ORGANISMS make complex behavioral and devel- are specialized for survival and dispersal in harsh enviopmental

Transcription

1 Copyright 2000 by the Genetics Society of America egl-4 Acts Through a Transforming Growth Factor- /SMAD Pathway in Caenorhabditis elegans to Regulate Multiple Neuronal Circuits in Response to Sensory Cues Susan A. Daniels,*,1 Michael Ailion,,1 James H. Thomas, and Piali Sengupta* *Department of Biology and Volen Center for Complex Systems, Brandeis University, Waltham, Massachusetts and Molecular and Cellular Biology Program of the University of Washington and Fred Hutchinson Cancer Research Center and Department of Genetics, University of Washington, Seattle, Washington Manuscript received April 6, 2000 Accepted for publication May 15, 2000 ABSTRACT Sensory cues regulate several aspects of behavior and development in Caenorhabditis elegans, including entry into and exit from an alternative developmental stage called the dauer larva. Three parallel pathways, including a TGF- -like pathway, regulate dauer formation. The mechanisms by which the activities of these pathways are regulated by sensory signals are largely unknown. The gene egl-4 was initially identified based on its egg-laying defects. We show here that egl-4 has many pleiotropies, including defects in chemosensory behavior, body size, synaptic transmission, and dauer formation. Our results are consistent with a role for egl-4 in relaying sensory cues to multiple behavioral and developmental circuits in C. elegans. By epistasis analysis, we also place egl-4 in the TGF- -like branch and show that a SMAD gene functions downstream of egl-4 in multiple egl-4-regulated pathways, including chemosensation. ORGANISMS make complex behavioral and devel- are specialized for survival and dispersal in harsh enviopmental decisions on the basis of sensory cues in ronmental conditions and can recover and resume retheir environment. The nematode Caenorhabditis elegans productive growth when conditions improve. responds to multiple types of sensory signals, including The decision to enter into or recover from the dauer chemical, mechanical, and thermal stimuli (for recent stage is made through the assessment of multiple paralreviews, see Driscoll and Kaplan 1997; Troemel lel sensory and developmental inputs. A high concentra- 1999). Responses to these stimuli are mediated by well- tion of a constitutively produced pheromone signals defined circuits consisting of ciliated sensory neurons, increased population density and is the primary chemointerneurons, and motor neurons (White et al. 1986). sensory signal regulating dauer formation (Golden and Functions of individual sensory neurons have been de- Riddle 1982, 1984b, 1985). In addition, high temperafined by laser killing experiments, and it has been shown ture and low levels of food indicate adverse conditions that distinct subsets of sensory neurons are required for and also promote dauer formation (Golden and Ridthe response to each type of stimulus (Chalfie et al. dle 1984a,b). Thus, regulation of dauer entry and exit 1985; Bargmann and Horvitz 1991a; Bargmann et al. requires integration of information from multiple sen- 1993; Kaplan and Horvitz 1993; Mori and Ohshima sory pathways and provides an excellent model system 1995). in which to investigate several aspects of neuronal func- Several aspects of C. elegans behavior and development tion. are regulated by environmental signals. Chemical cues Three signaling pathways that act in parallel to regudirect movement toward sources of food and away from late dauer formation have been defined (Figure 1; Vowtoxic compounds. Behaviors such as locomotion, pha- els and Thomas 1992; Thomas et al. 1993; Gottlieb ryngeal pumping, defecation, foraging, and egg laying and Ruvkun 1994; Riddle and Albert 1997). Most are also modulated by sensory cues (Horvitz et al. 1982; major players in each of these pathways have been iden- Avery and Horvitz 1990; Thomas 1990; Liu and Thomas tified by studying mutants that either enter the dauer 1994). In addition, environmental stimuli regulate de- state inappropriately under noninducing conditions velopmental decisions such as entry into and exit from [dauer-formation constitutive (Daf-c)], or fail to enter an alternative third larval stage called the dauer larva the dauer state under inducing conditions [dauer-for- (for review, see Riddle and Albert 1997). Dauer larvae mation defective (Daf-d)]. The group I Daf-c genes are thought to act by activating the ASJ chemosensory neurons among others to promote dauer formation under Corresponding author: Piali Sengupta, Department of Biology and inducing conditions (Vowels and Thomas 1992; Thomas Volen Center for Complex Systems, Brandeis University, 415 South St., Waltham, MA sengupta@brandeis.edu et al. 1993; Schackwitz et al. 1996). The group II 1 These authors contributed equally to this work. Daf-c genes constitute a transforming growth factor- Genetics 156: (September 2000)

2 124 S. A. Daniels et al. Figure 1. Parallel pathways mediate dauer formation. Genes described in this work are included in each pathway. GC, guanylyl cyclase; Hsp90, heat shock protein 90; TGF- R, TGF- receptor; NHR, nuclear hormone receptor; insulin R, insulin receptor; PI3 kinase, phosphoinositide-3- OH kinase. See text for additional details. (TGF- )-like signaling pathway and function to repress pal sensory neurons and genes regulating dauer formation dauer formation under noninducing conditions via the have been identified, it is unclear how multiple ASI, ADF, and ASG neurons (Bargmann and Horvitz sensory and developmental inputs are translated into 1991b; Schackwitz et al. 1996; Riddle and Albert modulation of each of these pathways. Since each 1997; Patterson and Padgett 2000). Under nonin- branch of the dauer pathway is regulated by parallel ducing conditions, the DAF-7 TGF- homolog is produced sensory inputs, there is likely to be functional redun- by the ASI neurons (Ren et al. 1996; Schackwitz dancy among genes that play roles in such regulatory et al. 1996). The DAF-7 signal is transduced via the DAF-4 events. One might also expect that these genes would TGF- type II and the DAF-1 TGF- type I receptors, as function in several aspects of neuronal function and well as the DAF-8 and DAF-14 SMAD proteins (Georgi thus would show pleiotropic mutant phenotypes. et al. 1990; Estevez et al. 1993; Riddle and Albert 1997; The genes unc-31, unc-64, and unc-3 were initially identified Inoue and Thomas 2000). This pathway antagonizes the on the basis of the impaired movement of mutants action of the DAF-3 SMAD protein and the daf-5 gene (Brenner 1974). unc-31 and unc-64 mutants show pleiotropic product (as yet uncloned) to repress dauer development phenotypes consistent with their general effects (Patterson et al. 1997). In a third pathway, an on neuronal function. unc-31 and unc-64 encode a insulin-like ligand(s) represses dauer formation via the CAPS-related protein and a syntaxin homolog, respec- DAF-2 insulin receptor and the AGE-1 phosphatidyli- tively; these proteins are required for calcium-regulated nositol-3-oh kinase (Morris et al. 1996; Kimura et al. secretion (Livingstone 1991; Avery et al. 1993; Ngu- 1997). This pathway is antagonized by the DAF-18 PTEN yen et al. 1995; Ann et al. 1997; Ogawa et al. 1998; phosphatase and the DAF-16 forkhead domain protein Saifee et al. 1998). Mutations in these genes also affect (Larsen et al. 1995; Lin et al. 1997; Ogg et al. 1997; the dauer pathway. While animals singly mutant in one Ogg and Ruvkun 1998; Gil et al. 1999; Rouault et of these genes form very few dauers under noninducing al. 1999). It is equally likely that these pathways act to conditions, double mutant combinations with other promote reproductive growth as opposed to repressing genes result in a strong synthetic Daf-c phenotype (Syn- dauer arrest. As shown in Figure 1, these pathways are Daf; Ailion et al. 1999). The Daf-c phenotype of uncthought to converge at the DAF-12 nuclear hormone 31 and unc-64 mutants is strongly suppressed by muta- receptor (Riddle and Albert 1997; Antebi et al. 1998). tions in daf-16, suggesting that these genes may regulate The signals from these parallel pathways are integrated insulin release in response to sensory and/or metabolic via unknown mechanisms and transduced to result in signals (Ailion et al. 1999). Similarly, unc-3 mutants coordinated developmental changes in multiple tissue also show a number of pleiotropies, including a Syn- types throughout the animal. Daf phenotype (I. Katsura, personal communication; To reflect environmental changes accurately, it is crucial this work). unc-3 encodes an Olf-1/EBF1-like transcriptory that the activities of the three major dauer regula- tion factor that regulates expression of genes such as pathways are appropriately regulated in response daf-7, as well as cholinergic genes in motor neurons (M. to sensory and developmental signals. The sensory cues Ailion and J. H. Thomas, unpublished results; P. Ren pheromone, temperature, and food have been shown and D. Riddle, personal communication; T. Starich to regulate expression of the DAF-7 TGF- ligand (Ren and J. Shaw, personal communication; K. Lickteig and et al. 1996; Schackwitz et al. 1996). Although the princi- D. Miller, personal communication; Prasad et al.

3 egl-4 Regulates Neuronal Functions ). unc-3 functions in the TGF- -mediated pathway were outcrossed a total of four times prior to further characterfor dauer formation, and the Daf-c phenotype of unc-3 ization. Population assays toward volatile and water-soluble chemicals were performed as described previously (Bargmutants is fully suppressed by daf-3 and daf-5 mutations mann and Horvitz 1991a; Bargmann et al. 1993). (M. Ailion and J. H. Thomas, unpublished results). Noncomplementation between odr-9 and egl-4: odr-9(ky27) Studying genes with weak Daf-c phenotypes can theresponse and odr-9(ky185) failed to complement for the defect in refore provide information about the mechanisms inarm toward diacetyl. odr-9(ky27) was mapped to the left of LGIV using standard mapping crosses. odr-9(ky27)/eglvolved in the modulation of dauer regulatory signals, 4(n478) and odr-9(ky185)/egl-4(n478) trans-heterozygotes failed in addition to providing insight into specific aspects of to complement for the phenotypes of chemotaxis defects, neuronal function. Here we describe characterization egg-laying defects, altered body size, darkened intestines, and of the gene egl-4. egl-4 was initially identified in screens hyperforaging behavior. All alleles of egl-4 were recessive for for mutants defective in egg-laying behavior (Trent et all phenotypes tested. al. 1983). We show that mutations in egl-4 result in sevadult hermaphrodites were mounted on agarose pads and Egg-laying assays: To count and stage eggs, N2 and egl-4 eral pleiotropies that include chemosensory defects, al- viewed under Nomarski optics at 400 magnification. Animals tered body length, and defects in synaptic transmission. were grown at 20, and first day adults were analyzed. To We also find that egl-4 plays a role in relaying sensory determine if egl-4 is responsive to food cues, single N2, eglcues to the egg-laying circuit. Moreover, we show that 4(n478), and flp-1(yn2) adult hermaphrodites were placed on similar to unc-31, unc-64, and unc-3 mutants, egl-4 mulawn of bacteria. Animals were allowed to lay eggs for 2 3 hr standard worm growth plates with either no food or a day-old tants are Syn-Daf and exhibit defects in dauer formation. at room temperature. Animals were then picked off the plate Finally, we show that egl-4 mutations deregulate the and the number of eggs laid was counted. TGF- branch of the dauer pathway, and that most deeggs Dauer assays: Age-synchronized animals were allowed to lay fects of egl-4 mutants, including chemosensory defects, at room temperature for 3 6 hr. Parent animals were are variably suppressed by mutations in daf-3 and daf-5. then removed and plates were incubated at the given assay temperatures. Dauer and nondauer animals were counted To our knowledge, this is the first report that implicates after 100 hr at 15, 65hrat20, 48hrat25, and 44 hr at a SMAD protein in chemosensory signaling. 27. This permitted the scoring of transient dauers that recover rapidly. Small differences in temperature 25 can make significant differences in the number of dauers formed, so each MATERIALS AND METHODS set of assays included all the relevant strains. All relevant comparisons are between strains assayed in parallel. Plates with Strains: Wild-type worms used were C. elegans variety Bristol, partially purified dauer pheromone were prepared as destrain N2. Worms were grown using standard methods (Bren- scribed (Vowels and Thomas 1994). Additional details on ner 1974). the protocol followed for dauer assays are provided elsewhere Strains carrying the following mutations were used in this (Ailion and Thomas 2000). work. Strains were obtained from the Caenorhabditis Genetics Serotonin assays: Serotonin was made as a 10 mg/ml stock Center unless noted otherwise. Mutations are listed by linkage solution in water and added to a final concentration of 1, 2, group: or 5 mg/ml to worm growth agar immediately before pouring. Plates were seeded with concentrated bacteria immediately LGI: daf-16(m27), daf-16(mgdf50), dpy-5(e61), egl-32(n155), before use. Dauer formation was assayed after 43 hr following unc-13(e450), unc-29(e1072), unc-75(e950). synchronous egglays. After counting, plates were returned to LGII: daf-5(e1385), dpy-10(e128), kyis37[odr-10-gfp::lin-15], 27 and incubated for an additional 4 days, after which the tph-1(mg280), tra-2(q276), unc-4(e120), unc-52(e444). number of nondauers was counted to score for dauer recovery. LGIII: daf-2(e1370), daf-7(e1372), tax-4(ks11), unc-64(e246). Aldicarb and levamisole assays: The effects of aldicarb and LGIV: daf-14(m77), egl-4(n477), egl-4(n479), egl-4(n612), levamisole were scored in acute paralysis assays as follows. For egl-4(n579), egl-4(n478), flp-1(yn2), osm-3(p802), unc-31(e928). both assays, plates were seeded with bacteria the day before LGV: daf-11(sa195), osm-6(p811). the assay. A total of 20 young adult animals were picked to LGX: daf-3(e1376), daf-12(m20), dpy-3(e127), kyis53[odr-10- each of two duplicate plates. Aldicarb was made as a 100 mm GFP(tagged)::lin-15], unc-1(e719), unc-3(e151), unc-6(e78), uncstock solution in 70% ethanol and added to a final concentra- 58(e665). tion of 0.5 or 1.0 mm to worm growth agar immediately before flp-1(yn2) was obtained from C. Li; tph-1(mg280) was obtained pouring. Animals were scored for movement and pharyngeal from J. Y. Sze and G. Ruvkun; kyis37 and kyis53 were generated pumping when prodded with a platinum wire after 6, 8, and in the laboratory of C. I. Bargmann. The strain carrying mex47-10 hr. To most clearly show the differences between resistant [daf-7-gfp::rol-6] was obtained from D. Riddle. unc-64(e246) and nonresistant strains, we plotted the percentage paralysis was outcrossed once to remove an unlinked temperature-sensifailure on 0.5 mm aldicarb at 10 hr, where paralyzed is defined as tive sterile mutation; the egl-4 alleles n477, n479, and n612 to move when prodded. Strains defined as resistant were outcrossed an additional two times before analysis. The were clearly different from wild type at all time points and following strains carrying multiple mutations were obtained concentrations. Levamisole was made as a 100 mm stock solu- from the Caenorhabditis Genetics Center: lin-1(e1275) unc- tion in water and added to agar to a final concentration of 33(e204) IV, dpy-9(e12) ced-2(e1752) lin-1(e1275) IV. The follow- 100 m. Acute paralysis was scored every 30 min for 2 hr. ing two strains were obtained from H. R. Horvitz: egl-32(n155) Paralysis was defined as the absence of any moving or pumping I; daf-3(e1376) X and egl-32(n155) I; daf-5(e1385) II. when animals were prodded with a platinum wire. Behavioral screens and assays: odr-9(ky27) and odr-9(ky185) Construction of double and triple mutant strains: Double were isolated in behavioral screens for mutants unable to mutants between egl-4 and various daf-c or daf-d mutations chemotax towards diacetyl, essentially as described previously were constructed and confirmed by the methods described (Bargmann et al. 1993; Sengupta et al. 1994, 1996). Mutants previously (Vowels and Thomas 1992; Thomas et al. 1993).

4 126 S. A. Daniels et al. Briefly, egl-4 double mutants with daf-c mutations were built alleles, and therefore likely represent strong loss-of-function by first constructing egl-4/ ; daf-c/ heterozygotes. egl-4 was mutations (Thomas et al. 1993; Gottlieb and Ruvkun 1994; homozygosed by picking Egl animals. Subsequently, the daf-c Ogg et al. 1997; Patterson et al. 1997). mutation was homozygosed by picking dauers and recovering Statistical analysis: In all analyses involving comparisons them. egl-4 double mutants with daf-d mutations were built among multiple groups, statistical significance was determined by constructing egl-4/ ; daf-d/m heterozygotes where m is a using the Bonferroni-Dunn multiple comparisons procedure, visible marker. egl-4 was homozygosed by picking Egl animals, with the significance level set at 5%. Analyses were performed and the daf-d mutation was homozygosed by picking animals using the Statview 4.5 application (Abacus Concepts, Berkeley, that failed to segregate the marker m. Markers used were CA). as follows: daf-16 dpy-5(e61) unc-75(e950) or unc-13(e450); daf-3 unc-1(e719) dpy-3(e27); daf-5 unc-52(e444); daf-12 unc-58(e665) or unc-6(e78). The unexpected strong suppression of the 27 Daf-c phenotype of egl-4(n479) in several double RESULTS mutants made us examine whether the 27 Daf-c mutation odr-9 and egl-4 are allelic: We identified two alleles of was actually present in the double mutants and hence whether the gene odr-9 (ky27 and ky185) in behavioral screens the 27 Daf-c mutation was identical to the egl-4 mutation. First, we confirmed that the 27 Daf-c mutation in n479 muchemical diacetyl (see materials and methods). We for mutants unable to respond to the volatile attractive tants mapped to the same region as egl-4. Second, we deconstructed the egl-4(n479); daf-3 and daf-16; egl-4(n479) strains to placed odr-9 in the same genetic interval as the prereisolate the egl-4 mutation. In both cases, all egl-4 homozygotes viously identified gene egl-4 using standard three-factor generated were strongly Daf-c at 27, indicating that the supmapping crosses (data not shown). Five alleles of egl-4 pressed strains do contain the daf-c mutation and that it is tightly linked to egl-4. (n477, n478, n479, n579, and n612) have been identified An egl-4 osm-3 double mutant was constructed by first generating in genetic screens for mutants with defects in egg-laying osm-3/egl-4 unc-33 heterozygotes. Egl non-unc Osm re- behavior (Trent et al. 1983). We found that odr-9 and combinant progeny were selected and homozygosed. The egl-4 are allelic. n478/ky27 and n478/ky185 trans-heteroegl-4; osm-6 double mutant was built by successively homozygoszygotes fail to complement for all phenotypes tested ing egl-4 and osm-6 by the Egl and Osm or Dyf phenotypes, respectively. tph-1 doubles were built by picking Egl (egl-4) or (see materials and methods). This gene is henceforth Unc (unc-31, unc-64, orunc-3) animals segregating from tph- referred to as egl-4. Here we present detailed character- 1/m; egl-4 or unc/ heterozygotes, where m was dpy-10(e128) ization of seven alleles of egl-4. unc-4(e120), except in the case of the unc-3 double, where egl-4 mutants are egg-laying defective: Since several m was tra-2(q276). Animals that failed to segregate m were alleles of egl-4 had been previously identified on the presumed to carry tph-1, which was also scored by a low-penebasis of their egg-laying defects, we further examined trance withered tail (Wit) phenotype. Triple mutants of eglthe egg-laying behavior of all egl-4 mutants. Egg-laying 4; unc-3 with daf-5 or daf-16 were built by picking dauers from egl-4/ ; unc-3/ ; daf-5/unc-52 or daf-16/ heterozygotes to behavior has been described as biphasic, with periods homozygose both egl-4 and unc-3 simultaneously. After dauers of active egg laying interspersed with inactive periods recovered, daf-5 was homozygosed by picking animals that (Waggoner et al. 1998). Induction of entry into the failed to segregate Unc animals, while daf-16 was homozygosed by picking partial dauers. The egl-32; egl-4 double mutant was active phase is regulated by sensory cues and is mediated constructed by crossing unc-13/ ; egl-4/ males with egl-32 by the neurotransmitters serotonin and FMRFamide- hermaphrodites. Non-Egl cross-progeny were picked individually, related neuropeptides (Horvitz et al. 1982; Trent et and those segregating Unc animals were kept. Egl animals al. 1983; Weinshenker et al. 1995; Waggoner et al. from these plates were again picked singly, and those segregat- 2000). Within the active phase, the rate of egg laying is ing Unc animals were selected as animals having the genotype regulated by an additional neurotransmitter, acetylchoegl-32/unc-13; egl-4/egl-4. egl-32 was homozygosed by picking animals that failed to segregate Unc progeny. Presence of the line (ACh; Trent et al. 1983; Weinshenker et al. 1995; appropriate single mutations was confirmed by complementation Waggoner et al. 1998). It has been reported previously testing for visible or behavioral phenotypes. Additional that egl-4 mutants exhibit normal rates of egg laying details on strain constructions are available upon request. within the active phase, but have longer latent periods The rationale behind the selection of alleles for some doubetween active phases (Waggoner et al. 1998). Consisble mutant constructions is as follows. The tph-1(mg280), unctent with this, egl-4 mutants experience transient bloat- 3(e151), unc-31(e928), daf-11(sa195), daf-14(m77), and dafing, where animals become filled with eggs but eventu- 16(mgDf50) alleles are likely null alleles (Avery et al. 1993; Ogg et al. 1997; Prasad et al. 1998; Birnby et al. 2000; Inoue ally lay most of their eggs. In Table 1 we examined this and Thomas 2000; Sze et al. 2000). daf-7(e1372) mutants have egg-laying phenotype in two ways. First, we counted the been shown previously to exhibit a Daf-c phenotype equivalent number of eggs retained in the uterus of adult hermaphin strength to that of animals carrying a predicted daf-7 null allele (Ren et al. 1996). daf-2(e1370) results in severe loss of rodites and found that egl-4 mutants retain approxi- daf-2 function; no clear daf-2 null mutations have been identified mately three times as many eggs as wild-type adults of (Kimura et al. 1997). It has been shown previously that comparable stage. We also examined the developmental the Daf-c phenotype of daf-7(e1372) and daf-2(e1370) mutants stages of eggs retained in the uterus of egl-4 mutants. is strongly enhanced in double mutant combinations with Typically, early events in embryogenesis occur in utero; mutations in parallel branches of the dauer pathway, but not with mutations in the same branch (Thomas et al. 1993). daffertilization). We find that 30% of the eggs retained eggs are laid during gastrulation (at min post- 3(e1376) and daf-16(m27) alleles have been shown to suppress Daf-c phenotypes to a similar degree as the respective null in egl-4 mutants are at the comma stage or later in

5 egl-4 Regulates Neuronal Functions 127 TABLE 1 Number and stage of eggs in N2 and egl-4 animals No. eggs in % late eggs in Genotype uterus uterus a N (20) 0 (40) ky (40) 33 1 (40) ky (40) 29 2 (40) n (40) 30 1 (40) n (40) 33 1 (40) n (40) 34 1 (40) n (40) 37 2 (40) n (40) 26 2 (40) Figure 2. Egg laying by egl-4 mutants is insensitive to food cues. Mean number of eggs laid per hour with or without food present is shown. The numbers of independent animals tested for each condition are indicated under the appropriate bars. Egg laying in wild-type animals is decreased in the ab- sence of food (P 0.001). Egg laying in egl-4 and flp-1 mutants is not significantly altered in the absence of food (P 0.05). P values were determined using the Mann-Whitney rank sum test. Number in parentheses is number of animals counted. a Late stage eggs are defined as eggs that are at or beyond the comma stage in development ( 400 min postfertilization or later). development ( 400 min postfertilization or later). Eggs at this late developmental stage are rarely if ever observed in the uterus of well-fed wild-type hermaphro- are sensed by the bilaterally symmetrical ciliated neuron dites. Thus, egl-4 mutants lay eggs at a later develop- types AWA and AWC (Bargmann et al. 1993). AWA mental stage, likely as a consequence of delayed active neurons mediate responses to the chemicals diacetyl egg-laying periods. All alleles appear to cause significant (2,3-butanedione) and pyrazine, while AWC neurons defects with no clear allelic series. are required for the attractive response to the volatile Since entry into the active phase of egg laying is regu- chemicals benzaldehyde, isoamyl alcohol, butanone, lated partly by serotonin, the defects of egl-4 mutants and 2,3-pentanedione. Both neuron types are required could result from pre- or postsynaptic defects in the for the response to the chemical trimethylthiazole. Sigserotonergic pathway. The egg-laying phenotype of egl-4 naling molecules such as olfactory receptors, G proteins, mutants is variably responsive to both serotonin and and ion channels that function in each of these neuron imipramine (a serotonin reuptake inhibitor; S. A. Dan- types have been identified previously (Coburn and iels and P. Sengupta, data not shown; Trent et al. Bargmann 1996; Komatsu et al. 1996; Sengupta et al. 1983), suggesting that egl-4 could function both pre- 1996; Colbert et al. 1997; Zwaal et al. 1997; Roayaie and postsynaptically. egl-4 could also act to potentiate et al. 1998; Jansen et al. 1999; Troemel et al. 1999). the effect of serotonin on initiating the active phase of We first examined the responses of all egl-4 mutants to odorants sensed by the AWA neurons. As shown in Figure 3, all egl-4 alleles tested exhibit very strong defects in the response to diacetyl and weaker but significant defects in the response to pyrazine. Since egl-4 mutants retain residual responses to pyrazine, egl-4 alleles likely affect a subset of functions rather than overall develop- ment of the AWA neurons. Diacetyl is recognized by the seven-transmembrane domain olfactory receptor ODR- 10, which is expressed specifically in the AWA neurons and is localized to their sensory cilia (Sengupta et al. 1996). To determine if the strong diacetyl defect of egl-4 results from defects in odr-10 expression or localization, we examined expression and subcellular localization of a green fluorescent protein (GFP)-tagged ODR-10 fusion protein in n478 animals. Both expression and localization of ODR-10 were unaltered in n478 mutants (data not shown). We have shown previously that while odr-10 null mutants fail to respond to 1 nl of diacetyl, they respond relatively normally to 100 nl of diacetyl (Sengupta et al. 1996). Diacetyl (1 nl) is sensed exclusively by the AWA neurons, while higher concentrations egg laying (Waggoner et al. 2000). Such a function has been ascribed to the FMRFamide-related neuropeptides encoded by the flp-1 gene. flp-1 plays a role in relaying sensory cues to the egg-laying circuit such that the rate of egg laying in flp-1 mutants is insensitive to food signals (Waggoner et al. 2000). To determine if egl-4 functions similarly, we compared the rate of egg laying in n478 mutants in the presence or absence of a bacterial food source. We find that egg laying by egl-4 mutants is insensitive to regulation by food cues (Figure 2). While egg laying by wild-type animals is significantly suppressed in the absence of food, the rate of egg laying in egl-4 mutants is unaffected. Thus, egl-4 may function to relay sensory cues to modulate the egg-laying circuit. egl-4 mutants exhibit multiple defects in chemosensory behaviors: The ky27 and ky185 alleles were isolated on the basis of their failure to respond to the volatile odorant diacetyl. We examined the chemosensory behaviors of all egl-4 alleles, and show that egl-4 mutants have widespread chemosensory defects. Attractive volatile chemicals: Attractive volatile chemicals

6 128 S. A. Daniels et al. Figure 3. Responses of egl-4 mutants to odorants sensed by the AWA neurons. Responses of egl-4 mutants to a point source of (A) 1 nl of diacetyl; (B) 1 l of 10 mg/ml pyrazine; (C) 1 l of 1:1000 dilution of trimethylthiazole; and (D) 100, 10, and 1 nl of diacetyl. Each data point represents the mean of at least six independent assays using 200 animals in each assay. Error bars equal the SEM. Single asterisks mark responses that are different from wild type at P 0.01; double asterisks mark responses that are different at P C.I., chemotaxis index. are sensed redundantly by the AWA and AWC neurons with defects in the structure of the sensory cilia often (P. Sengupta and C. I. Bargmann, unpublished results). fail to fill with dye (Perkins et al. 1986; Starich et al. However, egl-4 mutants fail to respond to all con- 1995). However, egl-4 mutants dye-fill normally, and no centrations of diacetyl tested (Figure 3D), consistent obvious defects in the morphology of the neurons were with defects in both the AWA and AWC chemosensory visible (data not shown). neurons. We also examined additional sensory behaviors. We next tested the responses of egl-4 mutants to addi- These included repulsion from volatile repellents, responses tional odorants sensed by the AWC neurons. While all to mechanical cues such as nose touch and alleles have normal responses to the odorant benzalde- osmotic shock, and responses to gentle body touch. hyde, they have weaker defects in the responses to butanone Sensory neurons mediating each of these behaviors have and isoamyl alcohol, and strong defects in the been identified (Chalfie et al. 1985; Bargmann et al. response to 2,3-pentanedione, an odorant structurally 1990; Kaplan and Horvitz 1993; Troemel et al. 1995, related to diacetyl (Figure 4). Overall, n478 has the 1997). egl-4 mutants were found to be wild type in their strongest defects and n477 has the weakest defects. All responses to these stimuli (data not shown), indicating egl-4 mutants except n478 exhibit wild-type response to that mutations in egl-4 affect the functions of a restricted trimethylthiazole (Figure 3C). subset of neurons. Attractive water-soluble chemicals: C. elegans is also at- egl-4 mutants are hypersensitive to dauer-inducing tracted to water-soluble chemicals such as NaCl and conditions: Dauer formation is dependent on the per- lysine (Ward 1973; Dusenbery 1974; Bargmann and ception of chemosensory cues such as dauer pheromone Horvitz 1991a). This behavior is mediated largely by and food. These cues are sensed by ciliated neurons the ASE ciliated neuron type, with minor contributions (Bargmann and Horvitz 1991b; Schackwitz et al. from additional neurons (ASG, ASI, ADF, and ASK; 1996). Mutants with chemosensory defects often have Bargmann and Horvitz 1991a). In addition to wide- defects in the regulation of dauer formation (Lewis spread defects in responses to volatile attractive chemicals, and Hodgkin 1977; Albert et al. 1981; Riddle et al. we found that all egl-4 mutants have strong defects 1981; Thomas 1993; Vowels and Thomas 1994; in their responses to NaCl and lysine (Figure 5). The Coburn et al. 1998). In addition to the chemosensory morphology of a subset of ciliated neurons (ASI, ASK, defects described above, egl-4 mutants also show defects ADL, AWB, ASH, and ASJ) can be visualized by filling in the dauer formation process. egl-4(n478) has been animals with the lipophilic dye DiO (Perkins et al. 1986; shown previously to be hypersensitive to dauer pheromone Herman and Hedgecock 1990). Neurons of mutants (Golden and Riddle 1984b). We verified this

7 egl-4 Regulates Neuronal Functions 129 Figure 4. Responses of egl-4 mutants to odorants sensed by the AWC neurons. Responses of egl-4 mutants to a point source of (A) 1 l of 1:200 benzaldehyde, (B) 1 l of 1:1000 dilution of butanone, (C) 1 l of 1:100 dilution of isoamyl alcohol, and (D) 1 l of 1:1000 dilution of 2,3-pentanedione. Each data point represents the mean of at least six independent assays using 200 animals in each assay. Error bars equal the SEM. Single asterisks mark responses that are different from wild type at P 0.01; double asterisks mark responses that are different at P this allelic order differs from that found for the response to volatile odorants. Mutations in other Daf-c genes also result in hypersen- sitivity to dauer pheromone (Golden and Riddle 1984b; Thomas et al. 1993). However, unlike most of these mutants, which are strongly Daf-c at 25, all egl-4 alleles except n479 form few or no dauers under noninducing conditions at this temperature. n479 exhibits a weak Daf-c phenotype at 25 (Figure 6A). Dauer formation is modulated by temperature, and it has been shown previously that several mutants with weak Daf-c phenotypes at 25 are strongly Daf-c at the elevated temperature of 27 (Ailion et al. 1999). Such mutants include unc-31, unc-64, and unc-3 (Ailion et al. 1999). We examined the Daf-c phenotypes of egl-4 mutants at 27. As shown in Figure 6B, all egl-4 alleles exhibit a Daf-c phenotype to varying degrees at 27. n479 is the strongest allele, forming close to 100% dauers at 27, while ky185 is the weakest allele, forming 25% dauers. The allelic series with respect to dauer formation at 27 is roughly similar to that determined by pheromone hypersensitivity. egl-4 is Syn-Daf with unc-3: Many weak Daf-c mutants have a Syn-Daf phenotype in double mutant combina- tions with unc-31, unc-64, and unc-3 mutants (I. Kats- ura, personal communication; Ailion et al. 1999). We wished to determine if egl-4 mutants are also Syn-Daf. and extended it by demonstrating that all alleles of egl-4 exhibit hypersensitivity to dauer pheromone (Figure 6A). While wild-type animals make 1% dauers upon addition of 1 l of dauer pheromone, at this concentration nearly 100% of egl-4(n479) animals form dauers. By this assay, the strengths of alleles are as follows: n479 ky27, n579, n477 n612 n478, ky185. We note that Figure 5. Responses of egl-4 mutants to water-soluble chemicals. Shown are responses of egl-4 mutants to 0.2 m NaCl and to 0.5 m lysine. Each data point represents the mean of five independent assays using 100 animals in each assay. Error bars equal the SEM. All responses differ from wild type at P

8 130 S. A. Daniels et al. Figure 6. Dauer formation phenotypes of egl-4 mutants. (A) The number of dauers formed at different concentrations of added pheromone is shown as a percentage of the total number of animals on the plate. Approximately animals were counted at each concentration of pheromone. (B) Shown are the number of dauers formed after 2 days at 27. See materials and methods for additional details. Approximately 100 animals of each genotype were counted. Numbers shown are from a single experiment. Experiments repeated on independent days show similar relative differences. egl-4 unc-31 and unc-64; egl-4 double mutants do not or unc-3 mutants. However, serotonin does appear to show Syn-Daf phenotypes (M. Ailion and J. H. Thomas, enhance dauer recovery of these mutants. Exogenous data not shown; I. Katsura, personal communication). serotonin leads to a dose-dependent increase in dauer However, we find that the egl-4(n478); unc-3(e151) double recovery of egl-4, unc-64, and unc-3 mutants, and to a mutant is strongly Syn-Daf, forming nearly 100% lesser extent of unc-31 mutants (Figure 7B). This sug- dauers at all temperatures (see Table 5). unc-3 has been gests that the effect of serotonin on dauer recovery is shown to regulate the expression of the DAF-7 TGF- not specific to egl-4. ligand (M. Ailion and J. H. Thomas, unpublished results; To determine if serotonin acts in parallel to egl-4, we P. Ren and D. Riddle, personal communication). analyzed double mutants with tph-1 (Table 2). Dauer We examined the expression of a daf-7::gfp fusion gene formation is enhanced in a tph-1; daf-7 double mutant, in egl-4(n478) mutants, and found that unlike unc-3 mu- suggesting that tph-1 acts in parallel to the group II Daf-c tants, mutations in egl-4 do not affect expression of the pathway (Sze et al. 2000). Similarly, we find that the DAF-7 TGF- ligand (data not shown). Daf-c phenotype of a tph-1; egl-4 double mutant is also Serotonin acts in parallel to egl-4 to regulate dauer strongly enhanced at all temperatures tested, indicating formation: egl-4 mutants exhibit a subset of the pheno- that tph-1 acts in parallel to egl-4 (Table 2). tph-1 mutants types associated with those of mutants with defects in also enhance dauer formation of unc-31, unc-64, and serotonin signaling. For example, tph-1 tryptophan hy- unc-3 mutants, suggesting that tph-1 acts in parallel to droxylase mutants that fail to synthesize serotonin have these genes as well. We find that the tph-1; unc-31 and egg-laying defects and a weak Daf-c phenotype similar tph-1; unc-64 double mutants are no longer temperature to that of egl-4 mutants (Sze et al. 2000). Serotonin has sensitive for dauer formation, while the tph-1; egl-4 dou- also been implicated in regulating foraging behavior ble mutant is weakly temperature sensitive. and male mating (Loer and Kenyon 1993; Duerr et egl-4 mutants exhibit synaptic transmission defects: al. 1999). We examined egl-4 mutants and found that unc-64 and unc-31 encode proteins that mediate synaptic they also have foraging and male mating defects (data transmission and other types of Ca 2 -regulated secretion not shown). This raised the possibility that egl-4 mutants (Livingstone 1991; Avery et al. 1993; Ann et al. 1997; have defects in serotonin signaling. The defects of tph- Ogawa et al. 1998; Saifee et al. 1998). Like egl-4, both 1 mutants can be rescued by the addition of exogenous mutants have multiple behavioral pleiotropies and are serotonin (Sze et al. 2000). Although serotonin does not Syn-Daf. unc-64 and unc-31 are resistant to aldicarb, an completely rescue the egg-laying defect of egl-4 mutants inhibitor of acetylcholinesterase, indicating that these (see above), we tested whether exogenous serotonin genes play a role in cholinergic transmission (Nguyen could rescue the 27 Daf-c phenotype of the egl-4(n479) et al. 1995; Miller et al. 1996; Saifee et al. 1998). To mutant. As shown in Figure 7A, serotonin does not test whether egl-4 also plays a role in synaptic transmis- rescue the 27 Daf-c phenotype of egl-4, unc-64, unc-31, sion, we examined the effects of acute exposure of egl-4

9 egl-4 Regulates Neuronal Functions 131 in the postsynaptic response to ACh. To determine whether egl-4 functions presynaptically or postsynaptically, we assayed sensitivity to the nicotinic ACh receptor agonist levamisole (Lewis et al. 1980a,b). A postsynaptic mutant unc-29(e1072) is completely resistant to levamisole (Fleming et al. 1997), whereas the unc-31(e928) mutant, which is involved in presynaptic release mechanisms, is sensitive. All three alleles of egl-4 tested are sensitive to levamisole, as are unc-3(e151) mutants, suggesting that egl-4 functions presynaptically in regulating synaptic transmission (Figure 8B). egl-4 mutants also appear to be hypersensitive to levamisole, similar to unc-31 mutants. However, unlike unc-31 mutants, egl-4 mutants exhibit adaptation at later time points, such that paralyzed animals resume pumping and slight movements of the nose. egl-4 mutants have increased body length: Body length in C. elegans is regulated via a DPP/BMP-mediated signaling pathway (Savage et al. 1996; Padgett et al. 1998; Patterson and Padgett 2000). This pathway is similar to the TGF- -mediated branch of the dauer pathway, but uses a different ligand and an independent set of receptors and SMAD signaling genes with the exception of the DAF-4 TGF- type II receptor, which is shared by both pathways. daf-4 mutants are Daf-c and have reduced body length (Estevez et al. 1993). In contrast, we noted that egl-4 alleles cause increased body length (Figure 9). egl-4 mutants are 20 30% longer than wildtype animals. n579 mutants show the strongest defect, being on average 31% longer than wild type, while the n477 and n612 mutants show the weakest phenotypes, being 17% longer. In comparison, lon-2(e678) mutants, which have been implicated in the pathway regulating body length, are 34% longer than wild-type animals. egl-4 functions in the group II branch of the dauer signaling pathway: Three parallel pathways that regulate dauer formation have been identified (see Introduction). In addition to the Daf-c phenotype, egl-4 mutants share additional phenotypes in common with both group I and group II Daf-c genes. Group I Daf-c mutants exhibit chemosensory defects to volatile and water-solu- ble chemicals, while group II Daf-c genes have defects in egg laying, have dark intestines (Din), and exhibit a clumpy behavior, in which animals tend to congregate in clumps (Trent et al. 1983; Thomas 1993; Vowels and Thomas 1994; Riddle and Albert 1997). egl-4 mutants, in addition to having chemosensory and egg-lay- ing defects, are also Din, but exhibit little if any clumpy behavior. To determine the pathway in which egl-4 func- tions, we made double mutants between egl-4 and muta- tions in each of these pathways. We examined suppres- sion or enhancement of different phenotypes of egl-4 in these double mutants. Dauer formation: The Daf-c phenotype of mutants that function in the group II pathway is fully suppressed by daf-3 and daf-5 mutations, while those in the other Figure 7. egl-4 mutants and serotonin. (A) Shown is the percentage of dauers formed at 27 in the presence of 5 mg/ ml serotonin in the plate. A total of animals of each genotype were counted in two independent assays. (B) The number of dauers on plates containing serotonin were counted after 4 days at 27 to assay recovery. Wild-type animals recover in 1 day at 27 in the presence or absence of serotonin. mutants to aldicarb. In Figure 8A, we show that the n479 and n612 mutants of egl-4 are strongly resistant to aldicarb while the ky27 mutant is less resistant. We also find that unc-3(e151) mutants are strongly resistant to aldicarb. In addition to the quantitative measurement of aldicarb resistance, egl-4 mutants are qualitatively less hypercontracted than wild-type animals on aldicarb. Resistance to aldicarb can arise either from a defect in presynaptic ACh synthesis or release, or from defects

10 132 S. A. Daniels et al. TABLE 2 tph-1 interactions with Syn-Daf genes Dauer formation (%) Genotype tph-1(mg280) 1 (289) 3 (245) 5 (171) egl-4(n479) 0 (268) 0 (425) 0 (225) unc-3(e151) 0 (433) 0 (402) 0 (412) unc-31(e928) 0 (487) 0 (306) 0 (345) unc-64(e246) 1 (198) 0 (154) 0 (182) tph-1(mg280); egl-4(n479) 47 (270) 75 (223) 76 (99) tph-1(mg280); unc-3(e151) 0 (244) 4 (216) 49 (134) tph-1(mg280); unc-31(e928) 65 (324) 67 (241) 64 (151) tph-1(mg280); unc-64(e246) 74 (281) 61 (244) 71 (214) Number in parentheses is number of animals counted. branches are not (Thomas et al. 1993). We first tested dauer formation in unc-31 and unc-64 mutants is also whether the dauer formation defects of egl-4 mutants suppressed by mutations in daf-12 and daf-16, the Daf-c are suppressed by daf-3 and daf-5. We found that muta- phenotype of these mutants is not suppressed by daf-3 tions in either daf-3 or daf-5 completely suppress egl-4 and daf-5 (Ailion et al. 1999). In contrast, as shown in dauer formation induced by pheromone (Table 3), sug- Table 4, dauer formation in egl-4 mutants at 27 is gesting that egl-4 functions in the group II pathway of strongly suppressed by daf-3 and daf-5 mutations, further dauer formation. We also assayed suppression of the confirming placement of egl-4 in the group II pathway. Daf-c phenotype of the strong n479 allele at 27. The Interestingly, egl-4 and daf-3 appear to mutually suppress Daf-c phenotype of n479 is suppressed by mutations in each other, since while the daf-3(e1376) mutant and eglthe Daf-d genes daf-3, daf-5, daf-16, and daf-12 (Table 4(n479) each form dauers at 27, the egl-4; daf-3 double 4). daf-12 mutations suppress the dauer phenotypes of mutant makes fewer dauers than either single mutant Daf-c mutants in all three branches, since these alone. Mutations in the Daf-d gene osm-6, which also branches are thought to converge at daf-12 (Riddle and lead to a Daf-c phenotype at 27 (M. Ailion and J. H. Albert 1997; Antebi et al. 1998). daf-16 mutations have Thomas, unpublished results; Apfeld and Kenyon been shown to suppress the 27 Daf-c phenotypes of 1999), do not suppress egl-4, demonstrating that such mutants in different branches of the pathway (Ailion mutual suppression is specific to daf-3. et al. 1999; Apfeld and Kenyon 1999). However, while We also determined whether the Syn-Daf phenotype Figure 8. Pharmacological analysis of synaptic transmission. (A) egl-4 mutants are resistant to aldicarb. The number of animals that are paralyzed on plates containing aldicarb is shown as a percentage of the total number of animals. Paralysis (defined as the absence of any movement or pumping when prodded) was scored after 10 hr on 0.5 mm aldicarb. Approximately 40 animals were counted for each genotype. (B) egl-4 mutants are sensitive to levamisole. Paralysis on 100 m levamisole was scored every 30 min. Approximately 40 animals were counted for each genotype.

11 egl-4 Regulates Neuronal Functions 133 TABLE 4 Suppression of egl-4 dauer formation at 27 % dauer formation Genotype N2 4 (187) 11 (163) daf-3(e1376) 30 (216) 96 (219) daf-5(e1385) 2 (191) 6 (206) daf-16(m27) a 1 (172) 1 (166) daf-12(m20) 0 (94) 0 (61) egl-4(n479) 97 (263) 100 (272) egl-4(n479); daf-3(e1376) 0 (151) 22 (174) Figure 9. egl-4 mutants have increased body length. The daf-5(e1385); egl-4(n479) 0 (260) 4 (239) mean body length (in millimeters) is shown for animals of daf-16(m27); egl-4(n479) a 1 (284) 2 (249) the indicated genotypes. At least 30 adult animals of each genotype were measured under 100 magnification using egl-4(n479); daf-12(m20) 0 (193) 0 (169) Nomarski optics and an eyepiece micrometer. Animals were measured 24 hr after the final molt. The mean body length Genotype % dauer formation at 26.7 of each egl-4 mutant is different from that of wild type at N2 4 (238) P osm-6(p811) 96 (192) egl-4(n479) 99 (289) egl-4(n479); osm-6(p811) 99 (272) of the egl-4; unc-3 double mutant is suppressed by muta- Number in parentheses is the number of animals counted. tions in daf-5 and daf-16. We find that mutations in a Partial dauers as described (Vowels and Thomas 1992). daf-5 completely suppress the egl-4(n478); unc-3(e151) synthetic dauer phenotype at either 15 or 25, while daf-16 mutations fail to suppress at 25 and only weakly Daf-c phenotype at 15. At this temperature, egl-4 clearly suppress at 15 (Table 5). These results further support enhances the Daf-c phenotypes of daf-11(sa195) and taxplacement of egl-4 in the group II branch. 4(ks11) (group I genes; Table 6). Although egl-4 does Since dauer formation is regulated by parallel pathnot enhance the Daf-c phenotype of daf-2(e1370) (a ways, there is strong enhancement of the Daf-c phenomember of the third branch) at 15, there is significant type in double mutants between genes in different pathenhancement at 20. These results suggest that egl-4 acts ways, but not between those acting in the same pathway in parallel to the group I branch and the daf-2 branch. (see Figure 1). For instance, while the Daf-c phenotypes We find that dauer formation is also slightly increased of daf-8 and daf-11 mutants are incompletely penetrant in egl-4(n478); daf-7(e1372) and egl-4(n478); daf-14(m77) at low temperatures, a daf-8; daf-11 double mutant forms double mutants (Table 6). This is difficult to interpret 100% dauers (Thomas et al. 1993). In contrast, the since daf-7 and daf-14 are strongly Daf-c on their own phenotype of a daf-8; daf-14 double mutant is still incomat 15. However, in both cases it appears that egl-4 does pletely penetrant since both of these genes function in enhance the phenotype, suggesting that egl-4 may act the TGF- pathway (Thomas et al. 1993). To further at least partially in parallel to the TGF- pathway, as confirm that egl-4 functions in the group II pathway, we well as the other two pathways. Since we do not know the built double mutants of egl-4 with Daf-c mutations in molecular nature of the egl-4(n478) allele, it is formally different pathways and looked for enhancement of the possible that the enhancement observed in the double mutants is due to the nonnull nature of this allele. TABLE 3 Chemosensory behaviors: Group I Daf-c mutants exhibit numerous chemosensory defects (Vowels and Thomas daf-3 and daf-5 suppress egl-4 dauer 1994; Coburn and Bargmann 1996). Group I genes formation on pheromone such as daf-11 encode components of a cgmp-mediated % dauer formation at 25.4 signaling pathway that function both in dauer formation as well as other chemosensory processes (Birnby et al. Genotype pheromone pheromone a 2000). To date, no genes in the TGF- or the insulin pathway have been implicated in other chemosensory N2 0 (180) 90 (189) daf-3(e1376) 0 (162) 0 (158) pathways (Trent et al. 1983; Ren et al. 1996; Schack- daf-5(e1385) 0 (155) 1 (143) witz et al. 1996; Sze et al. 2000; Tissenbaum et al. 2000). egl-4(n478) 1 (123) 100 (81) We further examined whether the chemosensory deegl-4(n478); daf-3(e1376) 0 (137) 0 (124) fects of egl-4(n478) are suppressed by daf-3 and daf-5 egl-4(n478); daf-5(e1385) 0 (99) 1 (77) mutations. We find that daf-3(e1376) completely suppresses Number in parentheses is the number of animals counted. all olfactory defects of egl-4(n478) (Figure 10). a Plates with 50 l pheromone. However, the effects of daf-5 mutations are more spe-