NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Riddle DL, Blumenthal T, Meyer BJ, et al., editors. C. elegans II. 2nd edition. Cold Spring Harbor (NY): Cold Spring Harbor Laboratory Press; 1997.

Cover of C. elegans II

C. elegans II. 2nd edition.

Show details

Section IVA Network of Gene Functions

A. Mutants

More than 30 genes controlling dauer larva formation have been identified. Mutations in daf (dauer formation) genes result either in the inability to form dauer larvae in response to crowding and starvation (dauer-defective, or Daf-d) or in the formation of dauer larvae at low population density in the presence of abundant food (dauer-constitutive, or Daf-c). Nonconditional dauer-constitutive mutants form dauer larvae independently of environmental cues. Temperature-sensitive dauer-constitutive mutants form dauer larvae at high frequency only at restrictive temperatures, and if such larvae are shifted to permissive temperatures, they exit from the dauer stage and resume growth (Swanson and Riddle 1981). The latter mutants are easily selected as detergent-resistant larvae formed in abundant food (Cassada 1975; Riddle 1977).

Identifying constitutively formed dauer larvae in visual screens (Riddle 1977; Malone and Thomas 1994; I. Caldicott and D. Riddle, in prep.) allows isolation of mutants that nonconditionally arrest development since heterozygous siblings can be used to propagate the mutant. In this way, nonconditional alleles of daf-23 (now called age-1 ) and daf-2 have been collected; these mutants arrest development at all temperatures. Mutants that form dauer-like (partial dauer) larvae nonconditionally have been identified, at least two of which, daf-9 and daf-15 , have been placed downstream from the daf-c and daf-d genes in the genetic pathway (Albert and Riddle 1988). Only temperature-sensitive alleles of daf-1 , -4, -7, -8, -11, -14, and -21 have been identified, suggesting that the null phenotype of these genes is temperature-sensitive. This model has been confirmed by molecular analysis of daf-1 , -4, and -7, since nonsense mutations, deletions, and transposon insertions all result in temperature-sensitive phenotypes (Georgi et al. 1990; Estevez et al. 1993; Ren et al. 1996). The relationship between dauer formation and thermotaxis behavior is intriguing. Some thermotaxis-defective mutants do not differ in their sensitivity to dauer-inducing pheromone, but others are either more sensitive or less sensitive (Golden and Riddle 1984a). tax-2 and tax-4 mutants have a slight Daf-c phenotype (Bargmann and Mori, this volume).

Virtually all Daf-c mutants obtained thus far are recessive. Exceptions are the single allele of daf-28 , which is weakly dominant and may be a gain-of-function allele (Malone and Thomas 1994), and dominant activating mutations in gpa-2 and gpa-3 , two genes that encode G-protein α subunits (Zwaal et al. 1997). Some Daf-c mutants are maternally rescued (all progeny of a daf/+ heterozygote grow to the adult at 25°C, but the Daf-c homozygotes produce all Daf-c progeny at the same temperature). The maternal-effect is complete for age-1 (Gottlieb and Ruvkun 1994), for some alleles of daf-1 (Riddle 1977), and the single allele of daf-21 , and it is partial for some alleles of daf-1 (Malone and Thomas 1994) and for daf-4 . Daf-d mutants are likewise recessive, as judged from their recessive suppression of Daf-c mutants. An exception is the weakly semidominant suppression of daf-2 and age-1 by daf-16 mutations (Gottlieb and Ruvkun 1994).

About half of the mutants selected by their dauer-defective phenotype also exhibit sensory defects involving chemotaxis, male mating, or osmotic avoidance (Riddle 1977; Culotti and Russell 1978; Albert et al. 1981). Chemotaxis, osmotic avoidance, thermotaxis, and dauer formation are all mediated by the amphids (for review, see Bargmann 1993; Bargmann and Mori, this volume). The sensory terminals of the exposed neurons consist of nonmotile cilia. A large number of mutations disrupt cilium morphogenesis, resulting not only in behavioral defects, but also in failure to absorb fluorescent dyes from the environment (Perkins et al. 1986). Of mutants in 13 genes identified as amphid dye-filling-defective (Dyf), 10 are also dauer-defective (Starich et al. 1995). The remaining three genes in the set have only one mutant allele each, so it is possible that stronger alleles might be dauer-defective as well. The Daf-c mutants daf-11 and daf-21 are also defective in chemotaxis (Vowels and Thomas 1994). Their gene products are thought to function in the amphid sensory cilia because the Dyf mutants are epistatic to mutations in these genes (Vowels and Thomas 1992; Thomas et al. 1993).

Mutants defective in both dauer larva formation and chemotaxis have been examined ultrastructurally, and a variety of morphological abnormalities in the dendritic endings of anterior and posterior sensory neurons were observed (Lewis and Hodgkin 1977; Albert et al. 1981; Perkins et al. 1986). Of all the anterior sense organs examined in mutants that are defective in both dauer larva formation and chemotaxis, only the amphids consistently displayed abnormal morphology. Mutations in one Daf-d gene, daf-6 , affect the amphid sheath cell so that it closes the amphid channel and creates a barrier between the amphid neurons and the environment (Albert et al. 1981). Genetic mosaic analysis also indicated that the daf-6 defect is localized to the sheath cells (Herman 1987). Killing these cells in wild-type L1 larvae with a laser prevents response to dauer-inducing pheromone, whereas killing the analogous phasmid sheath cells in the tail has no effect (Vowels and Thomas 1994).

Although a high level of pheromone is the primary dauer-inducing stimulus, under nondauer-inducing conditions, the source and quantity of food become more important and can affect mutant phenotypes, as well as wild-type development. E. coli strain OP50, a uracil auxotroph, was chosen by Brenner (1974) as a convenient laboratory food source for C. elegans because it promoted optimal development of the organism, and its reduced growth on uracil-limiting agar medium facilitated visual identification of even small larvae. By limiting the available bacterial food supply even further, Cassada and Russell (1975) were able to induce dauer larva formation in wild type at a population density that did not induce dauer larva formation in the presence of more food.

Recent results indicate that not all bacterial strains have the same sensory or nutritive value, which apparently can explain discrepancies in the penetrance (percent dauer formation) reported for some Daf-c mutant strains at 15°C (Swanson and Riddle 1981; Thomas et al. 1993). The apparent phenotypic differences do not reflect genetic divergence between laboratory sublines but are due to subtle differences in culture conditions. Given the importance of the ratio of pheromone to food in the formation of dauer larvae, the strain of bacteria used as a food source must be taken into account when experimental results are interpreted, or when comparing the results obtained in different laboratories. Indeed, raising animals at 27°C rather than 25°C induces constitutive dauer formation in some mutants that cannot otherwise be scored as Daf-c (Malone et al. 1996). Such observations underscore the importance of environmental conditions in dauer larva formation and raise the possibility that other cues also affect the process.

B. Genetic Pathways

Studies of the phenotypes resulting from combinations of temperature-sensitive daf-c and nonconditional daf-d mutations have allowed the daf genes to be ordered into complex branched pathways (Riddle et al. 1981; Vowels and Thomas 1992; Thomas et al. 1993; Gottlieb and Ruvkun 1994; Larsen et al. 1995; Grenache et al. 1996). A unified version of the pathways is shown in Figure 3. Tests for genetic epistasis ask whether a given daf-c; daf-d double mutant forms dauer larvae or grows to the adult at 25°C (Riddle et al. 1981).

The first gene in the pathway, daf-22 , is involved in production of the dauer-inducing pheromone (Golden and Riddle 1985). daf-22 is unique among daf-d mutants because its phenotype can be cured by addition of exogenous pheromone to the culture medium. daf-6 also acts at an early phase of the response (Vowels and Thomas 1992). As previously mentioned, the amphidial pores in daf-6 mutants are blocked, preventing sensation of pheromone.

Downstream from daf-6 , two parallel genetic pathways corresponding to two partially distinct neuronal signaling pathways have been identified (Thomas et al. 1993; Schackwitz et al. 1996). One branch includes daf-11 and daf-21 mutations. Dyf mutations that affect development of the sensory cilia suppress the temperature-sensitive daf-c mutations in daf-11 and daf-21 , indicating that mutations in these two genes require normal cilium structure to induce dauer larva formation in the presence of food (Vowels and Thomas 1992; Thomas et al. 1993). The second branch of daf-c genes includes daf-1 , -4, -7, -8, and -14, which are not suppressed by dyf mutations. These genes encode components of the TGF-β signaling pathway that are expressed in some sensory neurons (see below). These daf-c mutations are fully suppressed by mutations in daf-3 and daf-5 , which only partially suppress the daf-c mutations in daf-11 and daf-21 .

The parallel nature of these two branches of the pathway was proposed on the basis of the phenotypes of animals doubly mutant for daf-11 or daf-21 and any of the second group of mutations ( daf-1 , -4, -7, -8, and -14). These animals form 100% dauer larvae at all temperatures, a Daf-c phenotype more severe than observed for the single mutants at 15°C (Thomas et al. 1993). Thus, the two groups of genes have partially redundant activities that can compensate for one another under some conditions, a model that conforms to their apparent parallel activities in different neurons (Schackwitz et al. 1996).

The two branches of the dauer formation pathway converge again on daf-12 , which fully suppresses the Daf-c phenotype of both groups of upstream mutations. The daf-12 gene encodes a nuclear hormone receptor that is required for wild-type or daf-c animals to form dauer larvae (Yeh 1991). daf-c mutations in daf-2 and age-1 have antagonistic activity with daf-12 (Gottlieb and Ruvkun 1994; Larsen et al. 1995). The age-1 gene encodes a phosphatidylinositol-3-OH (PI3) kinase catalytic subunit (Morris et al. 1996) of the class that is activated in tyrosine kinase signaling pathways (Kapellar and Cantley 1994). At this point, it is unclear whether the link between age-1 phosphatidylinositol signaling and TGF-β signaling ( daf-1 , -4, and -7) is direct or indirect.

A simple model relating the genetics to the physiology is that the two parallel groups of daf-c genes detect and relay environmental signals to the animal. At low temperatures, either parallel pathway can operate independently, but at high temperatures, both are required to prevent dauer larva formation. The ultimate result of these pathways is to control the relative activities of the daf-12 and daf-2 genes. When pheromone levels are high, DAF-12 is active, DAF-2 is inactive, and dauer larvae form. When pheromone levels are low, DAF-12 is inactive, DAF-2 is active, and nondauer development ensues. Interestingly, daf-2; daf-12 double mutants often arrest as young larvae, as though the activity of one of these two genes is necessary to get the animal through larval development, either on the dauer pathway or on the nondauer pathway (Vowels and Thomas 1992; Larsen et al. 1995).

daf-c mutations in age-1 and daf-2 , but not other daf-c genes, double adult longevity and decrease sensitivity to UV irradiation. All three phenotypes are suppressed by a mutation in daf-16 (Kenyon et al. 1993; Larsen et al. 1995; Murakami and Johnson 1996). Different combinations of daf-12 and daf-2 alleles either enhance or suppress the extended life span, leading Larsen et al. (1995) to propose a direct interaction between daf-2 and daf-12 gene products. The age-1 gene has long been studied by means of one mutant allele (hx546) that extends longevity by 50% (Friedman and Johnson 1988a; Johnson et al. 1991), but this mutant lacks a Daf-c phenotype unless exposed to temperatures above the normal growth range (Malone et al. 1996). Only recently has age-1 been found to be allelic with daf-23 (Malone et al. 1996; Morris et al. 1996).

The genetic pathways diverge after daf-3 and daf-5 , with daf-12 activity required for dauer formation, but only daf-3 and daf-5 are required for expression of other aspects of the daf-1 group Daf-c mutant phenotype, including display of an L1 surface antigen on later larval stages (Grenache et al. 1996) and defects in behavior and appearance (Trent et al. 1983; Thomas et al. 1993).

Why is the dauer formation pathway so complex, and why were discrepancies observed between different laboratory groups? Many of the daf gene products probably affect response to a subset of environmental cues. The exquisite sensitivity of dauer formation to environmental conditions and the partial redundancy of the neurons that sense these conditions mean that different genes may shift in importance under different culture conditions. Furthermore, gene interactions are not always easily interpreted. First, many of the mutations used for epistasis tests were probably not null, which could weaken the interpretation for those genes. Second, some daf-c; daf-d double-mutant combinations clearly show intermediate phenotypes (coexpression), whose interpretation will necessarily be ambiguous (Vowels and Thomas 1992; Larsen et al. 1995). The relationship between these dauer-like forms and normal dauer larvae is not obvious. For example, if the dauer-like forms do not undergo dauer sensory remodeling, they might respond differently to sensory cues than do true dauer larvae. Certain daf-c; daf-d double mutants form dauer or dauer-like larvae transiently and then resume development (Vowels and Thomas 1992; Larsen et al. 1995). The appearance of transient dauer larvae makes the timing of scoring important.

An additional problem is that in many cases, the double-mutant phenotype is neither fully daf-c nor fully daf-d (partially penetrant suppression), which raises the question of the amount of suppression necessary to define an epistatic relationship. Several conventions have been used to define an epistatic relationship. For example, Riddle et al. (1981) considered 20% suppression of a Daf-c phenotype (scored in asynchronous populations after 3 days of growth) sufficient to place a Daf-d mutant downstream in the pathway in the absence of any information about which alleles might be null. Larsen et al. (1995) used 66% suppression of entry into the dauer stage 50 hours from the time eggs were laid (scoring synchronously developing populations over a 5-day period) as the minimum to define an epistatic Daf-d mutant. A useful strategy for epistasis tests has been to construct daf-c; daf-d strains that do not carry other genetic markers, and score synchronous populations for growth or dauer larva formation at a time when transient dauer formation can be observed (Vowels and Thomas 1992; Larsen et al. 1995).

C. Sensory Neurons with Overlapping Functions

Laser microsurgery has been used to identify specifically which amphid neurons participate in dauer signaling. In these studies, corresponding neurons in both amphids were killed. When neurons ADF, ASG, and ASI were killed early in the L1 stage, wild-type animals developed into dauer larvae at 20°C in abundant food, mimicking the Daf-c phenotype (Bargmann and Horvitz 1991b). However, these dauer larvae recovered within 1 day unless the ASJ neurons were also killed. Hence, under conditions favoring nondauer development, amphid neurons ADF, ASI, and ASG signal to inhibit dauer formation, and ASJ signals exit from the dauer stage. Mutants that fail to develop normal dendritic cilia are Daf-d because they fail to respond to pheromone (Golden and Riddle 1984a). It is argued that pheromone inhibits the promotion of nondauer development by ADF, ASI, and ASG. Killing the ADF, ASG, ASI, and ASJ neurons in Daf-d mutants resulted in a Daf-c phenocopy in the case of daf-22 and daf-10 , but not in the case of daf-3 , daf-5 , daf-12 , or daf-16 (Bargmann and Horvitz 1991b). This suggests that the latter genes act downstream from these cells, and they are required for dauer formation in the absence of the dauer-inhibiting neuronal signal (Fig. 3).

Figure 3. Genetic pathway for dauer larva formation, assembled from Malone et al.

Figure 3

Genetic pathway for dauer larva formation, assembled from Malone et al. (1996), Gottlieb and Ruvkun (1994), Larsen et al. (1995), and Grenache et al. (1996). The pathway is drawn to show wild-type gene functions that stimulate (more...)

Bargmann and Horvitz (1991b) observed that the function of either ASI or ADF alone was sufficient to prevent dauer formation at 20°C (i.e., in two sets of L1 larvae, ADF, ASG, and ASJ were killed and ASI, ASG, and ASJ were killed, respectively). The ASG neuron, on the other hand, only prevented 40% of ADF + ASI + ASJ killed L1 larvae from developing into dauer larvae. Hence, under standard laboratory conditions at 20°C, either ADF or ASI cell function is sufficient for continuous development, and ASG has an overlapping function, perhaps as an evolutionary fine-tuning mechanism for adapting pheromone response to different environments. Indeed, wild-type Caenorhabditis strains from different environments have been observed to differ quantitatively in their response to pheromone (Fodor et al. 1983).

Further evidence for a role of ADF in dauer formation was provided by a phenotypic analysis of osm-3 osmotic avoidance mutants (Shakir et al. 1993a). osm-3 encodes a kinesin-like protein that is expressed in chemosensory neurons during neurogenesis (Tabish et al. 1995). Severe osm-3 mutants are deficient in amphid dye filling and they are dauer-defective (always signaling nondauer development). Weaker alleles retain some ability to concentrate fluorescein isothiocyanate (FITC) in ADF neurons. The ability of ADF to absorb dye correlated with the ability to form dauer larvae in starved, crowded cultures. These results indicate that ADF is likely to be important under dauer-inducing conditions at 20°C, but why was pheromone-induced inhibition of ADF alone sufficient to permit dauer larva formation by the osm-3 mutant? One might have predicted that ASI (and ASG), which did not absorb dye, would have constitutively signaled nondauer development because they remained pheromone-insensitive.

The difference in assay conditions may account for the apparent difference in the function of ASI. Bargmann and Horvitz (1991b) assayed constitutive dauer formation of laser-operated animals in abundant food, whereas Shakir et al. (1993a) assayed dauer formation in starved cultures. ASI may be less important under these conditions than under unstarved conditions, or dye filling might not be a reliable test of pheromone-sensing ability of ASI. It is intriguing to consider that the relative roles of ADF, ASI, and ASG in signaling might change with food abundance. In abundant food, response to high pheromone levels might require inhibition of both ADF and ASI, but when food is depleted, inhibition of ADF might be sufficient to permit dauer formation because the competitive food signal is reduced.

Another possibility that could account for the difference between laser microsurgery results and mutant analysis concerns the late time of cell death after laser killing (early to mid L1). If ADF not only regulated dauer formation, but also signaled ASI in the embryo or L1 to make ASI competent to signal nondauer development, a weak osm-3 mutation could eliminate both functions, whereas laser killing might eliminate only the later function.

The experiments of Bargmann and Horvitz (1991b) were designed to detect dauer-inhibiting neurons, not dauer-promoting neurons. Thomas et al. (1993) proposed parallel pathways for dauer formation based on genetic analysis. Genes like daf-1 , which encode components of TGF-β signaling, were proposed to define one pathway, whereas the daf-d genes affecting chemotaxis and the daf-c genes daf-11 and daf-21 , which also affect chemotaxis (Vowels and Thomas 1994), were proposed to define a parallel pathway. Schackwitz et al. (1996) hypothesized that the Daf-c phenotype of daf-11 mutants results from the activity of dauer-promoting neurons and that loss of daf-11 function activates these neurons, resulting in constitutive dauer formation. Indeed, killing ASJ, but not other amphid neurons, resulted in suppression of the daf-11 and daf-21 Daf-c phenotypes. This suppression was enhanced by killing neurons ASK and ADL. In contrast, killing ADF and ASI enhanced the Daf-c phenotype of daf-11 and daf-21 mutants. Finally, killing ASJ in wild-type larvae impaired their response to the dauer-inducing pheromone.

The results of the neuron ablation studies suggest that parallel pathways for control of dauer larva formation function in different sets of amphid neurons. The model proposed by Schackwitz et al. (1996) is that mutations in the daf-1 group (see Fig. 3) disrupt the inhibition of dauer formation by ADF, ASI, and ASG. Laser-operated daf-1 and daf-7 mutants, in which ADF, ASI, and ASG were left intact, still expressed the Daf-c phenotype. In contrast, the dauer-promoting neuron, ASJ, is required for the Daf-c phenotype of daf-11 and daf-21 mutants to be expressed. In low levels of pheromone, the dauer-inhibiting neurons ADF, ASI, and ASG are active, and the dauer-promoting neuron ASJ is inactive. Conversely, in high levels of pheromone, ADF, ASI, and ASG are inactive and ASJ is active. Such parallel pathways have been found in the vertebrate visual system, where they are thought to enhance contrast sensitivity and improve rapid response to light changes (Schiller et al. 1986). Similarly, the fidelity of the dauer/nondauer switch may be enhanced by the parallel neural pathways.

A dauer-promoting role for ASJ in the L1 stage is surprising because these neurons function to promote dauer recovery (Bargmann and Horvitz 1991b). One possibility is that ASJ changes its function between the L1 and dauer stages. In this case, daf-11 and daf-21 would inactivate ASJ to help prevent dauer formation in low pheromone, but in the dauer larva, these genes would activate ASJ in low pheromone to promote recovery (Schackwitz et al. 1996). The temperature-sensitive daf-11 mutants, in fact, recover very poorly from the dauer stage when shifted to lower temperature (Vowels and Thomas 1994). An answer to the puzzle of ASJ functions may emerge from an understanding of cross-talk with neurons of opposite function, such as ASI.

D. Gene Products and Expression Patterns

1. TGF-β Signaling

The first daf gene cloned, daf-1 , had the hallmarks of a cell surface receptor with a signal peptide, a single transmembrane segment, and a cytoplasmic region with a protein serine/threonine kinase domain (Georgi et al. 1990). These and other sequence features have subsequently proved to identify it as a member of the TGF-β receptor family (Kingsley 1994; Massagué et al. 1994). The daf-4 gene encodes another receptor serine/threonine kinase corresponding to a type II TGF-β receptor (Estevez et al. 1993). The type II TGF-β receptor binds its ligand, recruits the type I receptor into a heterodimer, and activates the type I receptor by phosphorylation (Wrana et al. 1994). The type I receptor then phosphorylates downstream components in the signal transduction pathway. The daf-4 receptor was found to bind human bone morphogenetic proteins BMP-2 and BMP-4 when expressed in monkey COS cells (Estevez et al. 1993). daf-4 did not bind TGF-β or activin, and the daf-1 receptor did not bind any ligand tested. The DAF-1 protein is the putative type I receptor because it possesses a “GS domain” similar to the site of phosphorylation on the TGF-β type I receptor, and it has a smaller extracellular domain than does the DAF-4 receptor, as would be expected for a type I subunit. Type I TGF-β receptors do not bind ligand in the absence of the type II receptor. Biochemical tests for direct interaction of DAF-1 and DAF-4 proteins have yet to be done, so the proposed relationship is based on the genetic results and by analogy with other receptors in the family.

The DAF-4 receptor participates directly in dauer/nondauer signaling, rather than playing an indirect part in neural development. Heat shock treatment of transgenic daf-4 mutant animals carrying a heat-shock-inducible daf-4 (+) cDNA construct demonstrated that expression of daf-4 in the L1 stage was sufficient to rescue the Daf-c phenotype, and expression in the dauer stage induced exit (Estevez et al. 1993).

The putative natural ligand for DAF-4 in its role in nondauer development is the DAF-7 protein, which is a novel member of the TGF-β superfamily (Ren et al. 1996). The daf-7 gene encodes a secreted 350-amino-acid precursor that is processed at a dibasic proteolytic cleavage site to produce a 116-amino-acid mature hormone subunit. Such processing has been confirmed by amino-terminal sequencing of recombinant DAF-7 protein expressed in insect cells (P. Ren and D. Riddle, unpubl.). In this protein family, the biologically active protein is a homodimer or heterodimer joined by a disulfide bridge (Roberts and Sporn 1990). DAF-7 has sequence similarity to the BMP group of ligands, which includes the Drosophila decapentaplegic (dpp) gene product (Padgett et al. 1993), but it lacks several amino acids invariant in this group. It also has sequence similarity to TGF-β, including two conserved cysteines found in TGF-β and activin, but not in other superfamily members (Kingsley 1994). It is conceivable that a protein similar to DAF-7 was the evolutionary precursor to both TGF-β and BMPs, which now represent structurally divergent subfamilies (Fig. 4).

Figure 4. Sequence relationships between ten members of the TGF-β superfamily shown by phylogenetic analysis.

Figure 4

Sequence relationships between ten members of the TGF-β superfamily shown by phylogenetic analysis. The ligand region alone (amino acids 250–350 of DAF-7) was used, starting (more...)

Northern blot analysis has shown that daf-1 and daf-4 mRNAs are present in all developmental stages (M. Estevez and D.L. Riddle, unpubl.), but daf-7 mRNA is regulated. The daf-7 mRNA is rare but most abundant in the L1 stage under conditions favoring nondauer development (Ren et al. 1996). Its steady-state level decreases in the L2, but the RNA is nearly absent in the egg, L2d, and later stages. Control of daf-7 transcription by food and/or pheromone is one mechanism by which the dauer/nondauer decision is made. Mutations in daf-1 , daf-4 , and daf-7 all result in a Daf-c phenotype, and they define one step in the genetic pathway (Fig. 3), as would be expected for a ligand and its receptor. A low-pheromone environment activates continuous development by promoting secretion of the daf-7 ligand. Ligand binding would, in turn, activate the DAF-1/DAF-4 receptor.

The repression of daf-7 transcription as a result of exposure to dauer-inducing pheromone was inferred from observation of transgenic animals carrying green fluorescent protein (GFP) reporter constructs (Ren et al. 1996; Schackwitz et al. 1996). The reporter gene was expressed in ASI neurons and preferentially under conditions favoring growth. Ren et al. (1996) observed GFP expression in hermaphrodite larvae from 4 to 5 hours after hatching, through the four larval stages and in adults, whereas Schackwitz et al. (1996) observed only low levels of adult expression. Neither daf-7 mRNA nor GFP expression was observed in embryos, suggesting that daf-7 does not have a role in sensory neuron development. During pheromone-induced dauer formation at 25°C, GFP expression was strongly suppressed in L1 larvae, both in the percentage of animals expressing GFP and in the intensity of fluorescence. As animals entered the dauer stage, weak GFP expression was no longer detected. The GFP expression results suggest that during the L1 stage, the amphid ASI neurons use DAF-7 to signal nondauer development in low pheromone and then again use DAF-7 to signal recovery from the dauer state in response to fresh food at higher growth temperatures (see Ren et al. 1996).

As described above, laser microsurgery performed at 20°C indicated that neurons ASI, ADF, and ASG were apparently redundant. The temperature itself may account for the difference between neuron ablation and GFP expression results or the surgery may not have killed ASI cells soon enough to prevent daf-7 expression, so that additional neurons (perhaps DAF-7 target cells) had to be disrupted to cause dauer formation. Alternatively, unknown transcriptional or translational regulatory elements for expression of GFP in other neurons may be absent from the reporter constructs.

In contrast with daf-7 , a daf-1::gfp fusion is expressed in more than 50 cells in the body. Many of these cells are interneurons, but amphid neurons also fluoresce in the transgenic animals (Fig. 5). The amphid cells that express GFP include ASI (C. Gunther and D. Riddle, unpubl.). Expression in ASI raises the possibility of an autocrine loop in which daf-7 progressively stimulates its own expression, although Ren et al. (1996) found that daf-7 (+) activity was not required for continuous reporter gene expression in transgenic animals.

Figure 5. GFP expression under control of the daf-1 promoter.

Figure 5

GFP expression under control of the daf-1 promoter. The head of the L1 larva is to the left; dorsal side is up. Genomic DNA upstream of the (more...)

The secretion of a TGF-β-like ligand by a sensory neuron, presumably to activate a receptor on interneurons (some of which are synaptically connected via the circumpharyngeal nerve ring), might seem surprising. As precedent for this type of neural function, activin has been implicated as a neurotransmitter/neuromodulator in central neural pathways in rats for the release of oxytocin in response to suckling (Sawchenko et al. 1988). DAF-7 might act as a neuromodulator to change the activity of neural circuits over time. Sufficient DAF-7 signaling through the L1 stage could raise the system to a threshold committing the animal to nondauer development by the L1 molt. A second use of DAF-7 could be to directly modulate the activity of other chemosensory neurons that may have either synergistic or competitive activities. If, for example, DAF-7 secretion by ASI beyond a certain threshold inactivated ASJ, it would ensure a coherent signal by silencing the competitive neurons.

2. Cyclic Nucleotide Signaling

Recently, cyclic nucleotide signaling has been implicated in the dauer/ nondauer switch by the cloning of additional genes for which mutant alleles convey a partial or complete Daf-c phenotype as well as other sensory defects. The daf-11 gene encodes a transmembrane guanylyl cyclase (D. Birnby and J. Thomas, pers. comm.); activating alleles of gpa-2 and gpa-3 , which encode α subunits of heterotrimeric G proteins, convey a dominant Daf-c phenotype ( Zwaal et al. 1997); and the sensory mutant genes tax-2 and tax-4 , which have a weak Daf-c phenotype, encode subunits of a cyclic-nucleotide-gated channel (Coburn and Bargmann 1996; Komatsu et al. 1996).

Heterotrimeric G proteins are components of seven transmembrane hormone and sensory receptor pathways. GTP-bound G-α proteins can regulate cGMP phosphodiesterase or they can activate adenylyl cyclase, and the resulting rise in cAMP can result in activation of protein kinase A. GTP-bound G-α can also activate phospholipase C, a key component in 3-phosphoinositide signaling pathways. The C. elegans gpa-2 and gpa-3 genes were cloned on the basis of their sequence similarity with mammalian G-α subunits (Zwaal et al. 1997). Expression of presumptive dominant mutations in transgenic animals resulted in a Daf-c phenotype, and this phenotype was suppressed by daf-d cilium structure mutations (just as daf-c mutations in daf-11 and daf-21 are suppressed), suggesting that GPA-3 functions in the amphid cilia. This location would be expected for a G protein coupled to the pheromone receptor. gpa-3 reporter gene fusions were, indeed, expressed in amphid neurons. Activated gpa-3 mutant strains were also Dyf, indicating that constitutive GPA-3 activity is deleterious to neuron function.

Loss of gpa-2 function results in dauer recovery in the presence of pheromone, but gpa-3 loss-of-function mutants retained pheromone responsiveness and were not able to recover under the same conditions. Hence, the neurons expressing gpa-2 are involved in both dauer formation and recovery, whereas amphid neurons expressing gpa-3 are involved only in dauer formation. Loss of gpa-2 or gpa-3 function results in reduced sensitivity to pheromone in the L1 stage and, under certain conditions, an altered food response. Mutations in daf-11 and daf-21 are partially suppressed by these loss-of-function mutations, but mutations in daf-1 or daf-8 are not, indicating that gpa-2 and gpa-3 may act downstream or in parallel to daf-11 and daf-21 and either upstream or in parallel to daf-1 and daf-8 (Zwaal et al. 1997). The daf-11 and daf-21 gene products might act through G proteins, or this suppression could be indirect, involving other neurons. gpa-2 and gpa-3 seem to be involved primarily in the sensory response to dauer-inducing pheromone, since inactivation of these genes interferes with the pheromone response, and constitutively active GPA-2 and GPA-3 lead to a Daf-c phenotype.

The tax-2 and tax-4 genes encode subunits of a cyclic-nucleotide-gated channel (Komatsu et al. 1996; Coburn and Bargmann 1996) that are required for expression of the daf-11 (guanylyl cyclase) mutant phenotype. Cyclic-nucleotide-gated channels are involved in olfactory and photosensory transduction in vertebrates (Nakamura and Gold 1987). The ASJ sensory neurons exhibit defects in axon outgrowth in tax-2 (Coburn and Bargmann 1996) and in daf-11 mutants (Bargmann and Mori, this volume). These results show that cyclic nucleotides are an important intracellular messenger in C. elegans sensory transduction and in chemosensory axon morphology.

E. Pleiotropy

1. Multiple Functions for daf Genes

Most of the dauer formation genes also have effects on other aspects of C. elegans biology. daf-11 and daf-21 exhibit chemotaxis and thermotaxis defects, consistent with their proposed role in sensory function (Vowels and Thomas 1994; Bargmann and Mori, this volume). The Dyf mutants, which have developmental defects in the sensory cilia, also have correspondingly broad defects in chemosensory function; dyf mutations suppress daf-11 and daf-21 mutations. In addition to the G-protein mutants discussed above, mutations in tax-2 and tax-4 also suppress daf-11 and daf-12 mutations, suggesting a role for these tax genes in thermotaxis, chemotaxis, and dauer formation (Bargmann and Mori, this volume).

Multiple roles in developmental or behavioral processes may not always be so obvious. For example, most Daf-c mutants are defective in egg laying (Egl), but a large number of additional Egl mutants, including egl-4 , lack a Daf-c phenotype (Trent et al. 1983). However, an egl-4 mutant is hypersensitive to dauer-inducing pheromone, indicating that this gene may play some part in promoting nondauer development that is not detectable under normal laboratory conditions (Golden and Riddle 1984a). It is now known that mutations in more than 50 genes, including egl-4 , convey a Daf-c phenotype, but only when combined in specific double mutants (synthetic Daf-c mutants) (I. Katsura, pers. comm.). Such genes include those required for chemotaxis, locomotion, and defecation. For example, an unc-31 ; aex-3 double mutant is strongly Daf-c (Avery 1993a), whereas the single mutants are not.

Recently, mig-7 mutants have been shown to be novel alleles of daf-12 . daf-12 function is required not only for dauer formation, but also for proper guidance of gonadal development by the distal tip cells and for proper development of the hypodermis (Antebi et al., this volume). Different daf-12 alleles have different effects on these processes. It is possible that daf-12 coordinates different developmental processes in time, like the heterochronic genes. The gonadal distal tip cells stop migrating at the dauer-specific molt in response to the signal for dauer morphogenesis, and hypodermal cell divisions that would normally occur after the molt do not occur. DAF-12 may also synchronize gonadal and hypodermal development at other larval stages. It is interesting that certain daf-12 mutations also double the longevity of certain daf-2 mutants (Larsen et al. 1995). Mutations in age-1 and daf-2 , but not other daf-c genes, increase adult longevity, increase thermotolerance, and decrease sensitivity to UV irradiation (Kenyon et al. 1993; Larsen et al. 1995; Lithgow et al. 1995; Murakami and Johnson 1996; Kenyon, this volume). One hypothesis is that daf-2 mutants express genes for efficient life maintenance in the adult that are normally only expressed in the dauer stage, and certain daf-12 alleles enhance this expression when combined with daf-2 mutations.

2. Multiple Functions for TGF-β Signaling

Mutants in the daf-1 group have an adult egg-laying defect (Trent et al. 1983) and dark intestinal cells, and individuals tend to accumulate ("clump") in groups near the edge of the bacterial lawn (Thomas et al. 1993). The mutants also express an L1 surface antigen on later larval stages (Grenache et al. 1996). These phenotypes are suppressed by daf-3 and daf-5 mutations, indicating that these two genes function in every aspect of the function of the daf-1 group. However, these phenotypes are not suppressed by daf-12 , which may be more selective for the dauer pathway (see Fig. 3).

Mutants in daf-4 , the type II TGF-β receptor, are not only Daf-c and Egl, but also small (Sma) with abnormal male tail phenotypes (Savage et al. 1996). The daf-1 (putative type I receptor) mutants possess the Daf-c and adult Egl phenotypes, but they are not Sma or defective in male mating. The presence of mRNAs for both these receptors in all developmental stages, including the adult, is consistent with their use in both larval and adult behaviors. The similarity of the Daf-4 adult phenotypes to sma-2 , sma-3 , and sma-4 mutants (which are not Daf-c) led Savage et al. (1996) to examine these genes as potentially encoding constituents of TGF-β signaling unrelated to dauer formation. These three genes encode members of the Smad family of proteins, the founding member of which is the product of the Mothers against decapentaplegic (Mad) gene from Drosophila (Sekelsky et al. 1995). In Drosophila, the Mad and decapentaplegic (dpp) phenotypes are similar, and loss-of-function mutations in Mad enhance the embryonic dorsal-ventral patterning defects and adult appendage defects resulting from dpp mutations. Vertebrate homologs of MAD include the human DPC4 (Smad4) tumor suppressor, which associates with either the Smad1 or Smad2 proteins in response to BMP and TGF-β, respectively (Lagna et al. 1996). Smad proteins contain a conserved carboxy-terminal transcriptional activation domain and are transported to the nucleus upon phosphorylation. Smad proteins may also form complexes with other transcription factors to activate activin-dependent transcription (Chen et al. 1996).

3. Dauer Signaling Downstream from DAF-4

Smad proteins act downstream from TGF-β family receptors to activate transcription in both vertebrate and invertebrate systems. Which genes might encode the DAF counterparts to the three SMA proteins? The most likely candidates are the daf-c genes associated with daf-1 and daf-4 in the genetic pathway (see Fig. 3). In fact, three daf genes have recently been found to encode Smad family members: daf-3 (A. Koweek et al., pers. comm.), daf-8 (A. Estevez and D. Riddle, unpubl.), and daf-14 (T. Inoue and J. Thomas, pers. comm.). Mutations in daf-8 and daf-14 are Daf-c, but daf-3 mutants are Daf-d. Given the known interactions between DPC4 and Smad1 or Smad2 (Lagna et al. 1996), it is attractive to speculate that DAF-8 and DAF-14 would inactivate DAF-3 (which is required for dauer formation) to signal nondauer development when activated by DAF-7 signaling through the DAF-1/DAF-4 receptor. This would be consistent with the observed epistasis of daf-3 mutations to daf-8 and daf-14 mutations. Alternatively, the daf-c Smad genes might be in a pathway to activate nondauer development, and the daf-3 Smad might be in a parallel pathway to activate dauer development. Blockage of either pathway would lead to constitutive activation of the other, but the double mutant would activate nondauer development by a mechanism not requiring daf-8 or daf-14 .

At least three Smad family proteins are involved in transducing daf-4 signals in the daf pathway and three others are in the sma pathway. The DAF-4 receptor apparently does not use the DAF-7 ligand or the DAF-1 receptor as its type I partner for specifying adult body size and male tail development. Such a type I receptor may be found among other sma genes. Activation of the distinct type I receptor would then result in activation of SMA-2, SMA-3, and SMA-4 proteins. The DAF-1 protein and the downstream Smad proteins in the dauer pathway are activated by the DAF-7 ligand to signal nondauer development, but these proteins apparently have no significant role in the sma and male tail pathways.

Copyright © 1997, Cold Spring Harbor Laboratory Press.
Bookshelf ID: NBK20185


  • PubReader
  • Print View
  • Cite this Page

Recent Activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...