Genetics, Vol. 155, 85-104, May 2000, Copyright © 2000
A Transmembrane Guanylyl Cyclase (DAF-11) and Hsp90 (DAF-21) Regulate a Common Set of Chemosensory Behaviors in Caenorhabditis elegans
Deborah A. Birnby1,2,a,
Elizabeth Malone Link1,3,a,
Jennifer J. Vowels2,a,
Hong Tiana,
Patrick L. Colacurcioa, and
James H. Thomasa
a Department of Genetics, University of Washington, Seattle, Washington 98195-7360
Corresponding author:
James H. Thomas, Department of Genetics, Box 357360, University of Washington, Seattle, WA 98195-7360., jht{at}genetics.washington.edu (E-mail)
Communicating editor: I. GREENWALD
 | ABSTRACT |
|---|
Caenorhabditis elegans daf-11 and daf-21 mutants share defects in specific chemosensory responses mediated by several classes of sensory neurons, indicating that these two genes have closely related functions in an assortment of chemosensory pathways. We report that daf-11 encodes one of a large family of C. elegans transmembrane guanylyl cyclases (TM-GCs). The cyclic GMP analogue 8-bromo-cGMP rescues a sensory defect in both daf-11 and daf-21 mutants, supporting a role for DAF-11 guanylyl cyclase activity in this process and further suggesting that daf-21 acts at a similar step. daf-11::gfp fusions are expressed in five identified pairs of chemosensory neurons in a pattern consistent with most daf-11 mutant phenotypes. We also show that daf-21 encodes the heat-shock protein 90 (Hsp90), a chaperone with numerous specific protein targets. We show that the viable chemosensory-deficient daf-21 mutation is an unusual allele resulting from a single amino acid substitution and that the daf-21 null phenotype is early larval lethality. These results demonstrate that cGMP is a prominent second messenger in C. elegans chemosensory transduction and suggest a previously unknown role for Hsp90 in regulating cGMP levels.
LIKE all free-living organisms, Caenorhabditis elegans responds to a variety of environmental stimuli. The presence of food affects locomotion, egg laying, and defecation (B. SAWIN, C. TRENT and H. R. HORVITZ, personal communication; LIU and THOMAS 1994 
). Specific volatile and nonvolatile chemicals attract or repel C. elegans in chemotaxis assays (WARD 1973 ; DUSENBERY 1974 ; BARGMANN et al. 1993 ), and a constitutively secreted pheromone regulates development (GOLDEN and RIDDLE 1982 , GOLDEN and RIDDLE 1984A ). daf-11 and daf-21 mutants have similar defects in several of these responses (reviewed in RIDDLE and ALBERT 1997 ), suggesting that the daf-11 and daf-21 gene products (DAF-11 and DAF-21) act at the same step to regulate chemosensory transduction in several types of sensory neurons.
In C. elegans, bilaterally symmetric pairs of ciliated sensory neurons in the head amphid sensilla mediate many chemosensory behaviors. For example, C. elegans is attracted to a variety of nonvolatile chemicals, including Cl-, cAMP, and biotin, which are sensed primarily by the ASE neurons (BARGMANN and HORVITZ 1991B ). Response to these attractants is defective in both daf-11 and daf-21 mutants (VOWELS and THOMAS 1994 ). C. elegans is also attracted to several volatile odorants, and different chemical classes are detected by the AWA, AWB, and AWC neurons (BARGMANN et al. 1993 ). daf-11 and daf-21 mutants do not respond to isoamyl alcohol and benzaldehyde, which are sensed by the AWC neurons, but respond normally to odorants sensed by the AWA neurons (VOWELS and THOMAS 1994 ). These mutant phenotypes suggest that both daf-11 and daf-21 mutants have functional deficits in two classes of sensory neurons, ASE and AWC.
The external environment also regulates formation of the C. elegans dauer larva, an alternative third-stage larva specialized for survival under harsh conditions (CASSADA and RUSSELL 1975 ). A constitutively secreted pheromone is the main inducer of dauer formation (GOLDEN and RIDDLE 1982 , GOLDEN and RIDDLE 1984A ). Temperature and food modulate the effect of this pheromone, with high temperature and low food levels favoring dauer formation (GOLDEN and RIDDLE 1984A , GOLDEN and RIDDLE 1984B ). daf-11 and daf-21 mutants are Daf-c (dauer formation constitutive), forming dauers even in the absence of inducing conditions (RIDDLE et al. 1981 ; THOMAS et al. 1993 ). In addition to daf-11 and daf-21, several other genes regulate dauer formation, and most of these genes have been ordered into a genetic pathway by analysis of double mutants (reviewed in RIDDLE and ALBERT 1997 ). The Daf-c genes daf-1, -4, -7, -8, and -14 are thought to act in parallel to daf-11 and -21 (THOMAS et al. 1993 ) and encode components of a TGF-ß-like signaling pathway (GEORGI et al. 1990 ; ESTEVEZ et al. 1993 ; REN et al. 1996 ; INOUE and THOMAS 2000 ; A. ESTEVEZ and D. L. RIDDLE, personal communication). The Daf-c genes daf-2 and age-1 are thought to act in parallel to or downstream of daf-11 and -21 (THOMAS et al. 1993 ; GOTTLIEB and RUVKUN 1994 ) and encode components of an insulin-like signaling pathway (KIMURA et al. 1994 ; MORRIS et al. 1996 ). It is notable that analyses of pleiotropic mutant phenotypes and double mutant interactions have all identified shared properties of daf-11 and daf-21 mutants that are distinct from other Daf-c mutants (reviewed in RIDDLE and ALBERT 1997 ).
Regulation of dauer formation is also mediated by particular classes of amphid sensory neurons, and the functions of these neurons have been determined by killing specified cells with a laser microbeam. When the ADF and ASI neurons are killed together, wild-type larvae form dauers constitutively (BARGMANN and HORVITZ 1991A ; SCHACKWITZ et al. 1996 ), indicating that these neurons normally repress dauer formation in the absence of inducing conditions. Varied evidence suggests that ADF and ASI mediate the function of the TGF-ß-like signaling branch of the dauer pathway (BARGMANN and HORVITZ 1991A ; THOMAS et al. 1993 ; SCHACKWITZ et al. 1996 ). In contrast, the ASJ neurons promote dauer formation, since killing these cells suppresses dauer formation in response to pheromone (SCHACKWITZ et al. 1996 ). Killing the ASJ neurons also suppresses the constitutive dauer formation of daf-11 and daf-21 mutants, suggesting that the Daf-c phenotype in these mutants results from activation of ASJ (SCHACKWITZ et al. 1996 ). Other Daf-c genes that function in the TGF-ß-like signaling pathway or the insulin-like signaling pathway are distinct in that their Daf-c phenotype does not depend on the ASJ neurons. The ASJ neurons are also required for recovery from the dauer state (BARGMANN and HORVITZ 1991A ) and daf-21 and most daf-11 dauers recover poorly (VOWELS and THOMAS 1994 ). Mutations that disrupt the amphid sensory cilia have been used in epistasis experiments to demonstrate that the Daf-c phenotype of daf-11 and daf-21 mutants requires intact ciliated sensory endings, suggesting that DAF-11 and DAF-21 function in the sensory endings to mediate an early step in chemosensory signal transduction (VOWELS and THOMAS 1992 ).
The various shared properties of daf-11 and daf-21 mutants indicate that these genes have closely related functions in several types of sensory neurons. Here we present molecular analyses of these two genes. We show that DAF-11 is homologous to transmembrane guanylyl cyclases (TM-GCs), which catalyze the formation of cyclic GMP (cGMP) from GTP. cGMP is a widely used second messenger that regulates kinases, other nucleotide cyclases, cyclic nucleotide phosphodiesterases, and cGMP-gated ion channels (reviewed in GOY 1991 ). We show that DAF-11 is one of a large family of TM-GCs predicted by the C. elegans Genome Sequencing Consortium, suggesting that cGMP is a common second messenger in C. elegans. Expression of a daf-11::gfp reporter fusion is consistent with the known roles of DAF-11 in ASE, AWC, and ASJ. We demonstrate that a cGMP analogue rescues the Daf-c phenotypes of daf-11 and daf-21 mutants, indicating that in both cases the mutant phenotype results from reduced levels of cGMP. We also show that daf-21 encodes heat-shock protein 90 (Hsp90), a chaperone protein that stabilizes many diverse protein targets. Our analysis indicates that the Daf-c daf-21 mutation is an unusual allele and that a null mutation is lethal. We suggest a model in which Hsp90 is required to stabilize the DAF-11/TM-GC or another signal transduction component that regulates cGMP levels.
| MATERIALS AND METHODS |
|---|
Culture, strains, and genetics:
C. elegans strain maintenance and genetic nomenclature were as described (BRENNER 1974 ; HORVITZ et al. 1979 ). The following strains were used: N2 (wild type), BW1435 dpy-17(e164) ncl-1(e1865) unc-36(e251); him-8(e1490); her-1(y101hv1) unc-42(e270); ctDp11, CB1364 daf-4(e1364), CB1370 daf-2(e1370), CB1372 daf-7(e1372), DR20 daf-12(m20), DR40 daf-1(m40), DR77 daf-14(m77), DR87 daf-11(m87), JT191 daf-28(sa191), JT195 daf-11(sa195), JT5436 daf-8(e1393), JT5850 dpy-11(e224) daf-21(p673), JT5996 sqt-3(sc63) daf-21(p673) unc-76(e911), JT6130 daf-21(p673), JT6412 daf-11(m84); daf-12(m20), JT6561 daf-11(sa195); daf-12(m20), JT6672 lon-3(e2175) daf-21(p673), JT6857 dpy-17(e164) ncl-1(e1865) unc-36(e251); daf-21(p673); ctDp11, JT6901 dpy-17(e164) ncl-1(e1865) unc-36(e251); daf-11(sa195); ctDp11, JT6917 daf-4(e1364), JT6918 daf-7(e1372), JT6919 daf-14(m77), JT7672 tax-4(ks11), JT7673 tax-4(ks28), JT7674 tax-4(p678), JT8708 lon-3(e2175) daf-21(p673); saEx192 [pEM1; pRF4], JT8710 lon-3(e2175) daf-21(p673) V; saEx193 [pEM1; pRF4], JT7839 tax-4(ks11), JT7840 tax-4(ks28), JT7841 tax-4(p678), JT8712 lon-3(e2175) daf-21(p673); saEx194 [pEM12; pRF4], JT8776 lin-15(n765); saEx207, JT8903 lin-15(n765); saEx237, JT8904 lin-15(n765); saEx238, JT9386 daf11(sa195ts); saEx289, KK 627 itDf2/nT1 n754, LL1008 daf-21(nr2081)/nT1 n754, and MT5813 nDf42/nT1 n754.
To map daf-21(p673) genetically, we crossed him-5 males to sqt-3 daf-21 unc-76/+ + + hermaphrodites, picked non-Sqt non-Unc hermaphrodite cross-progeny, and then picked Sqt non-Unc and Unc non-Sqt recombinants in the next generation. The identification of recombinants was not biased by the daf-21 genotype due to maternal rescue of the daf-21 phenotype. We isolated strains homozygous for each recombinant chromosome and then scored the Him and Daf-c phenotypes. Of 23 Sqt non-Uncs, 15 were Him non-Daf, 6 were non-Him non-Daf, and 2 were Daf non-Him. Of 39 Unc non-Sqts, 25 were Daf non-Him, 9 were Daf Him, and 5 were Him non-Daf. These results are summarized in Fig 5A, and they indicate that daf-21 is roughly two-thirds of the way between him-5 and unc-76.
View larger version (65K):
In this window
In a new window
Download PPT slide
|
Figure 1.
The DAF-11 protein and homology to transmembrane guanylyl cyclases. (A) Splicing pattern of the daf-11 mRNA, as determined by comparison of genomic DNA sequence to cDNA sequence. (B) Predicted DAF-11 protein sequence. The putative transmembrane domain is boxed, the kinase homology domain is underlined, and the guanylyl cyclase domain is underlined twice. (C) Alignment of the guanylyl cyclase domain of DAF-11 and its closest relative outside of C. elegans, the sea urchin speract receptor (accession no. p16065).
|
|
View larger version (35K):
In this window
In a new window
Download PPT slide
|
Figure 2.
Effects of 8-bromo-cGMP on dauer formation. (A) Dose response curve for daf-11, daf-21, and daf-8 mutants at 25° and on the wild type with pheromone. (B) Effect on other Daf-c mutants at 25°. Data for daf-11(sa195) and daf-8(e1393) animals are the same as in A and are shown for comparison. Bars indicate standard errors of the mean.
|
|
View larger version (182K):
In this window
In a new window
Download PPT slide
|
Figure 3.
Expression of daf-11::gfp fusion transgene. (A and B) Fluorescence (A) and Nomarski (B) images of an L1 animal. Arrows in A and B point to the same locations. Expression is seen in the dendrites of amphid neurons (long arrows), in an AWC cell body (short white arrow in B), and an ASJ cell body (brightest spot toward the right in A and short black arrow in B). ASI and ASK (seen in A) are out of the plane of focus. (C and D) Fluorescence (C) and Nomarski (D) images of an adult. Arrows in D point to the sites of fluorescence in C, which are sensory cilia. Expression was also in the cell body (not shown). All photographs are of strain JT9386. (E) With the full length daf-11::gfp fusion, expression was observed in all cells listed, though infrequently in AWB. The short (fifth exon) daf-11::gfp fusion was expressed only in AWB, AWC, ASI, and ASK. NA, not applicable; NT, not tested.
|
|
View larger version (38K):
In this window
In a new window
Download PPT slide
|
Figure 4.
Avoidance of the volatile repellent 2-nonanone. Response of wild type, daf-11; daf-12 double mutants and the daf-12 single mutant control. The daf-12(m20) allele was used in all cases. Response is calculated as [(no. of worms in the third of the plate containing 2-nonanone) - (no. of worms in the third of the plate opposite to the 2-nonanone)]/(total no. of worms). A strong avoidance response is -1. Each data point is the average of two to four assays with >300 worms total. Bars indicate the standard deviation among assays.
|
|
To test the nature of the daf-21(p673) mutation, we used deficiencies that delete the daf-21 gene. Wild-type males were crossed to heterozygous hermaphrodites carrying chromosome V deficiencies balanced by the nT1 n754 translocation. The dominant Unc phenotype of n754 was used to infer the genotype of the progeny. Non-Unc males (Df/+) were crossed to dpy-11 daf-21(p673) hermaphrodites and the progeny were raised and scored at 20° or 25°. Individuals were picked and daf-21(p673)/Df strains were identified based on the segregation of dead eggs. Dpy progeny were assumed to be daf-21(p673) homozygotes, since dpy-11 is linked to daf-21. We tested three deficiencies that delete the daf-21 region: nDf42, itDf2, and yDf8. All gave similar results.
The daf-21(nr2081) deletion was constructed in trans to the nT1 n754 balancer translocation for strain maintenance. Wild-type males were crossed to daf-21(nr2081)/+ heterozygotes and male progeny were crossed to nDf42/nT1 n754 hermaphrodites. Unc progeny were picked, allowed to have progeny, and screened by PCR for daf-21(nr2081). This procedure also served to outcross daf-21(nr2081) twice. We constructed daf-21(p673)/daf-21(nr2081) heterozygotes by crossing wild-type males to daf-21(nr2081)/nT1 n754 hermaphrodites and then crossing non-Unc male progeny to daf-21(p673) hermaphrodites. Non-dauer hermaphrodite progeny were picked individually and transferred to new plates each day for 3 days and were then used for PCR assays of nr2081. Three heterozygotes were obtained and in each case the progeny at 20° were similar. A total of 16% arrested as L1 or L2 larvae, 70% formed dauers, and 14% developed as non-dauers (n = 161).
To create worms with the genotype daf-21(p673)/daf-21(p673)/+, we used ctDp11, a free duplication containing part of chromosome V (including daf-21, her-1, and unc-42) and part of chromosome III (including dpy-17, ncl-1, and unc-36). We crossed dpy-17 ncl-1 unc-36; him-8; her-1 unc-24; ctDp11 males to dpy-17 ncl-1 unc-36; daf-21(p673) hermaphrodites. We picked wild-type progeny [dpy-17 ncl-1 unc-36him-8/+; her-1 unc-42 +/+ + daf-21(p673); ctDp11] and allowed them to self-fertilize. In the next generation, we picked non-Dpy non-Unc dauers (dpy-17 ncl-1 unc-36; daf-21; ctDp11) and then chose strains that segregated no males [him-8(+)]. To assay the Daf-c phenotype at 20°, we let parents carrying the duplication (p673/p673/+) lay eggs overnight and scored the progeny after 3 days. We found 7.2% of the progeny formed dauers (n = 568), in contrast to progeny of daf-21(p673)/+ mothers, which never form dauers.
PCR, DNA sequencing, and DNA oligonucleotides:
Sequencing was performed by the ABI dye terminator cycle sequencing method (Perkin-Elmer, Norwalk, CT) using either AmpliTaq DNA polymerase or AmpliTaq DNA polymerase FS. The PCR products were analyzed by the University of Washington Biochemistry and Pharmacology DNA Sequencing Facilities and by Axys Pharmaceuticals. PCR and sequencing primers were obtained from various sources, and sequences of all listed primers are available on request.
Molecular identification of daf-11:
daf-11(m597) was isolated by P. Albert and D. Riddle from a strain with active transposition of Tc1. Southern blots of genomic DNA from outcrossed m597 strains were probed with Tc1 DNA, and a 0.38-kb NdeI fragment was found to contain a 1.6-kb Tc1 insertion that cosegregated with m597 in recombinants. Genomic DNA from daf-11(m597) was digested with NdeI and used for nested inverse PCR with TC1 primers, and the products were ligated at low concentration to encourage intramolecular ligation. The PCR primers for the first round were oriented outward from the ends of Tc1 (OLG34 and OLG35). The primer for the second round (OLG23) is derived from the inverted repeat sequence at the ends of Tc1 and also oriented outward. The resulting PCR product was cloned into pBluescript II KS+ (Stratagene, La Jolla, CA) that had been digested with EcoRI and treated with Klenow to generate blunt ends to create plasmid pTJ277. We verified that pTJ277 contained DNA flanking the m597 Tc1 by using it as probe on Southern blots of genomic DNA from m597, from four phenotypic revertants of daf-11(m597), and from genetic recombinants that retained m597 but removed most of the rest of the Tc1-mutagenized chromosome V. In each case, polymorphisms were detected consistent with a Tc1 insertion only in the daf-11(m597) strains.
Determination of the daf-11 cDNA sequence:
pTJ277 was used as a probe to isolate genomic and cDNA phage. Five phages with overlapping genomic inserts were isolated at a frequency of 1.8 x 10-4 from the Stratagene 
FIXII genomic library. Sequence analysis was performed on subclones from one of these phages to determine part of the daf-11 genomic sequence. This sequence was also compared to that generated by the C. ELEGANS SEQUENCING CONSORTIUM (1998), which sequenced the region containing daf-11 during the course of this work. daf-11 corresponds to the predicted gene B0240.3. From ~106 plaques from one cDNA library (A. FIRE, personal communication) and 3 x 105 plaques from another (BARSTEAD and WATERSTON 1989 ), one daf-11 clone was recovered. This cDNA insert was subcloned into pBluescript II using the KpnI and SacI sites in the flanking phage DNA to create plasmid pTJ342, which was sequenced. Since this cDNA appeared incomplete on both ends, an additional 9 x 105 plaques from a mixed-stage C. elegans cDNA library (Stratagene) were probed with a 2.4-kb ClaI-HindIII genomic fragment containing most of the cyclase domain. One daf-11 cDNA clone was recovered, and its insert was excised into pBluescript II using helper phage K07, to generate plasmid pTJ584. On the basis of the presence of a poly(A) tail, pTJ584 appears complete at the 3' end, and it was sequenced to determine the 3' untranslated region (UTR) and exon boundaries at the 3' end of the gene. daf-11 mRNA sequence at the 5' end was determined from reverse transcriptase-PCR (RT-PCR) products generated with the GIBCO-BRL (Gaithersburg, MD) 5' RACE system (FROHMAN et al. 1988 ). Mixed-stage RNA for the RT-PCR procedure was isolated either by the method of D. PILGRIM (personal communication) or essentially by the method of MILLER et al. 1988 . The 5' end was isolated in two pieces, one with primers from the coding region and the most 5' segment with a gene-specific primer and a primer to the C. elegans splice leader SL1. Sequencing of bulk RT-PCR product was performed on exons 1, 8, 9, 10, and parts of 2, 7, and 11. This sequence showed no evidence of a mixed population of cDNA. cDNA sequence from the rest of the gene was based on only one cDNA for each section, so alternative splicing cannot be ruled out for these regions.
Determination of daf-21 cDNA sequence:
The 5.8-kb BamHI insert in pEM1 was used as a probe to isolate cDNA phage from a mixed-stage C. elegans cDNA library (Stratagene). Approximately 3% of plaques hybridized to this probe that contains C47E8.4 and daf-21/Hsp90. The inserts for 12 positive plaques were excised into pBluescript II using helper phage K07. On the basis of restriction digest with EcoRI, we found that 11 of the 12 were clearly related. We sequenced the longest, pEM29, with gene-specific primers designed on the basis of the Genefinder (P. GREEN, personal communication) prediction for the Hsp90 coding sequence. This analysis confirmed all of the intron-exon splicing predictions and revealed a 3' untranslated region of at least 117 bp. Our analysis of 12 publicly available 3' expressed sequence tags showed that the five most extensive sequences end at the same point as pEM29, suggesting this is the true 3' end of the mRNA. We did not find evidence of a poly(A) tail. The 5' end of pEM29 is exactly at the predicted ATG start codon. To determine the true 5' end of daf-21, we used the GIBCO-BRL 5' RACE system for RT-PCR (FROHMAN et al. 1988 ). Mixed-stage RNA was isolated as above. A daf-21-specific primer (OLG 344) was used for first strand cDNA synthesis, and the 5' end was amplified with a second daf21-specific primer (OLG367) and an SL1-specific primer (OLG193). Sequencing of the RT-PCR product showed a 5' untranslated leader of 4 bp between the SL1 spliced leader and the daf-21 ATG.
Sequence of daf-11 and daf-21 mutations:
DNA was amplified by PCR from total genomic DNA (WOOD 1988 ) or from lysed worms, and bulk PCR products were sequenced. Mutations were sequenced on both strands with gene-specific primers. For daf-11(m87), daf-11(p169), and daf-11(sa203) only the coding region of the cyclase domain was sequenced. For daf-11(sa195), all coding sequence upstream of the nonsense mutation was sequenced, and for daf-11(m84) the entire coding sequence was determined. For daf-21(p673) the genomic sequence spanning the entire coding region was sequenced. For daf-21(nr2081), only the region immediately adjacent to the deletion was sequenced.
Modification of Genefinder predictions of gcy genes:
We analyzed the Genefinder predictions for gcy-1 to gcy-18, all of the TM-GC genes fully sequenced at the time. The GCY-1 through GCY-18 proteins and a selected set of previously reported TM-GCs were subjected to an initial multiple alignment with CLUSTALW 1.4. We manually inspected the resulting alignments for regions in which specific Genefinder-predicted proteins appeared to have deletions of conserved sequence or insertions of unconserved sequence in a conserved region. Typically, these stood out dramatically in a multiple alignment, and all but one occurred precisely at Genefinder-predicted exon-intron boundaries. In these cases, we manually scanned genomic sequence for splice sites that would produce clearly improved alignments. In four cases, such an alternative splice was found. F23H12.6 (GCY-13) had an anomalous insertion of VSRHENP after predicted amino acid (aa) 614; use of a splice donor 21 nucleotides (nt) upstream (2215 nt from the predicted ATG, splice junction sequence AGT ^ GTGAGTC) precisely eliminated the anomaly. F23H12.6 (GCY-13) also had an anomalous deletion of 10 amino acids (following the F23H12.6 predicted sequence FFSDVVGFT); use of a splice donor 30 nt downstream (4495 nt from the predicted ATG, splice junction sequence CAG ^ GCGAGTT) added 10 amino acids (VLANKSTPLQ) that restored typical similarity to other TM-GCs. [This alteration is questionable because it requires the use of an unusual splice donor sequence. To test the validity of this change, we performed a BLASTP search of the nr GenBank data set with an 18-amino-acid sequence (VGFTVLANKSTPLQVVNL) centered on the inserted 10 amino acids. All 34 hits with P values <0.9 were TM-GCs.] ZK455.2 (GCY-9) had an anomalous insertion of VCKLRQKII after predicted aa941, which was precisely removed by using a splice donor 27 nt upstream (4742 nt from the predicted ATG, splice junction sequence CAG ^ GTTTGCC). B0024.6 (GCY-6) had an anomalous insertion of 18 amino acids (RKIF QKSTNISSSFHLFS) after predicted aa1161 that was precisely removed by creating a new 54-nt intron (starting 5798 nt from the predicted ATG, splice donor sequence AAC ^ GTAAAAT and splice acceptor sequence GTTTCAG ^ CTG). While this approach to amending Genefinder predictions is not rigorous and may have ignored some errors, we think it highly likely that the revised protein predictions more closely approximate the true structures. As a partial test of validity we performed TBLASTN searches (default parameters), using as the query the protein most closely related to the gene under analysis. This permits detection of conserved protein coding regions in a manner independent of Genefinder predictions. In each case, the TBLASTN match to the target gene verified our exon assignment.
The gcy names for TM-GCs correspond to cosmid names from the C. elegans Sequencing Consortium as follows (see also YU et al. 1997 ): daf-11 = B0240.3; gcy-1 = AH6.1; gcy-2 = R134.2; gcy-3 = R134.1; gcy-4 = ZK970.5; gcy-5 = ZK970.6; gcy-6 = B0024.6; gcy-7 = F52E1.4; gcy-8 = C49H3.2; gcy-9 = ZK455.2; gcy-10 = odr-1 = R01E6.1; gcy-11 = C30G4.3 (missing TM domain; see text); gcy-12 = F08B1.2; gcy-13 = F23H12.6; gcy-14 = ZC412.2; gcy-15 = ZC239.7; gcy-16 = F27H7.c; gcy-17 = W03-F11.2; gcy-18 = ZK896.8; gcy-19 = C17F4.6; gcy-20 = F21H7.9; gcy-21 = F22E5.3; gcy-22 = T03D8.5; gcy-23 = T26C12.4; gcy-24 = W03F11.2; gcy-25:Y105C5B.a; gcy-26 = ZK896.8; gcy-27 = C06A12.4. For completeness, we also assigned names to the seven predicted soluble guanylyl cyclases as follows: gcy-31 = T07D1.1; gcy-32 = C06B3.8; gcy-33 = F57F5.2; gcy-34 = M04G-12.3; gcy-35 = T04D3.4; gcy-36 = C46E1.2; gcy-37 = C54E4.3.
8-Bromo-cGMP assays:
Plates (2 cm) were filled with 2 ml NGM agar (WOOD 1988 ) with 8-bromo-cGMP (Sigma, St. Louis) added to a given concentration from a freshly made 250 mM stock. The next day, a fresh overnight stock of Escherichia coli OP50 in Luria broth (LB) was harvested and resuspended at 5% (w/v) in sterile H2O. Twenty microliters of this bacteria was spotted onto each plate and allowed to dry for a few hours. Eight to 20 gravid hermaphrodites were picked on and allowed to lay eggs at room temperature for up to 3 hr and then removed. These plates, generally containing 50–120 eggs, were placed in a sealed box at 25°. Since 8-bromo-cGMP reduced growth synchrony, plates grown at 25° were scored at various times, generally several times for each plate, between 36 and 52 hr after egg laying. Since 8-bromo-cGMP induced dauer recovery in daf-11(sa195) dauers (data not shown), one set of experiments was scored particularly frequently to be certain that the drug blocks dauer formation rather than inducing rapid recovery. In all experiments, obvious dauers, L3s, or older animals were removed at each inspection and counted; the remaining L1 and L2 animals were left to continue growth. At the end of 52 hr, animals that were still L1s or L2s were scored as arrested. The frequency of dauer formation was based only on the dauers, L3s, and older animals and was averaged across all similar assay plates. For most data points, more than 100 worms were counted and assays were performed on at least 2 different days. The following were exceptions: one daf-8 intermediate concentration and daf-11(m87) 1.25 mM assays (done only 1 day each) and daf-8 2.5 mM assay (45 worms). 8-Bromo-cAMP (Sigma) assays were done in the same manner.
In several 8-bromo-cGMP assays, many animals arrested as L1 or L2 larvae or formed larvae with only some characteristics of dauers. We interpret this to mean that 8-bromo-cGMP can affect non-dauer development as well as execution of the dauer developmental program. These animals generally constituted <25% of the total, and they were excluded from the data presented. The following were exceptions to this low frequency: daf-11(m87) 0.6 mM 8-bromo-cGMP (34% not counted); daf-11(sa195) 1.25 and 2.5 mM 8-bromo-cGMP (about 40% not counted); daf-21(p673) 1.25, 2.5, and 5 mM 8-bromo-cGMP (about 50% not counted); daf-4(e1364) 5 mM 8-bromo-cGMP (42% not counted); daf-8(e1393) 2.5 and 5 mM 8-bromo-cGMP (about 50% not counted); daf-2(e1370) 5 mM 8-bromo-cGMP (62% not counted); tax-4(ks11) 5 mM 8-bromo-cGMP (67% not counted); tax-4(p678) 5 mM 8-bromo-cGMP (43% not counted). Conclusions based on these data were essentially unchanged if the partial-dauer animals were counted as dauers.
Construction of daf-11::gfp fusions:
The 4.2 kb upstream of daf-11 plus the entire genomic coding region was cloned in two pieces from PCR products generated from the daf-11-containing cosmid W04E7. Taq polymerase was mixed 100:1 with Pfu polymerase to reduce the mutation rate (after BARNES 1994 ). Primers were designed with restriction enzyme sites to facilitate cloning. OLG376 (XmaI site added) and OLG378 (XhoI site added) were used to amplify the 5' part and OLG377 (XmaI site added) and OLG363 (BamHI site added) were used to amplify the 3' part. OLG376 and OLG377 each changed two nucleotides in intron 6 to produce the XmaI site used in cloning. Sequential cloning of these two PCR products joined them at the shared XmaI site to reconstruct the daf-11 genomic structure. This fragment was subcloned into the SalI and BamHI sites of pPD95.70 to make pTJ642, an in-frame fusion of a nuclear localization signal and green fluorescent protein (GFP) to the DAF-11 C terminus, and into pPD95.79 to generate pTJ643, a similar fusion without a nuclear localization sequence (NLS). A shorter translational daf-11::gfp fusion was generated by cloning a single PCR product (primers OLG311 and OLG310) into the NsiI and BamHI sites of pPD95.67 to create pTJ536. A. FIRE, S. XU, J. AHNN and G. SEYDOUX (personal communication) provided all of the fusion vectors.
Construction of daf-21 subclones:
We constructed subclones of T10E3, a daf-21 rescuing cosmid, by digesting it with either BamHI, PstI, or StuI and shotgun cloning fragments into pBluescript SK+ (Stratagene). One fully rescuing subclone (pEM1) was recovered, which contained a 5.8-kb BamHI fragment. pEM1 was digested with XmaI and a 3.7-kb fragment (shown in Fig 5) extending from the insert XmaI site to the vector XmaI site was subcloned into pBluescript SK+ to create pEM12.
Construction and analysis of transgenic worms:
daf-11 transgenics were generated by injection (MELLO et al. 1991 ) of lin-15(n765) animals, using a lin-15 rescuing plasmid (pbLH98 at 60 ng/µl) as a transformation marker (HUANG et al. 1994 ). The concentrations of key experimental DNAs were: full-length daf-11::gfp fusion (pTJ643) at 200 ng/µl, and daf-11 (exon 5)::gfp fusion (pTJ536) at 50 ng/µl. We also analyzed a full-length daf-11::nls-gfp fusion (pTJ642), injected at 200 ng/µl, which gave nuclear-localized patterns that were otherwise similar to the non-NLS fusion. For pTJ643, injected daf-11(sa195); lin-15(n765) animals were grown at 20° for 1 day and then transferred to 25° and screened for non-Muv (lin-15-rescued) or non-Daf (daf-11-rescued) progeny. For Fig 3E, GFP expression was analyzed in two independent lines containing the full-length, NLS construct [JT8903 lin-15 (n765); saEx237 and JT8904 lin-15(n765); saEx238]. Animals were grown at 25° to enable identification of transgenic animals. GFP expression was observed using epifluorescence microscopy and fluorescent cells were identified by Nomarski microscopy on the same animal. Photographs were taken on Kodak Elite II slide film (ISO 400) and were digitally formatted. Adobe Photoshop was used to adjust brightness and contrast, add annotation, and convert color to gray scale.
For analysis of pheromone effects on GFP expression in daf-11 mutants, lin-15(n765); saEx238 or daf-11(sa195); saEx289 animals were grown on 50–80 µl of pheromone at 25° as described (THOMAS et al. 1993 ), conditions sufficient to induce over 90% dauer formation in the wild type. GFP expression in L1 larvae was observed 1 day after eggs were laid, and dauers were counted and assayed for GFP expression 2 days after eggs were laid.
daf-21 transgenics were generated similarly except that the dominant rol-6(su1006dm) marker (pRF4, 200 ng/µl) was used as the marker. daf-21(p673) mutants have few progeny and injections in this background yielded few transformants, so we injected into the wild type with cosmids (5 ng/µl, individually or in pools) or plasmid subclones (15 or 20 ng/µl). We then crossed the heritable transgenic arrays into a daf-21(p673) background. We used a linked, visible marker to follow daf-21 in crosses because maternal rescue prohibited scoring the Daf-c phenotype in the progeny of daf-21/+ mothers. lon-3 daf-21/++ males were crossed to transgenic Rol hermaphrodites (saEx*), and Rol cross-progeny were allowed to self-fertilize. From plates that segregated Lons, we identified strains of the genotype lon-3 daf-21; saEx*. We found that the Lon phenotype is best scored in adults, when it largely suppresses the Rol phenotype. Therefore, in these strains there are Rol larvae, and most of the adults are Lon non-Rol. In all cases, some dauers segregated indicating that daf-21(p673) must be homozygous. To assay rescue, we determined the frequency of dauer formation in synchronous larval populations grown uncrowded at 25°. Strains with full rescue had only 10–30% dauers, strains with partial rescue had 40–50% dauers, and strains with no rescue had >80% dauers.
Rescue of dauer formation and dauer recovery by saEx289:
Plates were prepared and assayed as for 8-bromo-cGMP assays (above), except that 80 µl of M9, pheromone solution, or a mixture (0, 10, or 20 units of pheromone) were added. Eggs were laid at room temperature for 3 hr and then plates were shifted to 25°. Dauers and L3 or older animals were counted and all L3 or older animals were removed once or twice per day starting at 48 hr. At 96 hr the number of dauers remaining on the plate was counted. The percentage of dauer formation was calculated as 100 x (no. of dauers)/(total no. of worms) at 48 hr, and the percentage of recovery was calculated as 100 x (total no. of recovered animals)/(total no. of recovered animals + dauers left on plate at 96 hr). Dauers that crawled up the side of the plate were not counted, as they did not have a chance to recover. Each data point was repeated on at least 2 days and on four to nine plates.
Chemotaxis assays:
Assays of chemotaxis to NaCl were performed as described (BARGMANN and HORVITZ 1991B ) except that worms were allowed to swim for 15–30 min on a chemotaxis plate without attractant between the last wash and the assay. This step allowed the worms to better acclimate to the assay conditions and led to more reproducible results (C. BARGMANN, personal communication). Volatile avoidance assays with 2-nonanone were performed as described (TROEMEL et al. 1997 ) except that worms were washed three times with S basal and once with water. Worms were grown and assays were performed at 20°. Because a large fraction of daf-11 mutant animals form dauers at 20°, daf-12 double mutants were assayed, with a daf-12 single mutant as control (as in VOWELS and THOMAS 1994 ).
Mosaic analysis:
Four gravid adults of genotype dpy-17(e164) ncl-1(e1865) unc-36(e251); daf-11(sa195); ctDp11 were placed on a plate at 25°. Three days later, non-Unc dauer mosaics were identified either visually or (in most cases) by flooding the plate with 1% SDS and picking animals that were alive and thrashed (1% SDS kills non-dauers and Unc dauers did not thrash). Mosaics were inspected by Nomarski microscopy, and cells were assayed for the presence or absence of the duplication by observing the Ncl (enlarged nucleolus) phenotype of several amphid cells. It was sometimes not possible to unambiguously identify all amphid neurons in each animal. In most cases, ASI, ADL, ASK, ASE, ASH, AWC, AUA, and ASJ neurons on each side were analyzed, and often several other cells were also analyzed.
Identification of the daf-21(nr2081) deletion mutation:
A library of mutagenized worms was screened by PCR with nested primers (LIU et al. 1999 ) to identify a daf-21 deletion called nr2081. The primers were C47E8.5 F1 (5' ATTCGTAATTCCGACCCTGC), C47E8.5 R1 (5' TTCTGTAGATGCGGGAAGCG), C47E8.5 F2 (5' TGCCAAATGAATCAAGCGGC), and C47E8.5 R2 (5' AAGCGTGAGATTGTGGCTCCTC) and resulted in a 2844-bp band in wild type. A mutant with a smaller band was identified, and sequencing of the mutant band showed an 860-bp deletion plus a 3-bp insertion. This mutation is predicted to remove amino acids 32–287 and to add 94 novel amino acids from another reading frame. To follow nr2081 in crosses, we used three primers, two that span the deletion and one within the deletion. Together these primers resulted in diagnostic bands for the mutant and wild type that could be scored simultaneously in heterozygotes.
| RESULTS |
|---|
daf-11 encodes a transmembrane guanylyl cyclase:
We cloned daf-11 using m597, a transposon-tagged allele isolated from a strain with active Tc1 transposition (kindly provided by P. Albert and D. Riddle). Southern blot analysis revealed a Tc1 insertion present in strains containing m597 and absent in recombinants and revertants not containing m597 (data not shown). We cloned genomic DNA flanking this Tc1 insertion by inverse PCR and used the flanking DNA to isolate phages from cDNA and genomic DNA libraries. The two daf-11 cDNAs isolated were both incomplete, but together permitted determination of the last 3 kb of the poly(A)-terminated daf-11 mRNA. We completed sequence at the 5' end of the mRNA from RT-PCR products. We also determined genomic sequence for most of the gene from subclones of genomic phage. The Genome Sequence Consortium completed the sequence of this region during the course of our work. Comparison of the genomic sequence to the cDNA sequence indicated that the daf-11 mRNA contains 17 exons and spans almost 8 kb of genomic DNA (Fig 1A). The mRNA contains one long open reading frame predicted to encode a 1077-amino-acid protein that is a member of the TM-GC family (Fig 1B and Fig C).
Transmembrane guanylyl cyclases catalyze the production of cyclic GMP from GTP and have been identified in animals from C. elegans to humans (reviewed in YUEN and GARBERS 1992 ; DREWETT and GARBERS 1994 ). They function in various signal transduction systems including a chemotaxis response in sea urchin sperm, regulation of blood pressure by natriuretic peptides, and visual signaling in mammals. TM-GCs consist of an N-terminal extracellular domain, a single transmembrane domain, and an intracellular region. The intracellular region contains a domain with homology to protein kinases (KHD, kinase homology domain) and a C-terminal GC domain that catalyzes the conversion of GTP to cGMP. The KHD lacks key residues required for kinase catalytic activity (YUEN and GARBERS 1992 ). The DAF-11 protein contains each of these characteristic domains (Fig 1B). As is true of most TM-GCs, strong DAF-11 homology to other guanylyl cyclases is restricted to the KHD and cyclase domains. An alignment of the cyclase domain of DAF-11 and its closest relative outside of C. elegans, the sea urchin speract receptor, is shown in Fig 1C. The speract receptor and DAF-11 share 22% identity and 47% similarity in the KHD and 43% identity and 64% similarity in the cyclase domain.
We used BLAST searches (ALTSCHUL et al. 1990 ) to identify other guanylyl cyclase-related genes in the nearly complete C. elegans genome sequence (C. ELEGANS SEQUENCING CONSORTIUM 1998 ). As of 10/99, these searches had identified a total of 28 C. elegans genes that are clearly related to TM-GCs. The 27 TM-GC genes other than daf-11 were named gcy-1 through gcy-27 (similar to guanylyl cyclase). Corresponding names from the C. elegans Genome Project are described in MATERIALS AND METHODS and by YU et al. 1997 . On the basis of our analysis of daf-11 and the fact that GFP fusions to several other gcy promoters are expressed in chemosensory neurons (YU et al. 1997 ), it seems likely that most of the C. elegans TM-GCs are involved in chemosensory signal transduction processes.
Seven other genes were identified with higher similarity to soluble guanylyl cyclases (not shown). Since over 95% of all C. elegans genes are included in this analysis (J. SULSTON and R. WATERSTON, personal communication), these observations suggest that all C. elegans guanylyl cyclases fall into one of these two previously defined families. Distinguishing between the TM and soluble GCs was straightforward based both on degree of identity in the cyclase domain and other structural features of each class of proteins.
Sequence changes in daf-11 mutants:
daf-11 alleles vary in their phenotypic effects (THOMAS et al. 1993 ; VOWELS and THOMAS 1994 ). To determine the specific effects of daf-11 mutations on protein function, we sequenced PCR products generated from mutant genomic DNA. Identified mutations are indicated in Fig 1B. Of the known daf-11 mutations, three of the sequenced alleles (m597, sa195, and m87) confer the strongest defects, consistent with a strong loss-of-function or null phenotype: a strong Daf-c phenotype at 25° but not at 15°, chemotaxis defects to nonvolatile and some volatile attractants, and severely defective dauer recovery at all temperatures (VOWELS and THOMAS 1994 ). sa195 is a nonsense mutation at Q450, which is predicted to truncate DAF-11 before the KHD and the cyclase domains; it is the best candidate for a daf-11 molecular null allele. m597 results from a Tc1 insertion in the KHD, in the codon I518. m87 is a missense mutation in the cyclase domain that changes S866 to F in a highly conserved region (Fig 1B and Fig C). A fourth allele, p169, has a missense mutation that changes the highly conserved G867 to R in the cyclase domain. The dauer phenotypes of the p169 mutant are consistent with a strong loss-of-function allele (strong Daf-c at 25°, weak Daf-c at 15°, poor dauer recovery), but other phenotypes have not been tested.
Two other daf-11 alleles confer unusual phenotypes. The daf-11(sa203) mutant was strongly Daf-c at 25°, but also formed 88% dauers at 15° (N = 234). sa203 has a nonsense mutation at Q904, just at the end of the conserved residues of the cyclase domain (Fig 1B and Fig C). We speculate that the sa203 mRNA expresses a truncated protein product that interferes with the function of another protein. Alternatively, a second undetected mutation may be responsible for the unusual phenotype. The daf-11(m84) mutant is also unusual: it is very strongly Daf-c at both 15° and 25°, the dauers recover much more quickly and efficiently than do other daf-11 mutants, and m84 adults have only weak defects in response to the volatile attractant isoamyl alcohol (VOWELS and THOMAS 1994 ). For each of these phenotypes, the unusual daf-11(m84) phenotypes are still seen when in trans to the strong alleles m87 and sa195. m84 is slightly dominant to the wild-type allele for its dauer formation phenotype (VOWELS and THOMAS 1994 ). These data suggest that the m84 mutation affects only some of the functions of DAF-11, and that it may encode a protein that interferes with the function of other proteins. daf-11(m84) causes a G806 to E change, affecting an unconserved residue in a relatively conserved part of the cyclase domain (Fig 1B and Fig C).
cGMP signaling is perturbed in daf-11 and daf-21 mutants:
To test the biological relevance of the DAF-11 sequence homology to TM-GCs and to test directly for a role for cGMP in chemosensation, we assayed the effect on dauer formation of the membrane permeant cGMP analogue 8-bromo-cGMP (Fig 2A). If the daf-11 mutant phenotype is due to a loss of guanylyl cyclase activity, supplementing with 8-bromo-cGMP might suppress the Daf-c phenotype. When grown at 25° with plentiful food without 8-bromo-cGMP, over 95% of daf-11(sa195) and daf-11(m87) animals formed dauers. In contrast, when 5 mM 8-bromo-cGMP was added to the growth medium, <1% formed dauers. Intermediate concentrations of the drug caused intermediate suppression of the Daf-c phenotype. 8-Bromo-cGMP similarly suppressed pheromone-induced dauer formation in the wild type in a dose-dependent manner, directly implicating cGMP in normal dauer formation.
We hypothesized that 8-bromo-cGMP suppressed dauer formation daf-11 mutants by substituting for the cGMP normally synthesized by the DAF-11 protein. This hypothesis predicts that 8-bromo-cGMP would not suppress mutations in genes that act downstream of or in parallel to daf-11. As predicted, we found that 8-bromo-cGMP did not prevent constitutive dauer formation in mutants for daf-1, daf-4, daf-7, daf-8, daf-14, daf-2, or daf-28 (Fig 2B). On the basis of gene interactions, it is thought that all of these genes act downstream of or in parallel to daf-1 (RIDDLE et al. 1981 ; THOMAS et al. 1993 ; GOTTLIEB and RUVKUN 1994 ; MALONE et al. 1996 ). In contrast, 8-bromo-cGMP fully suppressed the Daf-c phenotype of daf-21(p673) (Fig 2A, and see below). To test whether this suppression is specific to cGMP, we tested the response of daf-11, daf-21, daf-1, daf-4, daf-7, daf-14, daf-2, and daf-28 mutants to 8-bromo-cAMP. At 5 mM, 8-bromo-cAMP caused developmental arrest in the wild type, but 0.5 mM 8-bromo-cAMP allowed normal growth and had no effect on dauer formation in any of the Daf-c strains (data not shown). These results implicate cGMP in dauer formation and provide direct evidence that DAF-11 functions in vivo as a guanylyl cyclase.
daf-11 expression in sensory neurons controlling dauer formation and dauer recovery:
To identify cells in which daf-11 is expressed, we constructed a fusion gene containing 4.2 kb upstream of daf-11 and the entire genomic coding region with GFP (CHALFIE et al. 1994 ) fused at the C terminus. A transgenic array containing this construct (saEx289) rescued the Daf-c phenotype of daf-11(sa195) animals (Fig 3E). All GFP-expressing cells were amphid neurons, as identified by comparison to the known positions and processes of all neurons (WHITE et al. 1986 ). GFP expression was visible in the ciliated sensory endings, as expected, as well as in the cell bodies and dendrites (Fig 3). In some adults, expression was seen only in the ciliated endings (Fig 3C and Fig D) and the cell bodies (not shown), indicating that within the dendritic compartment the protein is preferentially localized to the ciliated endings. The expression pattern we observed was generally consistent with phenotypic evidence for daf-11 function in specific sensory pathways, as discussed below and summarized in Fig 3E.
The GFP fusion protein was reproducibly expressed in the amphid neuron classes ASJ and ASI, both implicated in regulating dauer formation. Constitutive dauer formation in daf-11 mutants is dependent on the function of the ASJ neurons (SCHACKWITZ et al. 1996 ), and expression of daf-11::gfp in ASJ neurons supports a model in which the DAF-11 guanylyl cyclase acts in ASJ to regulate dauer pheromone response. A simple possibility is that cGMP produced by DAF-11 inhibits the dauer-promoting activity of ASJ neurons, thus preventing dauer formation, and that dauer pheromone or daf-11 mutations reduce the cGMP level, activating the ASJ neurons and inducing dauer formation.
In contrast to the ASJ neurons, the ASI neurons function together with another neuron class, ADF, to repress dauer formation in the absence of dauer-inducing conditions (BARGMANN and HORVITZ 1991A ; SCHACKWITZ et al. 1996 ). Various results suggest that ASI neurons repress dauer formation by secreting DAF-7, a TGF-ß related protein (BARGMANN and HORVITZ 1991A ; THOMAS et al. 1993 ; REN et al. 1996 ; SCHACKWITZ et al. 1996 ). Expression of daf-11::gfp in ASI was unexpected, since genetic evidence supports a daf-11 function that acts in parallel to daf-7 (THOMAS et al. 1993 ; SCHACKWITZ et al. 1996 ). Several explanations might reconcile these findings. First, one of the many other TM-GCs that have been identified in C. elegans (above; YU et al. 1997 ) may function in ASI neurons redundantly with the DAF-11 TM-GC, thus masking a role for daf-11 in ASI neurons. Second, it is possible that daf-11 function in ASI neurons is unrelated to dauer formation. Third, expression of daf-11::gfp in ASI could be an artifact of our assay: an unexpectedly large fraction of gfp fusions to neuronally expressed genes show expression in ASI neurons (TROEMEL et al. 1995 ; D. A. BIRNBY, E. M. LINK, J. J. VOWELS, H. TIAN, P. L. COLACURCIO and J. H. THOMAS, unpublished data). As an independent approach to testing daf-11 function in dauer formation, we undertook mosaic analysis of daf-11(sa195). Unfortunately, analysis of several putative mosaics from over 250,000 animals screened (see MATERIALS AND METHODS) led to only two conclusions. First, daf-11(sa195) is very slightly dominant in the strain used for mosaic analysis, complicating interpretation of mosaic losses. Second, loss of daf-11 in any of several amphid neurons (including the ASJ class) slightly increased the likelihood of dauer formation, with no clear pattern of critical cells (data not shown). Taken together with previous studies (SCHACKWITZ et al. 1996 ), these results suggest that daf-11 function in preventing dauer formation may be primarily in ASJ neurons, but that daf-11 also acts in other cells.
Under appropriate environmental conditions, dauers recover and resume their normal life cycle (CASSADA and RUSSELL 1975 ). Cell-kill experiments show that ASJ is the only amphid neuron class essential for this recovery (BARGMANN and HORVITZ 1991A ). Strong daf-11 mutants are severely defective in recovery from the dauer state (VOWELS and THOMAS 1994 ). We hypothesized that daf-11 might function in ASJ neurons to promote dauer recovery. To test this hypothesis, we assayed expression of daf-11::gfp in dauers that were induced by dauer pheromone at 25°. Expression in ASJ of the rescuing fusion array saEx238 was analyzed in 13 dauers (26 cells). We saw definite expression in ASJ in 6 cells and expression in 12 more cells whose positions were consistent with ASJ. The arrangement of neurons in the dauer larva is somewhat different from that in the well-described L1 larva, and positive identification was not always possible. These results show that daf-11 is expressed in ASJ neurons in most or all dauers and is consistent with a function of DAF-11 in ASJ neurons to promote recovery from the dauer state.
The dauer-inducing pheromone does not affect daf-11::gfp:
Exposing animals to the dauer-inducing pheromone strongly reduces expression of daf-7::gfp fusions, suggesting that the effects of the pheromone on this TGF-ß pathway are mediated at the level of daf-7 expression (SCHACKWITZ et al. 1996 ; REN et al. 1996 ). In contrast, exposure to dauer pheromone had no effect on expression of a full-length daf-11::gfp fusion in any cells (data not shown), suggesting that pheromone affects this pathway at a level other than daf-11 transcription. Since 8-bromo-cGMP suppresses the Daf-c phenotype of daf-11 mutants and blocks the dauer-inducing effects of the pheromone in wild-type larvae, dauer pheromone presumably reduces cGMP levels in daf-11-expressing neurons. It is possible that pheromone directly inactivates DAF-11 guanylyl cyclase activity, perhaps by binding to the extracellular domain. However, given the pleiotropy of daf-11 mutations, this is an unlikely general model for daf-11 function. We favor a model in which pheromone regulates cGMP levels indirectly, by binding an unidentified receptor or receptors and initiating a transduction process that ultimately stimulates a cGMP phosphodiesterase or reduces DAF-11 guanylyl cyclase activity. A phosphodiesterase is involved in cGMP-mediated visual transduction in mammals, where it is activated by light to lower cGMP levels in rod outer segments (MIKI et al. 1975 ).
daf-11 expression in cells regulating chemotaxis and 2-nonanone avoidance:
daf-11::gfp expression was also seen in AWC and ASK neurons (Fig 3E). The AWC class is required for chemotaxis toward the volatile attractants isoamyl alcohol and benzaldehyde. Since daf-11 mutants are defective in response to these attractants, expression in AWC neurons was predicted. There is evidence that the ASK neurons play a minor role in chemotaxis to lysine and possibly other nonvolatile attractants (BARGMANN and HORVITZ 1991B ). The ASK class may also play a small part in promoting dauer formation in response to pheromone (SCHACKWITZ et al. 1996 ). daf-11 may function in ASK neurons in one of these responses. Expression in AWC and ASK was variable, which may indicate lower levels of daf-11 expression in these neurons or may result from incomplete regulatory sequences in the transgene.
A shorter daf-11 fusion construct, with the same promoter region plus only the first 136 codons of daf-11 (up to the fifth exon) fused to gfp, was expressed consistently in the AWC and AWB neurons and was expressed occasionally in the ASI and ASK neurons (Fig 3E). (Another transgenic array containing the same construct had consistent expression in ASI neurons, but the strain stopped expressing detectable GFP before quantititative data were collected.) Reinspection of strains carrying the full-length daf-11 fusion also revealed occasional expression in AWB cells. Presumably the differences in expression among fusions result from internal daf-11 regulatory sequences, but we have not further investigated this. The AWB neurons are required for avoidance of the repulsive compound 2-nonanone (TROEMEL et al. 1997 ), prompting us to test this response in daf-11 mutants. To facilitate growth, the strains tested also contained a daf-12 mutation to prevent dauer formation, and the daf-12 single mutant was tested as a control (see MATERIALS AND METHODS). We found that both daf-11(sa195) and daf-11(m84) mutants are defective in response to either undiluted or a 10-1 dilution of 2-nonanone (Fig 4). Together these results indicate a role for daf-11 in AWB sensory transduction.
In addition to their other phenotypes, daf-11 mutants are defective in chemotaxis toward nonvolatile chemicals, a process mediated primarily by the ASE neurons. However, daf-11::gfp expression in ASE cells was only rarely seen in larvae bearing the longer fusion and was never seen (0/21 cells) in adults grown at 20° (the temperature at which chemotaxis assays are performed). This lack of expression could be because the fusion lacks regulatory sequences or because the effect of the daf-11 mutation on chemotaxis is not mediated through ASE. We found that the longer fusion could partially rescue the daf-11 chemotaxis defect (Fig 3E), supporting the idea of an ASE-independent effect of daf-11 on chemotaxis. However, daf-11::gfp expression is weak and variable in some cells and it remains possible that a low level of daf-11 expression in ASE neurons was not identified by our reporter assay.
daf-21 encodes an Hsp90:
Genetic analysis has indicated that daf-21 acts at the same step as daf-11 in the dauer formation pathway (THOMAS et al. 1993 ), and daf-11 and daf-21(p673) mutants have nearly identical defects in sensing odorants (VOWELS and THOMAS 1994 ). Furthermore, the suppression of the daf-21(p673) Daf-c phenotype by 8-bromo-cGMP suggests that the daf-21(p673) mutation, like daf-11 mutations, reduces cGMP levels. Therefore, we also cloned the daf-21 gene. We genetically mapped daf-21 with respect to three cloned genes (see MATERIALS AND METHODS) and used the genetic distances to infer an approximate physical position for daf-21 (Fig 5A). We used transformation rescue with cosmids and cosmid subclones to localize daf-21 to a 5.8-kb genomic fragment that fully rescues the Daf-c phenotype of daf-21(p673) (Fig 5B and Fig C). The C. ELEGANS SEQUENCING CONSORTIUM (1998) predicted two divergently transcribed genes in this interval: C47E8.4 and C47E8.5, which encodes the C. elegans heat-shock protein 90. Two lines of evidence indicate that daf-21 corresponds to the Hsp90 gene. First, a subclone (pEM12) that contains the Hsp90 coding region and about 1 kb of upstream sequences partially rescues the daf-21 Daf-c phenotype, despite lacking >70% of the C47E8.4 coding region (Fig 5C). Second, by sequencing daf-21(p673) genomic DNA, we identified a missense mutation in the Hsp90 coding region. This mutation changes E292 to K, a dramatic charge change in a highly conserved amino acid that is likely to affect protein function (Fig 5D and Fig 6).
View larger version (56K):
In this window
In a new window
Download PPT slide
|
Figure 6.
The Hsp90 protein family. Selected members of the Hsp90 protein family are represented as boxes. The two shaded regions in each protein are the conserved N-terminal (left) and C-terminal (right) domains. The number of amino acids in the charged linker region between the two domains is indicated. The percent identity between pairs of proteins is shown for both the N-terminal and C-terminal domains. In the upper representation of DAF-21, the location of the E292K mutation in daf-21(p673) is shown by the black circle, and the extent of the deletion in daf-21(nr2081) is shown by the heavy line. Protein sequences were aligned using the ClustalW program in MacVector (Oxford Molecular Group).
|
|
Hsp90 proteins consist of two highly conserved domains connected by a charged linker region (reviewed in SCHEIBEL and BUCHNER 1998 ; BUCHNER 1999 ). Both the N-terminal and C-terminal domains contain chaperone sites (YOUNG et al. 1997 ; SCHEIBEL et al. 1998 ). The N-terminal domain includes an ATP-binding pocket and a cleft that is large enough to accommodate polypeptides (PRODROMOU et al. 1997A , PRODROMOU et al. 1997B ; STEBBINS 1997). ATP binding is required for Hsp90 function (OBERMANN et al. 1998 ; GRENERT et al. 1999 ) and induces peptide dissociation from the N-terminal domain (SCHEIBEL et al. 1998 ). The charged linker varies in length from 0 amino acids in E. coli to 50 amino acids in humans. This charged region enhances binding of the amino-terminal domain to denatured protein and mediates an effect of bound peptide on ATP affinity (SCHEIBEL et al. 1999 ). The C-terminal domain binds promiscuously to partially folded proteins in an ATP-independent manner (SCHEIBEL et al. 1998 ) and is required for Hsp90 dimerization (NEMOTO et al. 1995 ; MENG et al. 1996 ; NEMOTO and SATO 1998 ). The C. elegans Hsp90 protein includes all of these domains (Fig 6), and the E292K change in p673 affects an early amino acid in the C-terminal domain.
In vertebrates, there are two Hsp90 cytoplasmic isoforms, 
and ß. In addition there is a third cytosolic relative, Trap-1/Hsp75, that lacks the charged region (CHEN et al. 1996 ; SONG et al. 1996 ). A fourth Hsp90 relative called GRP94/GP96 is found in the endoplasmic reticulum. DAF-21/Hsp90 is 74 and 76% identical to human Hsp90 and Hsp90ß, respectively, and is clearly most closely related to these proteins (Fig 6). There are no other predicted C. elegans genes with this degree of similarity to Hsp90, indicating that there is a single C. elegans Hsp90 ortholog. However, there are predicted C. elegans genes orthologous to Trap-1/Hsp75 (R151.7) and GRP94/GP96 (T05E11.3; Fig 6). Hsp90 is a highly abundant protein whose expression is increased further under conditions of stress. Consistent with abundant expression, we and others have identified many daf-21/Hsp90 cDNAs (WATERSTON et al. 1992 ; W. R. MCCOMBIE, J. M. KELLEY, L. AUBIN, M. GOSCOECHEA, M. G. FITZGERALD, A. WU, M. D. ADAMS, M. DUBNICK, A. R. KERLAVAGE, J. C. VENTER and C. A. FIELDS, personal communication; Y. KOHARA, H. MITSUKI, A. NISHIGAKI, T. MOTOHASHI, A. SUGIMOTO and H. TABARA, personal communication). Approximately 3% of clones in a mixed stage cDNA library hybridized to the pEM1 rescuing subclone. It is intriguing that previous studies showed daf-21/Hsp90 transcription is elevated in dauers (DALLEY and GOLOMB 1992 ), though this is unlikely to explain the daf-21(p673) Daf-c phenotype. Sequencing of a cDNA and RT-PCR products confirmed the predicted intron-exon structure and showed that the mRNA is trans-spliced to the C. elegans spliced leader SL1.
daf-21(p673) is not a loss-of-function mutation:
It is known from genetic studies in yeast and Drosophila that Hsp90 is required for viability (BORKOVITCH et al. 1989; CUTFORTH and RUBIN 1994 ; VAN DER STRATEN et al. 1997 ). In contrast, daf-21(p673) mutants have an assortment of sensory defects and reduced fertility, but otherwise grow nicely in the laboratory. We investigated whether daf-21(p673) is a loss-of-function mutation by testing daf-21(p673) in trans to three deficiencies that delete the gene. We crossed Df/+ males to daf-21(p673) homozygous hermaphrodites. If p673 were a loss-of-function allele, we would expect that the daf-21(p673)/Df progeny would form dauers or display some more severe phenotype such as larval arrest (if p673 were a partial loss-of-function allele enhanced by the Df). However, we found that the daf-21(p673)/Df animals were perfectly viable and did not form dauers (Table 1). This result indicates that two copies of daf-21(p673) are required for the Daf-c phenotype and suggests that the Daf-c phenotype is not due to loss of daf-21 function. The deficiencies did have obvious effects on the daf-21(p673) phenotype, however. daf-21(p673) homozygotes have reduced fertility (VOWELS 1994 ), and this phenotype was enhanced by the deficiencies such that the daf-21(p673)/Df heterozygotes are nearly sterile at 25°. In addition, although daf-21(p673)/+ animals never have dauer progeny, ~40% of daf-21(p673) homozygotes from daf-21(p673)/Df mothers form dauers at 20°. These results also confirm that the deficiencies delete the daf-21 gene, as expected from genetic map data. We conclude that daf-21(p673) is not a simple loss-of-function mutation and that dosage of the mutant allele is important in determining the phenotype.
To determine the null phenotype of daf-21, we obtained a deletion mutation kindly provided by Axys Pharmaceuticals, NemaPharm Group. The mutant was isolated according to the method of LIU et al. 1999 (see MATERIALS AND METHODS) and contained an 860-bp deletion plus a 3-bp insertion. This mutation is predicted to remove amino acids 32–287 and to add 94 novel amino acids from another reading frame. daf-21(nr2081) deletion homozygotes arrested growth at the L2 to L3 larval stages (Table 1), indicating that Hsp90 is essential in C. elegans, as it is in yeast and Drosophila (BORKOVITCH et al. 1989; CUTFORTH and RUBIN 1994 ; VAN DER STRATEN et al. 1997 ). To confirm that p673 and nr2081 are allelic, we examined daf-21(p673)/daf-21(nr2081) heterozygotes. We found that these animals have many (70%, n = 161) dauer progeny at 20°, indicating that these mutations fail to complement for the dauer-constitutive phenotype. Of the remaining progeny, about half arrested as L1 or L2 larvae and about half developed as non-dauers. We note that in combination with daf-21(p673), the daf-21(nr2081) deletion mutation has a stronger effect on dauer formation than any of the three large deficiencies tested,
TOP
ABSTRACT
MATERIALS AND METHODS