A Caenorhabditis elegans type I TGFβ receptor can function in the absence of type II kinase to promote larval development

Genetic and genomic analyses of Caenorhabditis elegans are providing insights into nematode biology and the organization and function of orthologous genes in vertebrates (Ruvkun and Holbert, 1998). However, functional similarity between the products of vertebrate gene families and their orthologs in evolutionarily distant nematodes may not always be conserved. Here we address the functional conservation of TGFβ receptor kinase signaling, which controls facultative diapause in C. elegans.

Dauer larvae are developmentally arrested, nonfeeding larvae specialized for survival and dispersal in an adverse environment. The dauer/nondauer developmental switch is influenced by temperature and by the abundance of food relative to the concentration of a dauer-inducing pheromone (Golden and Riddle, 1984a,b). DAF-7, a growth factor in the TGFβ superfamily, is expressed in ASI chemosensory neurons and promotes nondauer development. Transcription of daf-7 is regulated by pheromone and is a critical element controlling passage through the dauer stage (Ren et al., 1996). daf-7 mutants have a temperature-sensitive (ts) dauer-constitutive (Daf-c) phenotype and arrest at the dauer stage in conditions that favor growth (Swanson and Riddle, 1981).

The DAF-7 growth signal is received and transduced by two transmembrane receptor kinases, DAF-1 (Georgi et al., 1990) and DAF-4 (Estevez et al., 1993), and three SMAD [SMA (Small)/MAD (Mothers Against Decapentaplegic)-related] proteins, DAF-8 (Estevez, 1997), DAF-14 (Inoue and Thomas, 2000) and DAF-3 (Patterson et al., 1997). Worms with mutations in daf-1, daf-4, daf-8 or daf-14 are ts Daf-c, indicating that the wild-type function of these genes is to inhibit dauer formation. These mutants also exhibit adult phenotypes, including defects in egg-laying and body fat accumulation (Kimura et al., 1997; Thomas et al., 1993). The DAF-3 SMAD functions to promote dauer formation and inhibit egg-laying. Mutations in daf-3 prevent the dauer formation and suppress the Daf-c and adult phenotypes of ligand, receptor and SMAD mutants (Thomas et al., 1993).

daf-4 is the only type II receptor gene in the C. elegans genome (C. elegans Sequencing Consortium, 1998) and its mutant phenotype is pleiotropic, reflecting the receptor’s role in more than one signaling pathway. Mutations in dbl-1/cet-1, which encodes a bone morphogenetic protein (BMP)-like ligand (Morita et al., 1999; Suzuki et al., 1999), daf-4, the type I receptor sma-6 (Krishna et al., 1999), and three SMAD genes, sma-2, sma-3 and sma-4 (Savage et al., 1996), all result in small adult body size (Sma) and abnormal male tail structures (Mab). The requirement for DAF-4 in Daf and Sma/Mab signaling suggests that it can bind either DAF-7 or DBL-1/CET-1. DAF-4 binds human BMPs when expressed in COS cells (Estevez et al., 1993). DBL-1/CET-1 is similar to BMPs, but DAF-7 is a novel TGFβ family member with sequence similarities to both TGFβ and BMPs (Ren et al., 1996). daf-4 mutants are the only ones with a Daf-c and Sma/Mab phenotype, but enhancement of the Daf-c phenotype in double mutant strains daf-1; sma-6 (Krishna et al., 1999) and daf-7; cet-1 (Morita et al., 1999) suggest that sma-6 and dbl-1/cet-1 function at a low level in dauer signaling.

The mechanism of TGFβ receptor activation (Wrana et al., 1994) has been shown to be conserved in the activin receptor (Attisano et al., 1996), a human BMP receptor (Liu et al., 1995) and the Thick-veins/Punt (TKV/PUNT) receptor (Ruberte et al., 1995) for Drosophila Decapentaplegic (DPP). In these examples, the type II receptor binds ligand and then associates with the type I receptor in a heteromeric ligand-receptor complex. Phosphorylation by the type II kinase activates the type I kinase, which in turn phosphorylates transcription-activating SMADs. Using this as a model for type I receptor activity, we tested the dependence of DAF-1 signaling on the function of the DAF-4 type II receptor in vivo. We conclude that signaling by the DAF-1 receptor can occur in the absence of DAF-4 kinase activity. Sequence differences between DAF-1 and other type I receptors, especially in regions that mediate the transition from the inactive to the active conformation of TβR-I, suggest the structural basis for this DAF-4-independent activity.

Nematodes were grown on NG agar plates seeded with the Escherichia coli uracil auxotroph OP50 (Brenner, 1974). Strains used in this work include the wild-type N2, daf-1 IV mutants listed in Table 1, daf-4(e1364)III, daf-7(e1372)III and lon-1(e185)III. To construct daf-4(e1364); daf-1(m40) and daf-4(e1364); daf-1(p168) double mutant strains, nondauer F₁ progeny from daf-1 hermaphrodites mated with daf-4(e1364)/+ males at 25.5°C were isolated. F₂daf-1 homozygotes were identified as Egl adults issuing from F₁ adults segregating Sma (daf-4) progeny. daf-4; daf-1 double mutants were identified as F₃ Sma, Egl adults at 15°C, isolated, and their progeny tested for dauer formation at 25.5°C. The genotype was confirmed by noncomplementation with daf-1(m40). A similar scheme was used to construct the daf-4(e1364); daf-7(e1372) double mutant, except that F₂daf-7 homozygotes were identified as dauer larvae at 25.5°C that recovered at 15°C to adults segregating Sma (daf-4) progeny.

All mutants were maintained at 15°C prior to analysis. For each strain, 10 or 20 gravid adults were transferred to a fresh plate and incubated at 15°C, 20°C, 22.5°C or 25.5°C. After 6 to 12 hours, the adults were removed, eggs were returned to the experimental temperature and populations scored for the presence of dauer larvae at the following times: 2 days at 25.5°C, 3 days at 22.5°C, 4 days at 20°C and 5 and 6 days at 15°C. Dauer larvae and adults were counted and removed at the time of scoring. Animals that were neither dauer nor adult were incubated overnight at the same temperature and scored on the following day.

The daf-1 promoter was isolated as a 2.6 kb SalI-BglII genomic fragment (2561 to 32 bp upstream of the ATG codon) and ligated upstream of gfp in pPD95.75 (kindly provided by A. Fire, J. Ahnn, G. Seydoux and S. Xu) to generate DR#268. The daf-4 promoter was first cloned as a 3.3 kb SalI-SwaI fragment (3445 to 110 bp upstream of the ATG codon) into pBluescript (Stratagene) and then excised with SalI and BamHI for ligation to pPD95.75 to generate DR#262. The daf-1 gene was cloned from cosmid F29C4 as a 9.8 kb BstEII fragment into a pBluescript vector modified to create a cloning site for the gene by cutting with PstI and BamHI and inserting a double-stranded oligonucleotide encoding two internal BstEII sites (DR#333). Each plasmid was injected into the ovaries of young adults at a concentration of 100 μg/ml. Expression patterns were observed in transgenic lines generated by injection of N2 animals with gfp reporter plasmids and pRF4 (Mello et al., 1991) expressing rol-6(su1006) as a transformation marker. To test for rescue of daf-4(e1364), daf-1(+) and pRF4 were coinjected into daf-4(e1364) +/+ lon-1(e185) heterozygotes and transformed daf-4(e1364) homozygous progeny were isolated as Sma Rol F₁ segregants. Heterozygotes were used because of the small brood size of injected homozygous daf-4(e1364) adults. Microscopy images of GFP expression were captured as described in Fig. 2B legend.

Genomic DNA used in chemical mismatch detection was prepared from mixed-stage populations of worms grown on plates at 15°C. For PCR amplification of daf-1 sequences, the following oligonucleotide pairs (written from 5′ to 3′) were used: AGGCGCCGAAACTTCCG-ACG with CGGTCACAGATTTCTCGTCC, AACACGTGCTAC-ACGGACGG with CACTTGCCGTCGACCTCATC, ATGAGG-TCGACGGCAAGTGG with TTCCGATCCTGAACTGTCAC, and AGAGAAGCGGCCACAGTGAT with CTCTAACCAAGAAGTGG-GCG.

Each PCR product was radiolabeled using T4 polynucleotide kinase and γ[³²P]ATP (6000 Ci/mmole). Heteroduplexes were generated by denaturing 10 ng of labeled DNA (wild-type or mutant) mixed with 50 ng of cold DNA (mutant or wild-type) in 50 μl of annealing buffer [0.3 M NaCl, 0.1 M Tris-HCl (pH 8), 0.1 mM EDTA] at 95°C, then reassociating at 65°C overnight. Chemical modification of mismatched C bases with hydroxylamine and cleavage with piperidine were carried out as described by Cotton et al. (1988). The products were resolved by electrophoresis in denaturing polyacrylamide gels and mismatch sites located by comparison of novel bands with radiolabeled DNA size standards. Mutations were identified by automated sequencing of PCR fragments.

For analysis of daf-4 mRNA from the e1364 allele, poly(A)-enriched RNA was used for reverse transcription and PCR amplification of a daf-4 cDNA with the Superscript One-Step RT-PCR System (GIBCOBRL). Primers corresponding to sequences in exon 1 (5′CACGGAAGACGATCGTCA3′) and exon 8 (5′GGAGCTGC-CGATCACTGA3′) yielded an 833 bp product that was cloned into the pGEM-T vector (Promega) and sequenced.

A synthetic peptide (CSSNDDSSRPLLG) consisting of an N-terminal Cys linked to twelve amino acids corresponding to the C-terminal sequence of the daf-1 protein was coupled to keyhole limpet hemocyanin using m-maleimido benzoyl-N-hydroxysuccinimide ester (Harlow and Lane, 1988). Two rabbits were injected with the conjugated peptide using the Ribi Adjuvant System (Ribi Immunochem Research, Inc., Hamilton MT) and preimmune and immune sera were evaluated in enzyme-linked immunosorbent assays (Harlow and Lane, 1988). Peptide-specific antibodies were purified on a Sepharose peptide affinity column prepared and used according to manufacturer’s instructions (Pierce Sulfolink). For Western analyses, mutant strains were grown on NG agar plates at 15°C and staged liquid cultures of wild-type N2 were grown at 20°C. Nematode extracts were prepared by suspending washed worms in 10 volumes of SDS sample buffer (Harlow and Lane, 1988) and immediately heating at 100°C for 3 minutes prior to loading onto an 8% SDS polyacrylamide gel. Extracts resolved by SDS-PAGE were transferred to nitrocellulose and the filter was blocked in 5% nonfat dry milk in phosphate-buffered saline prior to incubation in a 1:800 dilution of affinity-purified peptide antibodies. The filter was incubated in a 1:3000 dilution of alkaline-phosphatase-conjugated goat anti-rabbit secondary antibody, and alkaline phosphatase activity was detected using bromochloroindolyl phosphate/nitro blue tetrazolium substrate (Harlow and Lane, 1988).

DAF-1 is more similar to type I than to type II receptors in the TGFβ family. DAF-1 has ten cysteine residues that are conserved in the extracellular domains of type I receptors for TGFβ, activin, BMPs and DPP, but not in type II receptors (Fig. 1A). Furthermore, DAF-1 has a glycine-serine rich sequence (GS domain) that is characteristic of type I receptors. Phosphorylation of serines in the GS domain by an associated type II receptor is required for activation of the TGFβ type I kinase (Wrana et al., 1994). However, DAF-1 is the most divergent member of the type I receptor clade (Fig. 1B). Whereas the similarity between the kinase domains of type I receptors averages 50%, DAF-1 is only 25% similar (Fig. 1C).

To address the question of whether DAF-1 and DAF-4 function in the same or in different cells, and to identify putative target cells for DAF-7 signaling, we compared the expression patterns of gfp fused to the daf-1 and daf-4 promoters and looked for areas of overlap. GFP expression mediated by a 2.6 kb fragment of daf-1 upstream regulatory sequence was observed in the head and the developing ventral nerve cord beginning in mid-stage embryos and continuing into adulthood (Fig. 2A). In the head, GFP was detected in more than twenty neurons in the anterior, lateral, ventral and retrovesicular ganglia (Fig. 2B,C). Fluorescent processes terminating at the tip of the head suggest that daf-1 is expressed in sensory neurons and in support cells in the amphids and inner labial sensilla (Fig. 2C). In the midbody, GFP was expressed in the ALM mechanosensory neurons (Fig. 2A) and the PVT neuron (Fig. 2E,F), as well as one additional neuron pair. In the lumbar ganglia of the tail, five cells expressing GFP included phasmid neurons and PLN and PLM mechanosensory neurons (Fig. 2E,F). The daf-1 promoter also conferred gfp expression in nonneuronal cells, including a membranous sheath surrounding the distal end of the intestine (Fig. 2A) and in the distal tip cell (DTC) of the gonad (Fig. 2D). In some lines, GFP was sometimes detected in the muscles of L4 and adult animals.

To compare the transcription pattern of the daf-1 promoter with the whole gene, the gfp cDNA was fused to the terminus of the daf-1 gene in a plasmid that included 2.6 kb of upstream sequence. Despite the fact that this transgene rescued the daf-1 Daf-c mutant phenotype, GFP fluorescence was not detected in rescued animals. Hence, our observations were limited to the daf-1 promoter fusion, which may not represent the expression pattern of the whole daf-1 gene if enhancer elements are present in introns or in 3′ sequences.

The temporal pattern of GFP expression directed by the daf-1 promoter was compared to endogenous DAF-1 protein detected in immunoblots. Antibodies to a DAF-1 C-terminal peptide detected a protein in wild-type worms that was not seen in a daf-1(e1146) null mutant or in the presence of excess DAF-1 peptide (Fig. 3A). The specificity of antibody reactivity with this 88 kDa protein, as well as its similarity to the predicted mass of DAF-1 (79 kDa) support its identity as endogenous DAF-1. In an analysis of synchronized populations of wild-type worms, the 88 kDa protein was detected in all developmental stages (Fig. 3B), although steady-state levels of DAF-1 appear highest in L1. Expression of GFP at all stages of development correlates with these immunological data.

GFP expression mediated by 3.3 kb of daf-4 5′ sequence matched the expression pattern of a daf-4::gfp gene fusion (Patterson et al., 1997), suggesting that daf-4 regulatory sequences conferring tissue-specificity are upstream of the transcription start site. GFP was expressed in the pharynx (Fig. 2G), intestine, hypodermis (Fig. 2H) and body wall muscles (Fig. 2I), in L1 through adult stages. In the head, GFP was seen in neurons of the lateral, vesicular and retrovesicular ganglia (Fig. 2G), although fluorescence from the pharynx hindered comparisons with the daf-1 pattern. Ventral cord neurons also were visible, as was the PVT neuron (Fig. 2J-K), but only two phasmid neurons were detected in the tail (Fig. 2J-K).

Whereas the nervous system is the primary site of gfp expression mediated by the daf-1 promoter, the daf-4 promoter directs expression more broadly, consistent with its other functions. In L1 larvae, when the dauer/nondauer decision is made, both daf-1 and daf-4 promoters are active in neurons in the head, as well as in the ventral cord and tail. Both promoters continue to express GFP in similar patterns in dauer larvae from starved plates (data not shown). Coexpression of daf-1 and daf-4 seems likely to occur in the subset of neurons expressing gfp with either promoter, consistent with the hypothesis that DAF-7 secreted from amphid ASI neurons is detected by other neurons, which then transmit growth signals to other cell types (Ren et al., 1996).

The divergent sequence of DAF-1 led us to question whether the regulation of DAF-1 signaling is like other type I receptors. To map amino acids critical for function, we sequenced fourteen daf-1 loss-of-function mutants; G-to-A transitions, common with EMS, were detected in thirteen (Table 1). Five mutations encoded premature stop codons, seven were missense and one (m214) affected the 5′ splice sequence of intron 7 (Fig. 1A). In immunoblots, full-length DAF-1 was detected in six missense mutants, whereas no 88 kDa species was detected in the five nonsense mutants, the Tc1 insertion, or the splice site mutant (Fig. 3C). Truncated polypeptides expressed by nonsense mutants would not be detected with the C-terminal peptide antibodies.

Four amino acid substitutions are within the span of ten cysteine residues conserved in the extracellular domain of all type I receptors (Fig. 1A). In daf-1(p168), a glycine-to-arginine substitution (G70R) is positioned between the third and fourth cysteines (C64 and C90). The daf-1(m138) mutation changes the fourth cysteine to tyrosine (C90Y). daf-1(m122) carries two mutations that result in substitutions in a glutamic acid (E105A) and glycine (G110E) proximal to the sixth cysteine (C111). The glycine is conserved in the extracellular domains of all BMP and TGFβ type I receptors, and may be important for proper configuration of the neighboring cysteine.

Although the extracellular domains of type I receptors share little similarity, the cytoplasmic regions are well conserved.

The crystal structure of TβR-I cytoplasmic domain (Huse et al., 1999) provides a model for predicting the effects of the daf-1 kinase mutations. TβR-I kinase, like other kinases, has two subdomains (Fig. 4A). The N lobe, defined by a twisted five-stranded β sheet and a single α helix (helix C), is primarily involved in coordination of Mg²⁺ -ATP. The C lobe positions the peptide substrate and contains many of the catalytic residues required for phosphotransfer. The inactive conformation of TβR-I is maintained by interactions between the type I-specific GS region and the activation segment, which stabilize helix C in a position that shifts the orientation of the N lobe β sheet relative to the C lobe, shearing the active site and distorting the ATP-binding pocket (Fig. 4B).

Two substitutions in DAF-1 affect the N lobe at the center twist of the TβR-I β sheet, where hydrophobic interactions with the GS domain helices are made. In daf-1(n690), a glycine conserved in other type I receptors is changed to glutamic acid (G352E). We predict that the longer side chain of the glutamic acid could disrupt the packing of the β sheet. In daf-1(sa40), the effect of the arginine-to-lysine (R349K) substitution is difficult to predict. All other type I receptors maintain a glycine at this position and, in TβR-I, substitution of glutamic acid for glycine prevents kinase-activating phosphorylation of TβR-I by TβR-II (Weis-Garcia and Massagué, 1996). The similarity in Daf-c phenotype between daf-1(sa40) and daf-4(e1364), a kinase deletion, supports the argument that R349 could promote DAF-1 phosphorylation by DAF-4.

In the C lobe of the DAF-1 kinase, daf-1(m575) mutates a glycine conserved in the activation sequence of all type I and type II receptors to aspartic acid (G465D). Modeling predicts that the substitution does not disrupt the local environment, but novel charge interactions could occur with an arginine in the catalytic loop (R332 in DAF-1) that is positioned near G465. In daf-1(m213), a proline invariant in all type I and II receptors, but not other kinases, is changed to serine (P546S). In TβR-I, the proline interacts with the aliphatic portion of the adjacent arginine and two conserved tryptophan residues (W354 and W241 of DAF-1) to form a hydrophobic region, which would be disrupted by substitution with serine.

To correlate structural changes in mutant DAF-1 proteins with functional changes in receptor activity in vivo, we quantified the Daf-c phenotype of daf-1 mutants at four temperatures (Table 1). Even nonsense mutations are not fully penetrant at lower growth temperatures, due to the temperature sensitivity of wild-type dauer formation (Golden and Riddle, 1984b). The phenotypes of four of the five nonsense mutants (e1287, e1146, m42 and m55), the Tc1 insertion (m412), and the splice-site mutant (m214) are similar, with approximately 30-75% of the populations forming dauer larvae at 15°C, 50-90% at 20°C, greater than 70% at 22.5°C and 100% at 25.5°C. This frequency represents the null phenotype of daf-1 under the conditions of our assay and indicates that DAF-1 signaling is not essential for growth at lower temperatures.

The m40 nonsense mutant, which encodes a UGA stop at W170, exhibits a weak phenotype relative to others. To test whether the strain carried a suppressor, it was crossed with wild type and resegregated. Two independent segregants were tested and exhibited the same Daf-c phenotype as the parent strain at 15° and 20°C. Since the m40 UGA codon is positioned upstream of transmembrane and cytosolic domain sequences, the weaker Daf-c phenotype may be due to low levels of DAF-1 protein produced by translational read-through (below the detection limits of Fig. 3 immunoblots). Like other animals, C. elegans has tRNA^[Ser]Sec that can insert selenocysteine at UGA codons (Lee et al., 1990). In prokaryotes, substitution of tryptophan at opal stops occurs due to third-position wobble in codon-anticodon recognition (Hirsh and Gold, 1971).

daf-1 missense alleles can be ordered by Daf-c phenotype in the following series: m122, m138>sa40>m213>p168>n690> m575 (Table 1). Extracellular domain mutants m122 and m138 form as many dauer larvae as the nonsense mutants, suggesting that this domain is important for DAF-1 signaling activity. The strong phenotype of sa40 belies the conservative arginine-to-lysine substitution, which might alter important interactions with DAF-4. The four remaining mutants (m213, p168, m575 and n690) all form few dauer larvae at 15°C, but display different profiles of temperature sensitivity. Some of these mutants may encode ts DAF-1 polypeptides that change conformation and disrupt function at higher temperatures. Immunodetection of these mutant proteins indicates that they are relatively stable.

Testing the dependence of DAF-1 signaling on phosphorylation by DAF-4 required a daf-4 mutant that eliminated kinase activity. A daf-4 deletion identified during PCR analysis of e1364 genomic DNA was characterized by sequencing a clone generated with primers that flank the deletion. The 212 bp gap deletes 31 bp of exon 5 and 181 bp of intron 5, including the 5′ splice sequence (Fig. 5A). To check splicing, we used RT-PCR to amplify daf-4 cDNA from e1364 RNA. The cDNA sequence showed that a cryptic 5′ splice sequence, 48 bp downstream from the deletion site, mediates the joining of exon 5 to exon 6, but also changes the reading frame of exon 6 (Fig. 5C). The putative translation product is a 208 amino acid polypeptide that includes amino acids 1-194 of DAF-4, followed by 14 amino acids and a UGA stop encoded by intron 5. A second stop (UAG) is encoded in exon 6, 63 bp downstream from the first stop. The polypeptide resulting from translation of the e1364 mRNA would not contain the membrane-spanning or kinase domains of DAF-4. Hence, we consider daf-4(e1364) to be a null allele with regard to kinase function. We have not tested whether a daf-4(e1364) product is made or is capable of binding ligand as a secreted receptor.

The daf-4(e1364) allele was used in combination with either of two hypomorphic daf-1 alleles, m40 or p168, to compare the phenotype of daf-4; daf-1 strains to daf-4 alone (Table 2). At 20°C, the Daf-c phenotype of daf-4(e1364) (Table 2) is less severe than daf-1 null mutants (Table 1) and similar to the daf-1(sa40) missense mutant, suggesting that daf-1 can signal in the absence of daf-4 kinase activity. If daf-1 requires daf-4 phosphorylation for signaling, no difference in the Daf-c phenotype of the daf-4; daf-1 double mutants relative to the daf-4(e1364) single mutant would be expected. However, if daf-4; daf-1 double mutants are more severe than daf-4; daf-1(+), we would conclude that daf-1(+) conveys a nondauer signal even in the absence of DAF-4 phosphorylation.

Significant enhancement of daf-4(e1364) dauer formation by the two different daf-1 alleles was observed at three growth temperatures (Table 2). At 22.5°C, almost all of the double mutant populations form dauer larvae, whereas only 68% of e1364 larvae arrest in the dauer stage (P<<0.001). Although the phenotypes of m40 and p168 alone are weak at 15°C and 20°C, the frequency of dauer formation in e1364; m40 and e1364; p168 strains is greater than in the e1364 single mutant. These results indicate that, in the absence of type II kinase activity, DAF-1 maintains signaling that promotes nondauer development, albeit not as efficiently as in the presence of the type II receptor.

Only BMP type I receptors have been shown to bind ligand and associate as homodimers in the absence of type II, although these receptors still require transphosphorylation for activation of downstream pathways (Graff et al., 1994; Koenig et al., 1994; Penton et al., 1994; ten Dijke et al., 1994a). Although we have not tested the ability of DAF-1 to bind DAF-7, the phenotypes of two daf-1 mutants (m138 and m122) with amino acid substitutions in the ligand-binding domain resemble daf-1 null alleles, indicating a role for ligand binding in phosphorylation-independent signaling. Furthermore, the Daf-c phenotype of the daf-7(e1372) null mutant (Ren et al., 1996) at 20°C and 22.5°C is more severe (P<<0.001) than daf-4(e1364) and similar to daf-4; daf-1 double mutants (Table 2).

The Daf-c phenotype of the e1364; e1372 double mutant at 20°C and 22.5°C is enhanced relative to daf-4(e1364) (P<<0.001), indicating that DAF-7 promotes nondauer signaling in the absence of DAF-4.

As a second test of the ability of DAF-1 to signal in the absence of DAF-4 phosphorylation, we increased daf-1(+) gene dosage in the daf-4(e1364) null mutant and tested for rescue of the Daf-c phenotype (Table 3). Any type I receptor dependent on type II activation should not be capable of rescuing a type II kinase mutant. Five transgenic lines carrying extrachromosomal arrays of daf-1(+) and the transformation marker rol-6(su1006) were assayed for dauer formation. At 20°C, dauer formation of daf-4(e1364) Rol progeny relative to nonRol siblings increased from 10% to 57% (P<<0.001), and also differed significantly from the Rol animals in rol-6(su1006) control lines (P<0.05). At 25.5°C, all progeny arrested in a dauer-like state, but an average of 15% of daf-1(+) Rol larvae resumed development after prolonged incubation times. Control rol-6(su1006) lines showed no such recovery. In no case did the overexpression of daf-1(+) rescue daf-4 completely, showing that wild-type signaling requires both daf-1 and daf-4. In a reciprocal experiment, a daf-1(e1287) nonsense mutant transformed with extrachromosomal daf-4(+) showed no rescue at 20°C or 25.5°C (data not shown), indicating that daf-4(+) is dependent on daf-1 activity for nondauer signaling.

Six daf-1 putative null mutants form more dauer progeny than daf-4(e1364) at 20°C. These data alone suggest that nondauer signaling through daf-1 can occur in the absence of daf-4 kinase activity. Further support for this hypothesis is found in two other genetic analyses. Mutations in daf-1 strongly enhance the Daf-c phenotype of a daf-4 kinase mutant (Table 2) and overexpression of daf-1(+) partially rescues a daf-4 mutant (Table 3). The observed activity of DAF-1 in the absence of DAF-4 transphosphorylation might be an evolutionary remnant, since DAF-1 diverged from other type I receptors soon after the split between type I and II ancestors (Newfeld et al., 1999). Alternatively, DAF-1’s activity might be an evolutionary innovation acquired during the long period since nematode receptors diverged from receptors in other organisms (ie. vertebrates and insects) represented in current phylogenies.

Our results suggest that DAF-1 signaling is dependent on ligand binding (Table 2). We predict that the role of DAF-7 is to promote formation of DAF-1 homodimers for signaling. It is possible that the truncated DAF-4 polypeptide encoded by daf-4(e1364) is secreted and maintains ligand-binding activity. Based on the dominant negative phenotype observed for kinase deletion mutants in the type II receptor family (Brand et al., 1993; de Winter et al., 1996; Suzuki et al., 1994), we predict that the effect of such a polypeptide would be inhibitory on DAF-1 activity, since it would compete with the DAF-1 receptor for ligand binding.

Despite significant overlap in GFP expression patterns, there are cells in which the daf-1 promoter, but not the daf-4 promoter, is active. GFP expression with the daf-1 promoter in mid-stage embryos, when the daf-4 promoter is not active, could be due to translation of maternal mRNAs in the embryo, since daf-1 mutants exhibit a strong maternal rescue phenotype whereas daf-4 mutants do not. In the nervous system, the daf-1 promoter is active in some neurons where daf-4 activity is not detected (e.g., compare tail neurons in Fig. 3E with J). Expression of GFP in the DTC of the gonad is also specific to the daf-1 promoter. Interestingly, gfp expression in embryos and the DTC was a difference noted between daf-3::gfp and daf-4::gfp expression patterns as well (Patterson et al., 1997). We propose that DAF-1 is a type I receptor kinase that activates growth-promoting SMAD proteins (DAF-8 and DAF-14), and that the function of DAF-4 as a type II receptor is to enhance the catalytic activity of DAF-1. If there are cells where daf-1 but not daf-4 is expressed, as suggested by the promoter analysis, DAF-1 may function as the DAF-7 receptor, since daf-4 encodes the only type II receptor in C. elegans genome (C. elegans Sequencing Consortium, 1998).

Although some daf-1 activity is seen in the daf-4 deletion mutant, wild-type signaling can only be achieved with daf-1 and daf-4 together. Mutations in either gene result in a Daf-c phenotype, and overexpression of daf-1(+) promotes daf-4(e1364) nondauer development at 25.5°C only after an extended pause in development. Overlap between gfp expression patterns with the daf-1 or daf-4 promoters supports a model in which these two receptors function as partners in the nervous system. Based on interactions that occur between other type I/type II receptor pairs, there are two likely mechanisms for the enhancement of DAF-1 activity by association with DAF-4. First, DAF-4 phosphorylation of DAF-1 serines in the GS domain may release any conformational inhibition of DAF-1 kinase activity. Second, like other BMP receptors (Koenig et al., 1994; Letsou et al., 1995; ten Dijke et al., 1994b), the binding affinity of DAF-7 for DAF-1/DAF-4 may be higher than for DAF-1 alone, stabilizing the heteromers and promoting more frequent interactions with SMAD proteins.

The overall divergence of the DAF-1 kinase domain from TβR-I and other type I receptors is depicted in the amino acid comparisons in Fig. 1. Although the three-dimensional structure of DAF-1 has not been determined, comparison with TβR-I provides insight into structural differences in DAF-1 that might contribute to its activity in the absence of DAF-4 transphorphorylation. Two regions of special interest are those that interact to establish the TβR-I inactive configuration: the GS domain and activation segment.

In TβR-I, helix GS2 is cradled on the outer surface of the N lobe β sheet (Fig. 4A) and forms a network of interactions with the β sheet and helix GS1 in the absence of GS domain phosphorylation. This hydrophobic core stabilizes the position of the GS loop, which wedges between helix C and the β4 strand of the sheet and shifts residues in the N lobe away from those in the catalytic loop (Fig. 4B), separating components of the active site (Huse et al., 1999). The sequence of the DAF-1 GS2 helix differs from other type I receptors, sharing only three amino acids in common with the eight residue consensus and ending with two glycines unique to DAF-1 (Fig. 1A). Because glycine is frequently found at the C-terminal position of helices, but rarely among the three amino acids adjacent to the terminus (Richardson and Richardson, 1988), we predict that DAF-1 helix GS2 is shorter than other type I receptors by at least two amino acids. A shorter GS2 helix could reduce the shift in the position of the C helix away from the N lobe β sheet, which would relieve distortion of the DAF-1 active site and may permit residual activity in the absence of phosphorylation by DAF-4.

In TβR-I, the rotation of the N lobe restricts ATP binding in the groove between the N and C lobes. R372, conserved in the activation sequence of all type I receptors except DAF-1 and SAX, contributes to the occlusion of this site by extending its side chain into the catalytic center and forming an ion pair with D351 (Huse et al., 1999), which is required for interaction with the magnesium ion of Mg²⁺ -ATP. In DAF-1, the position of R372 is occupied by lysine (K463) (see Fig. 1A). In the TβR-I crystal structure, substitution of lysine for R372 would increase the D351 bond distance by approximately one angstrom, weakening the strength of the salt bridge that prevents Mg²⁺ -ATP interaction. This amino acid difference that might confer greater flexibility to the conformation of DAF-1 and prevent complete inhibition of the kinase. However, resolution of DAF-1 three-dimensional structure will be required to determine the physical basis for the type II-independent signaling activity that we have observed.

DAF-4 functions as a type II receptor in two signaling pathways that require distinct ligands (DAF-7 or DBL-1/CET-1), type I receptors (DAF-1 or SMA-6), and SMAD proteins (DAF-8, DAF-14 and DAF-3 or SMA-2, SMA-3 and SMA-4). Enhancement of the Daf-c phenotype of double mutants has been observed between daf-7 and dbl-1/cet-1 (Morita et al., 1999) and between the daf-1 and sma-6 (Krishna et al., 1999). A reciprocal enhancement of the Sma/Mab phenotype would be expected in each double mutant, however these data have not been reported, possibly due to the difficulty in measuring small changes in body size.

DAF-1 and SMA-6 are paralogues (homologues that are descendants of an ancient gene duplication). The weak interactions observed between components of the two pathways may be a vestige of this ancestral relationship. If DBL-1/CET-1 can bind DAF-1 receptor complexes (albeit less effectively than DAF-7) to activate nondauer signaling, loss of such activation could account for the enhancement of the Daf-c phenotype observed in ligand double mutant strains. The Daf-c enhancement in daf-1; sma-6 double mutants requires that some cross reactivity also exists between SMA-6 and DAF-8 or DAF-14, or between SMA-2, SMA-3 or SMA-4 and downstream interactors that function to promote nondauer development.

Unlike other type I receptors, the activity of DAF-1 in the absence of type II kinase activity provides a secondary response system to ligand. The flexibility of maintaining two kinds of receptors with differing signal potencies could provide greater precision in deciding the developmental fate of larvae. Since the signaling activity of DAF-1 alone is not as strong as DAF-1 with DAF-4, neighboring cells presenting DAF-1 homoreceptors or DAF-1/DAF-4 heteroreceptors at their surface will vary in their degree of intracellular activation of SMAD proteins. Individual cells capable of changing the ratio of DAF-1 and DAF-4 receptors on their surface could adjust the strength of intracellular signals at different times of development. This system allows for greater specialization of each cell’s response to DAF-7.

The long-standing puzzle of how genetically identical individuals display different sensitivities to dauer-inducing pheromone (Golden and Riddle, 1984a) might be explained by slightly different levels of type I versus type II expression among corresponding cells in sibling progeny. Further modulation of the dauer/nondauer decision occurs by integration of the DAF-7 signal with others from pathways that respond to thermal stress and food availability (Kimura et al., 1997), which also contribute to the differential development of worms in environments of questionable suitability for growth.

We thank J. Sue McCrone for excellent technical assistance, Carol Moyer for her skill in handling rabbits, and Rabbit 7235. Annette Estevez constructed the daf-4(e1364)III; daf-1(m40)IV double mutant, and Patrice Albert reviewed strategies for strain construction and phenotypic analysis. We thank Hannah Alexander for advice on antibody production, Pei-feng Ren for help with microinjection, and Tom Quinn and John Tanner for assistance with protein modeling software. We especially thank Morgan Huse for comments on the manuscript and for Fig. 4 (modified from Huse et al., 1999). DNA sequencing and confocal microscopy services were provided by the Molecular Biology Program core research facilities at the University of Missouri. This work was supported by DHHS grant GM60151. C. V. G. and L. L. G. were recipients of Individual National Research Service Awards (GM17491 and GM12583, respectively) from the National Institutes of Health