate, in contrast to osm-7(n1515)(data not shown). Second, catp-1 mutants do not delay their development in the presence of DMPP. Under physiological conditions, catp-1(kr17) animals develop at normal rate during all larval stages,except at the L2 stage which is considerably extended compared with wild type. L2 lengthening could then account for the DMPP insensitivity of catp-1 mutants by occluding the reduction of development speed normally caused by activation of nicotinic receptors. Interestingly, the two daf-2 alleles that tested DMPP resistant are also slow growing: at 20°C, daf-2(m41) and daf-2(e1370) need one additional day to reach adulthood compared with wild type(Gems et al., 1998) (A.F.R. and J.L.B., unpublished). Part of this delay is due to an extended L2 stage,which is likely to be different from L2d as dauer formation was marginal in these conditions (A.F.R. and J.L.B., unpublished). As in catp-1mutants, L2 lengthening in daf-2 might account for the resistance to DMPP.

Signaling downstream of the insulin and IGF receptors has been extensively studied in vertebrates at the cellular and molecular levels. When activated,these receptor tyrosine kinases (RTKs) are able to phosphorylate multiple intracellular substrates, including the insulin receptor substrate proteins(IRS), Shc, Gab-1, Cbl and APS. Upon tyrosine phosphorylation, each of these substrates can recruit a distinct subset of signaling proteins containing Src homology 2 (SH2) domains and initiate different signaling pathways, among which are the Akt/PKB and the MAPK pathways(Lizcano and Alessi, 2002; Saltiel and Pessin, 2002). Each of these pathways plays a separate role in the different cellular effects of insulin and IGF-1. Most of the insulin and IGF receptors transduction machinery described in vertebrates is conserved in C. elegans. However, the genetics of dauer formation and aging have defined a linear pathway for insulin signaling, which consists of elements both necessary and sufficient for dauer formation and aging under the control of daf-2. Ultimately, DAF-2/InsR activation leads to the segregation of DAF-16/FOXO in the cytoplasm, preventing its interaction with transcriptional targets. This pathway is required to implement DMPP toxicity as daf-16(0)suppresses daf-2(e1370) DMPP resistance(Fig. 8A).

Despite the functional importance of the DAF-16/FOXO-dependent pathway for DAF-2/InsR signal transduction, increasing evidence substantiates the existence of DAF-16-independent DAF-2/InsR pathways(Gerisch and Antebi, 2004; Yu and Larsen, 2001). Our results, together with two recent studies(Hopper, 2006; Nanji et al., 2005), support the role of the Ras pathway in DAF-2/InsR signaling during larval development. First, we demonstrated that activated Ras efficiently suppressed the constitutive dauer formation of daf-2(m41) mutants but not of daf-2(e1370), as previously reported. Second, activated Ras was fully suppressing the DMPP resistance of daf-2(m41) but weakly suppressed the resistance of daf-2(e1370). Such genetic interaction data must be interpreted cautiously because we could not test the interactions between the null alleles of daf-2 and let-60/Ras which are both lethal early during development. However, these results raise the possibility that two signaling branches bifurcate downstream to DAF-2/InsR, one independent of LET-60/Ras and one involving LET-60/Ras. According to this simple model, the LET-60/Ras-dependent pathway would be prominently reduced in daf-2(m41) while both pathways would be depressed in daf-2(e1370) mutants. As suggested by Nanji et al.(Nanji et al., 2005), such differential alteration of the coupling between DAF-2/InsR and the Ras-MAPK pathway might participate in the phenotypic differences observed between daf-2 mutant alleles as DAF-2/InsR probably positively regulates LET-60/Ras activity to control both L2 developmental timing and dauer formation (Fig. 8).

Genetic interaction data indicate that CATP-1 participates in two processes regulated by DAF-2/InsR signaling during C. elegans larval development: catp-1 positively interacts with daf-2 in the control of L2 developmental timing and negatively interacts in the decision to form dauer larvae. In both cases, CATP-1 function seems to interact with the Ras-MAPK pathway discussed above. Like daf-2 mutants, the DMPP resistance of catp-1 mutants is suppressed by let-60(gf). However, catp-1(kr17) DMPP resistance was independent of DAF-16/FOXO,thus indicating an interaction of CATP-1 with a DAF-16/FOXO-independent DAF-2 pathway (Fig. 8A). Opposite to their interaction in the control of L2 developmental timing, CATP-1 negatively interacts with LET-60/Ras signaling to control entry into the dauer diapause,as both catp-1(kr17) and let-60(gf) suppress constitutive dauer formation in a daf-2(m41) mutant. Such dual interaction illustrates the complexity of the network that integrates insulin/IGF-1 signaling into developmental decision at the organism level.

One striking feature of the interactions between catp-1,let-60/ras and daf-2 is the high degree of allele specificity,as previously reported for many phenotypic traits of the daf-2mutants. Despite extensive analysis of multiple daf-2 mutant alleles,the relationship between the molecular lesions of the DAF-2 receptor and mutant phenotypes remains poorly understood. In our study, we observed the strongest interaction with m41, which corresponds to a G383E mutation in the Cys-rich region of the ectodomain(Yu and Larsen, 2001), outside of the ligand-binding interface (McKern et al., 2006). Three other mutations in the ectodomain(Kimura et al., 1997; Scott et al., 2002) show weak(m596) or no (e1368 and m577) genetic interaction with catp-1(kr17). Similarly, mutants of the kinase domain display either weak (e1391) or no (e1370) genetic interaction with catp-1(kr17). If differences in relative levels of disruption of LET-60/Ras and PI3-kinase signaling account for phenotypic differences based on genetic data (Nanji et al.,2005), the cellular and molecular mechanisms at work are unknown. For example, the m41 mutation in the ectodomain might affect the binding of one or a group of the many C. elegans insulin-like peptides that preferentially activate the Ras-MAPK pathway. Alternatively, the different mutations might cause subtle changes of the overall receptor activity, and phenotypic differences might arise from differential coupling with intracellular signaling pathways among the cells executing insulin/IGF-1-dependent programs. The multiplicity of the functions of DAF-2 in C. elegans and the prominence of cell non-autonomous processes has hampered such analysis so far.

Sequence analysis unambiguously identifies CATP-1 as an α-subunit of the Na⁺/K⁺- and H⁺/K⁺-pump P-type ATPase family. Four additional C. elegans genes are predicted to encode closely related proteins, including eat-6 which codes for a bona fide Na⁺/K⁺-ATPase α-subunit required for proper excitability of pharyngeal muscle cells(Davis et al., 1995). The three other predicted genes have not been analyzed. The primary function of these ATPases is to maintain ionic gradients across plasma membranes and intracellular homeostasis. By sequence comparison, CATP-1 is equally distant from Na⁺/K⁺- and H⁺/K⁺-ATPases,thus precluding a prediction of which ions might be transported by this protein. However, our results suggest that the identified functions of CATP-1 are mostly independent from its transport function, as recently proposed for some signaling functions of the vertebrate Na⁺/K⁺-ATPase in response to cardiac glycosides(Schoner, 2002).

Cardiac glycosides are the most widely employed therapy for cardiac insufficiency. They block the Na⁺/K⁺-ATPase, which affects a wide range of cellular phenomena(Kaplan, 2002; Xie and Askari, 2002). Those effects have long been thought to result from the change in cell Na⁺ content, which can lead to changes in intracellular pH (via the Na⁺/H⁺ exchange system), or intracellular Ca²⁺ (via the Na⁺/Ca²⁺ exchange system)(Kaplan, 2002; Scheiner-Bobis and Schoner,2001). Such ionic changes would subsequently initiate various signaling responses. However, disrupting Na⁺/K⁺-ATPase function has also been reported to modulate several classical intracellular signaling pathways such as ERK and Src pathways in different lines of cultured cells with no obvious alteration of intracellular cation concentrations(reviewed by Xie and Askari,2002). Recent evidence substantiates a direct interaction of the Na⁺/K⁺-ATPase with intracellular signaling machineries,independently of its pump function. The cytoplasmic region of the Na⁺/K⁺-ATPase was shown to interact in vitro and in vivo with the soluble tyrosine kinase Src. This interaction would maintain Src in an inactive form. Ouabain, an inhibitor of the Na⁺/K⁺-ATPase, is proposed to cause Src release and activation of Src and downstream effectors such as ERK and FAK. Interestingly,the inhibition of Src activity could be achieved with a pump-dead mutant as efficiently as with the wild-type protein(Liang et al., 2006; Tian et al., 2006). In our study, we mutated a CATP-1 residue which is essential for the catalytic cycle,similar to the Liang study, or a residue absolutely required for ATP binding,hence catalytic activity. Both mutant proteins were able to partially rescue catp-1(kr17) defects. Partial rescue could be explained by some contribution of the pump function. Alternatively, it might result from inappropriate expression in transgenic animals or from additional modifications of CATP-1 properties in the mutant forms. In any case, our data provide evidence that CATP-1 can function independently of its pump function to regulate L2 developmental timing.

Although a scaffolding function has been postulated for the Na⁺/K⁺-ATPase at septate junctions in the Drosophila tracheal system(Genova and Fehon, 2003; Paul et al., 2003), this is to our knowledge the first in vivo demonstration of a signaling function of a P-type ATPase that is independent of its pump function. Whether CATP-1 interacts directly of indirectly with the Ras-MAPK and DAF-2/InsR machinery remains equally possible at this stage. However, it is tempting to speculate that the cytoplasmic region of CATP-1 could serve as a membrane-bound docking platform to recruit components or modulators of the Ras-MAPK and/or DAF-2/InsR signaling pathways. The multiple developmental phenotypes of catp-1mutants offer an interesting opportunity to dissect the mechanisms underlying signaling pathway modulation by cation-transporting ATPases in vivo.

After this manuscript was accepted, Paul et al. reported an in vivo pump-independent function of the Na,K-ATPase for epithelial junction function in Drosophila (Paul et al.,2007).

Supplementary material for this article is available at http://dev.biologists.org/cgi/content/full/134/5/867/

We thank Ian Johnstone for the gift of a Pdpy-7 plasmid. Some nematode strains used in this work were provided by the Caenorhabdidtis Genetics Center, which is funded by the NIH National Center for Research Resources(NCRR). B. Matthieu is thanked for help with confocal microscopy, H. Gendrot for technical help, M.A. Félix, M. Labouesse, I. Katic and V. Robert for critical reading of the manuscript. A.F.R. was supported by a fellowship from theMinistère de la Recherche and by the Association pour la Recherche contre le Cancer. This work was funded by an AVENIR grant from the Institut National de la Santé et de la Recherche Médicale and by the ACI BDPI from the Ministère de la Recherche.