Distinct roles of transcription factors EGL-46 and DAF-19 in specifying the functionality of a polycystin-expressing sensory neuron necessary for C. elegans male vulva location behavior

Because of its simple nervous system with invariant cell lineage and position, C. elegans provides an excellent model to study how diverse neuronal subtypes are specified (Sulston and Horvitz, 1977; Sulston,1983). The anatomy and interconnectivity of all 118 hermaphrodite neuron types are known (White,1986), as are the molecular details of many neuronal subtypes(Chalfie, 1995). The C. elegans male has 79 additional neurons, falling into 37 classes(Sulston et al., 1980). Most of those male-specific neurons are located in the tail region and contribute to specific motor output during mating behavior(Sulston and White, 1980; Loer and Kenyon, 1993; Liu and Sternberg, 1995; Garcia et al., 2001; Garcia and Sternberg,2003).

During mating, the C. elegans male scans for the vulva by touching the hermaphrodite with the ventral side of his tail and backing along her body. If the vulva is not found, he turns at the hermaphrodite head or tail and scans the other side (Liu and Sternberg, 1995). The male hook sensillum is a copulatory structure that is located just anterior to the cloaca and mediates vulval location behavior (Liu and Sternberg,1995). Intact wild-type males usually stop at their first or second vulval encounter. When the hook sensillum is ablated, operated males circle the hermaphrodite multiple times and fail to stop at the vulva(Liu and Sternberg, 1995). This defect is referred to as the Lov (location of the vulva defective)phenotype (Barr and Sternberg,1999). The hook sensillum consists of five cells, including a structural cell and two ciliated sensory neurons HOA and HOB(Sulston et al., 1980). The two hook neurons have large nuclei and send dendrites into the hook structure;however, their anatomy can be distinguished by cell morphology and synaptic contacts (Sulston et al.,1980). Ablation of either HOA or HOB results in a Lov phenotype,indicating that HOA and HOB have non-redundant functions(Liu and Sternberg, 1995).

The C. elegans homologues of human autosomal dominant polycystic kidney disease genes PKD1 (lov-1) and PKD2(pkd-2) are expressed in the HOB hook neuron(Barr and Sternberg, 1999; Barr et al., 2001; Kaletta et al., 2003). Human PKD genes, which encode divergent members of the TRP family of cation channels, possibly act in signal transduction important for renal epithelial differentiation, as mutations in PKD1 and PKD2 are associated with pathogenic renal cyst formation (reviewed by Wu,2001). In C. elegans, lov-1 and pkd-2 mutations disrupt vulva location behavior, consistent with a defect in HOB sensory function (Barr and Sternberg,1999; Barr et al.,2001). Although LOV-1 and PKD-2 are localized in sensory cilia and cell bodies, the ultrastructure of cilia and dendrites appears normal in lov-1 and pkd-2 mutants(Barr et al., 2001).

Another class of genes required for vulva location affects the formation of ciliated endings in sensory neurons. This class includes che-3, daf-10,osm-5 and osm-6 (Barr and Sternberg, 1999). che-3, osm-5 and osm-6 are required for most or all sensory cilia(Lewis and Hodgkin, 1977; Perkins et al., 1986), while daf-10 only functions in a subset of ciliated sensory neurons(Albert et al., 1981). The hermaphrodite expression of osm-5, a homolog of the mouse autosomal recessive polycystic kidney disease (ARPKD) gene(Haycraft et al., 2001; Qin et al., 2001), and osm-6 has been shown to be regulated by a RFX transcription factor DAF-19, which plays a critical role in general sensory cilium differentiation(Swoboda et al., 2000; Haycraft et al., 2001).

We report the isolation of an allele of egl-46, a putative zinc-finger transcription factor, in a screen for loci required for fate specification of C. elegans hook neuron HOB. egl-46 was previously characterized as a gene when mutated affecting the development of two mechanosensory neurons (FLP cells) (Wu et al., 2001), as well as having defects in the hermaphrodite HSN egg-laying motoneurons (Desai et al.,1988; Desai and Horvitz,1989). We demonstrate that EGL-46 and the transcription enhancer factor (TEF) homolog EGL-44 are expressed in the HOB hook neuron and are required for expression of genes encoding polycystins LOV-1 and PKD-2,homeodomain protein CEH-26, and neuropeptide-like protein NLP-8. egl-44 and egl-46 mutants are defective in vulva location behavior during mating, suggesting compromised normal HOB function. This HOB-specific pathway is distinct from the DAF-19-regulated general cilia formation pathway in sensory neurons. We found that daf-19 acts independently of egl-44 and egl-46 to affect expression of downstream genes in the HOB-specific program, indicating that general and cell-specific regulatory factors work in concert to establish cell-specific features crucial for HOB neuronal function in sensory behavior.

Nematodes were cultured at 20°C as described(Brenner, 1974). All strains used contain him-5(e1490) V to obtain males, except for egl-46(n1127), in which case we used him-8(e1489) IV(Hodgkin et al., 1979). The following alleles were used in this study: daf-19(m86) II(Perkins et al., 1986); egl-44(n1080) II, egl-46(n1127) V(Desai and Horvitz, 1989); pha-1(e2123ts) III (Schnabel and Schnabel, 1990); unc-119(ed4) III(Maduro and Pilgrim, 1995); unc-31(e169) IV, unc-46(e177) V, dpy-11(e224) V(Brenner, 1974); unc-68(e540) V (Lewis et al.,1980); unc-42(e270) V(Riddle and Brenner, 1978); osm-5(p813) X (Dusenbery,1980); and lin-15(n765ts) X(Ferguson and Horvitz, 1985). Integrated GFP fusions or extrachromosomal GFP arrays were: nIs133(pkd-2::gfp) I, nIs128 (pkd-2::gfp) II (H. Schwartz and H. R. Horvitz, personal communication); mnIs17 (osm-6::gfp) V(Collet et al., 1998); syEx301 (lov-1::gfp) (Barr and Sternberg, 1999); myEx256 (osm-5::gfp)(Qin et al., 2001); rtEx227 (nlp-8::gfp) (Nathoo et al., 2001); and saEx490 (daf-19::gfp)(Swoboda et al., 2000).

A 6.4 kb fragment of ceh-26 containing 5277 bp 5′ flanking sequence plus coding sequence to the fourth exon was amplified by long-range PCR using primers P26-22 (GTCCTTTGGCCAATCCCGGGGATCCAGAGCTACTGTTACTTTCAGGGC)and P26-23 (GCCTGCAGAACATTGGCATGTGGCGTCACGGG). BamHI-digested pPD95.77 was joined to the ceh-26 fragment by primer extension and linear amplification (Cassata et al.,1998). The product was cut with PstI and circularized to give plasmid pRFP7. pRFP7 (100 ng/μl) was co-injected with dpy-20(20 ng/μl) into dpy-20(e2017) hermaphrodites as described(Mello et al., 1991). Integration of a transgenic line yielded strain TB1200 with ceh-26::gfp integrated transgene chIs1200 linked to chromosome III. chIs1200 was crossed into him-5(e1490) to yield strain TB1225.

The sy628 allele was generated by mutagenizing the strain TB1225 carrying the HOB marker ceh-26::gfp with EMS using standard protocols(Rosenbluth et al., 1983). In particular, we picked males descended from each single hermaphrodite daughter of mutagenized parents and examined them under a conventional epi-fluorescence microscope for GFP expression. Three-factor mapping of sy628 on linkage group V used alleles of unc-46, dpy-11, unc-68 and unc-42: unc-46 (16/16 recombinants) dpy-11 (0/16 recombinants) sy628; dpy-11 (0/44) sy628 (44/44) unc-42; dpy-11 (4/10) sy628 (6/10) unc-68. During the mapping experiments, the presence of sy628 mutation was determined by loss of ceh-26::gfp expression in HOB.

The ∼0.6 map unit interval between dpy-11 and unc-68was covered by 17 cosmids, including 97 identified genes or predicted coding sequences(www.wormbase.org,version WS74). The sy628 hermaphrodites had a mild egg-laying defective (Egl) phenotype. A previously identified gene associated with an Egl phenotype, egl-46, is located in the middle of that interval. Cosmid K11G9, which contains the entire egl-46 locus, was injected into the strain PS3568 ceh-26::gfp; egl-46(sy628) him-5(e1490) at 40 ng/μl using pmyo-2::gfp plasmid pPD118.33 (5.5 ng/μl) as co-transformation marker (Mello et al.,1991). Three stable lines were obtained from individual F1 progeny that expressed myo-2::gfp in pharynx. Injection of cosmid K11G9 restored the ceh-26::gfp expression in HOB in 76/81 males from three independent transgenic lines. Injection of another cosmid in the same interval, F44C4, which contains a different predicted zinc-finger transcription factor, showed no rescue of HOB expression of ceh-26::gfp in fourteen stable transgenic lines (n=172). Those transgenic lines had a nonsex-specific ectopic expression of ceh-26::gfp in a neuron anterior to HOB, most likely PVT. It is not clear that this ectopic expression is due to injected F44C4 cosmid or interaction between pmyo-2::gfp plasmid and F44C4 cosmid. Cosmids were obtained from the Sanger Institute (Cambridge, UK).

To test for complementation, PS3568 ceh-26::gfp; egl-46(sy628)him-5(e1490) males were crossed to MT2316 egl-46(n1127)hermaphrodites. F1 hermaphrodites with CEH-26::GFP expression were cross progeny, and were examined for an Egl phenotype. F1 males were analyzed by HOB expression of ceh-26::gfp. All 79 sy628/n1127 heterozygous males examined lacked ceh-26::gfp expression in HOB, and heterozygous hermaphrodites were Egl. Thus, sy628 and n1127 fail to complement.

A 2318 bp genomic DNA fragment containing the entire egl-46 coding region was PCR amplified from sy628 mutant DNA using the pair of primers 5′-CTCCCCTTCTTGTAAGGTGTCTT-3′ and 5′-AATTCACTCAGCAATTTGGAAAA-3′. The PCR products from six independent PCR reactions were separately purified using QIAquick PCR purification kit and were pooled together for direct sequencing. Two nested primers, 5′-TTTCGTTCACATCTACCGTAACC-3′ at the 5′ end of the gene and 5′-CGGGGGAAATTGTAAAGAGTTAG-3′ at the 3′ end, and two internal primers, the reverse primer 5′-CCTCTTATGTGCCTTCGTTTTG-3′ at 109-131 bp of the intron 2 and the forward primer 5′-GCTAATGACACCGAGAAAACGAAC-3′ at 274-297 bp of the same intron, were used for sequencing. This sequencing therefore did not cover the 189 bp gap in the intron 2 between reverse and forward primers. The PCR primers and two outside sequencing primers were picked by an oligo design program in the C. elegans genome project at the Sanger Institute(www.sanger.ac.uk/Projects/C_elegans/). The two internal sequencing primers were obtained using Macvector software(Oxford Molecular Group). The G-to-A lesion site at nucleotide 165 of the first exon was observed in both strands.

The N-terminal cfp::egl-46 translational fusion plasmid TU#627 and yfp::egl-44 fusion plasmid TU#628 were kindly provided by Ji Wu and Martin Chalfie. Plasmid DNAs of TU#627 and TU#628 were injected separately into the strain unc-119(ed4); him-5(e1490) at 49 ng/μl. We used 50 ng/μl of pDP#MM016B, a plasmid containing a wild-type copy of the unc-119 gene, as the co-injection marker. Transgenic animals were recognized by rescue of the Unc phenotype of unc-119(Maduro and Pilgrim, 1995). Three independent lines were obtained for each construct and the male expression pattern in those lines was characterized. Transgenic animals generated with the same CFP and YFP plasmids but with myo-2::gfp as a transformation marker had similar expression patterns in the male tail.

The mating behavior of mutant or control males was observed with sluggish unc-31 adult hermaphrodites. All males were isolated from hermaphrodites at the L4 stage and were kept on fresh plates in groups of∼30 animals before observation. For the mating assay, a virgin adult male(12-36 hr post L4 lethargus) was placed on a 0.5 cm bacterial lawn with five 24-hour-old unc-31 hermaphrodite adults(Barr and Sternberg, 1999; Garcia et al., 2001). Each individual male was watched under a Zeiss Stemi SV11 or Wild M420 `Macroscope'for ten vulva encounters or until he stopped at the vulva (pausing for more than 1 second or inserting his spicules), whichever came first. The vulva location efficiency of individuals for a population was calculated as described by Barr and Sternberg (Barr and Sternberg, 1999). To facilitate calculation, the vulva location efficiency of males with more than 10 vulva encounters (pass all ten vulva encounters) was considered to be 0 (actual value≤1/11). The Wilcoxon(Mann-Whitney) test was used to determine statistical significance.

GFP expression was analyzed by conventional fluorescence microscopy (Zeiss Axioskop) using a Chroma Technology High Q GFP long-pass filter set (450 nm excitation, 505 nm emission). CFP and YFP were visualized using a Chroma Technology CFP filter set `31044v2' (exciter D436/20, emitter D480/40,beamsplitter 455dclp) and an YFP set `41029' (exciter HQ 500/20, emitter HQ520lp, beamsplitter Q515lp).

The male tail is remodeled during the L4 stage, undergoing a series of changes in cell shape and position(Sulston et al., 1980). By the late L4 stage, most of the cells that function in adults reach their final locations, and initiate morphological changes to form the adult tail structures. At this stage, a homeodomain-containing putative transcription factor ceh-26 (Bürglin,1994) begins to be expressed and perdures through the adulthood in the HOB hook neuron (Fig. 1A1,A2). Therefore, the presence of CEH-26::GFP indicates a differentiated neuronal fate of HOB. Non-sex-specific expression of ceh-26::gfp is mostly in nuclei of the head (R. F. P. and T. R. B.,unpublished). To identify genes involved in HOB fate specification, we performed a screen for mutants with altered expression of ceh-26::gfpin the HOB cell. This pattern allows for a rapid visual inspection of GFP fluorescence under a compound microscope in the male tail.

One of the mutants recovered from this screen, sy628, failed to express ceh-26::gfp in the HOB neuron of homozygous males with complete penetrance (Fig. 1A3; Table 1). No effect on non-sex-specific ceh-26::gfp expression (e.g., the head nuclei) was observed in sy628 animals, suggesting that the sy628mutation does not cause a general defect in expression of GFP transgenes or of ceh-26 (data not shown). Anatomical examination of sy628males at the third and the fourth larval stages showed that P10.ppap, the presumptive HOB neuron in wild-type animals, was present and occupied its normal position in sy628 mutants. In addition, the hook structure and overall tail morphology appeared normal under Nomarski optics. Initial observations indicated that sy628 males had a decreased mating efficiency. Analysis of their mating behavior determined that sy628males were deficient in vulval location (the Lov phenotype), but had no obvious defect in other steps of mating, such as response, turning, spicule insertion, or sperm transfer. About 97% of wild-type control males stopped at the vulva during the first two vulva encounters (88% vulva location efficiency), as opposed to only 39% of sy628 males(Fig. 2A). On average, sy628 males required more than five encounters to find the vulva,with an overall vulva location efficiency of 36%.

We mapped sy628 to linkage group V between dpy-11 and unc-68, and identified it as an allele of egl-46 (see Materials and methods). egl-46 encodes a putative C2H2-type zinc-finger transcription factor homologous to human and mouse IA1 protein,mouse MLT1 protein and Drosophila Nerfin 1 protein(Wu et al., 2001). The lesion in sy628 mutants was a G-to-A transition at position 165 of the first exon (161-TCTGGAACCCAACGC-175), which changes a tryptophan codon UGG to an UGA opal stop codon. This residue is located at position 55 out of 286 of the inferred EGL-46 protein, before the putative glutamine-rich transcriptional activation domain (residues 61 to 75) and other conserved domains (Wu et al., 2001). This early stop is not necessarily a null allele.

We confirmed the male phenotypes of the egl-46 mutant using a different allele, n1127, which alters the splicing donor of intron 2,located before the region encoding the three zinc fingers of EGL-46 protein(Wu et al., 2001). n1127 and sy628 failed to complement (see Materials and methods). Desai and Horvitz (Desai and Horvitz, 1989) found that n1127 has a decreased male mating efficiency (∼50%). We observed that n1127 males had a Lov phenotype similar to sy628 mutants(Fig. 2B). The vulva location efficiency of n1127 males was 39% (n=17), compared with 94%(n=16) for the control males. There was a marked decrease of ceh-26::gfp expression in n1127 HOB neurons(Table 1). Only two out of 118 n1127 homozygous males examined retained a faint GFP expression in HOB. No altered expression of ceh-26::gfp was detected in cells other than HOB in n1127 mutants.

The hermaphrodite expression pattern of egl-46 has been described by Wu et al. (Wu et al.,2001). Using an egl-46::cfp construct, we analyzed its expression in males and found a similar pattern for non-sex-specific expression (such as the FLP cells, ventral cord neurons and PVD). Both HOA and HOB are born from a single precursor cell (P10.p) at the late L3 stage, and they differentiate into their neuronal fates during the L4 stage. egl-46::cfp was expressed in the HOB neuron beginning at the L4 stage and continuing throughout adulthood (Fig. 3A-D), consistent with the timing of HOB differentiation, and a potential role in the maintenance of HOB function. No detectable expression was seen in the HOA hook neuron. The egl-46 mating defect is reminiscent of ablation of a hook neuron(Liu and Sternberg, 1995). Based on expression of egl-46 gene in a single hook neuron, we infer that the Lov phenotype of egl-46 mutant males is probably due to impaired HOB function.

The dependence of ceh-26::gfp expression on EGL-46 activity suggested that the defective HOB sensory behavior caused by an egl-46mutation could result from loss of HOB-specific gene expression. The C. elegans polycystin genes lov-1 and pkd-2 are expressed in HOB and are required for vulva location(Barr and Sternberg, 1999)(Table 1; Fig. 1B1,B2). To test whether egl-46 regulates these two genes, we used GFP transgenes to visualize their expression in an egl-46(sy628) mutant background. sy628 mutants lacked expression of lov-1::gfp in the HOB neuron (Table 1): only one out of 92 animals examined had detectable expression. The expression of pkd-2::gfp in HOB was also greatly reduced by the sy628 and n1127 mutations of egl-46(Table 1; Fig. 1B3). A neuropeptide-like protein-encoding gene, nlp-8, is also expressed in HOB as well as in the non-sex-specific neuron PVT in the tail(Nathoo et al., 2001). The PVT expression of nlp-8::gfp was not affected by egl-46mutations; however the HOB expression of nlp-8::gfp was absent in about half of egl-46(sy628) males and was decreased in the remainder(Table 1). Therefore, EGL-46 activity is necessary for the HOB expression of all three genes, and the lack of lov-1 and pkd-2 expression could account for the mating defect of egl-46 mutants.

We also observed male-specific egl-46::cfp expression in ciliated ray neurons. The C. elegans male has nine pairs of rays (ray 1-9 for both the left and right sides), each associated with an A-type neuron and a B-type neuron (RnA and RnB, n=1-9)(Sulston et al., 1980). egl-46::cfp was observed in one of the two ray neurons for each ray(Fig. 3D,E); this neuron is probably a B-type neuron because of its co-localization with pkd-2::gfp (data not shown), which is known to be expressed in these neurons (Barr and Sternberg,1999). However, egl-46 regulation was not necessary for lov-1 and pkd-2 expression in ray neurons(Fig. 1B3). lov-1 and pkd-2 mutants show deficiencies in both response and vulva location during mating, correlating with their expression in the B-type ray neurons(except ray 6) and the HOB hook neuron(Barr and Sternberg, 1999; Barr et al., 2001). By contrast, despite egl-46 expression in ray neurons, no obvious defect in either ray neuron expression of PKD genes or response behavior of the mating was detected in egl-46 mutant males. egl-46 might play a major role in HOB sensory specification, but some other factors function in ray neurons.

Wu et al. (Wu et al., 2001)reported that egl-46 acts with egl-44 to specify subtypes of mechanosensory neurons, and for HSN development in hermaphrodites(Desai and Horvitz, 1989). egl-44 encodes a transcription enhancer factor of the TEA domain class (Bürglin, 1991) and is orthologous to the mammalian TEF factors(Wu et al., 2001). We therefore examined the behavior of egl-44(n1080) males, and found that this egl-44 mutation reduced vulva location behavior(Fig. 2C). Similar to egl-46 mutants, egl-44 mutant males passed the vulva frequently and it took an egl-44(n1080) male about five encounters on average to locate the vulva. Specifically, egl-44 mutant males had an overall 43% vulva location efficiency, while control males (wild-type for the egl-44 locus) had an 88% vulva location efficiency.

We determined the male tail expression pattern of the egl-44 gene with the yfp construct described by Wu et al.(Wu et al., 2001). Expression of egl-44 overlapped with but was not identical to that of egl-46. At the L4 stage, the four neurons PVX, PVY, HOA and HOB are positioned in a signature anterior-to-posterior row at the middle left side(Sulston et al., 1980). egl-44::yfp fluorescence was obvious in HOB, PVX and PVY, with HOB usually the brightest, but was barely visible in HOA(Fig. 4A,B). As stated above, egl-46::cfp was only present in HOB. A few neurons anterior to PVX(e.g. PVV) had faint egl-44::yfp expression, as did several cells from the B and Y lineage, including PCB, PCC and PCh(Fig. 4C,D). These cells did not express egl-46::cfp. In addition, almost all the descendants of the ray precursor cells (Rn) expressed egl-44::yfp, including the ray neurons (RnA and RnB) and the ray structure cells (Rnst), all of which are derived from the anterior daughter Rn.a, as well as posterior daughter Rn.p hypodermal cells (Fig. 4E,F;data not shown). EGL-46 showed a more limited expression in the ray lineage. In adults, egl-44::yfp was still expressed in HOB, RnA, RnB and Rnst cells. Hypodermal Rn.p cells no longer displayed bright YFP expression in adults, possibly because of their fusion with the tail hypodermal syncytium. Owing to dramatic changes in cell shapes and positions during the extensive male tail remodeling at the L4-adult transition, the faint egl-44::yfp expression in PCB, PCC and PCh was hard to trace in adults. Overall, egl-44::yfp was expressed more extensively in the male tail than was egl-46. However, a mutation in egl-44 did not result in broader defects in male mating behavior than did an egl-46 mutation.

Based on its behavioral phenotype and its expression in HOB, egl-44 might regulate HOB fate specification, similar to egl-46. We therefore examined HOB-specific gene expression in an egl-44(n1080) mutant background, and found that egl-44mutants displayed a significant decrease in HOB-specific expression of ceh-26::gfp. 50% (49/99) of egl-44(n1080) males lacked ceh-26::gfp in HOB, while the remaining 50% (50/99) had weak HOB expression (Fig. 1A4; Table 1). Reduction of lov-1::gfp expression in HOB by an egl-44 mutation was striking, but only a small effect on pkd-2 and nlp-8expression was observed (Table 1; Fig. 1B4). The lesion in egl-44(n1080) allele is a missense mutation. It is possible that the residual EGL-44 activity in n1080 mutants leaded to an incomplete reduction of HOB gene expression. egl-44 has six differently spliced isoforms(www.wormbase.org,version WS74). The n1040 mutation affects four of them. Currently, we have no information about which isoform might be dominate in the HOB neuron. Expression of the `c' form egl-44 cDNA under control of the 3.1kb egl-46 promoter gave an ambiguous result, with only about 10%restoration of ceh-26::gfp expression in HOB in each of three transgenic lines (data not shown). egl-44 mutants were not defective in ray B neuron expression of lov-1 and pkd-2.

Even though the egl-46 mutations caused a more severe defect in HOB gene expression than did an egl-44 mutation, the Lov phenotypes are similar in male mutants. One possibility is that incomplete decrease of gene expression in the HOB neuron by the egl-44 mutation could reduce the HOB function enough to display a comparable Lov phenotype; however, we cannot rule out the possibility that EGL-44 and EGL-46 might have some distinct targets in HOB. In addition, the faint EGL-44 expression in the HOA hook neuron, as well as in the PCB and PCC neurons of the postcloacal sensilla, might also contribute to the vulva location activity(Liu and Sternberg, 1995). The Lov phenotype is not synergistic in the egl-44; egl-46 double mutant,and there was no observable difference in the efficiency of vulva location compared with single mutants (data not shown). By contrast, C. elegans males with HOB ablated have a 0% vulva location efficiency(Liu and Sternberg, 1995). Both egl-44 and egl-46 mutants had an incomplete loss of nlp-8::gfp expression, but no further elimination of nlp-8::gfp expression was seen in an egl-44; egl-46 double mutant background (Table 1). This lack of enhancement for the Lov phenotype and a defect in nlp-8expression indicates that egl-44 and egl-46 act at least partially in a common pathway for HOB specification. The egl-44;egl-46 double mutant males seemed less active than each of the single mutants and took longer to initiate mating behavior, which might be due to insufficient function of the ray neurons in the double mutant.

In the non-sex-specific FLP cells, wild-type egl-44 is required for normal egl-46 expression (Wu et al., 2001). To determine whether egl-44 and egl-46 regulate each other's expression in the HOB neuron, we introduced an extrachromosomal egl-46::cfp array into an egl-44 mutant, and an egl-44::yfp array into an egl-46 mutant. The timing and relative brightness of egl-46::cfp expression in HOB was not affected in an egl-44(n1080) mutant background compared with a wild-type background,but CFP expression in FLP neurons was reduced. Similarly, no change in the HOB expression of egl-44::yfp was observed in egl-46(sy628)males. We infer that there is no interdependence of egl-44 and egl-46 expression in HOB.

Genes that are expressed in HOB and mutate to a Lov phenotype can be grouped into two separate pathways (Barr and Sternberg, 1999) (this work). osm-5 and osm-6 belong to a general ciliogenic pathway common to all ciliated neurons, including HOA and HOB (Collet et al., 1998; Qin et al.,2001). The other genes discussed above, including egl-44,egl-46, lov-1 and pkd-2, define a program specific for HOB differentiation. We thus asked if there are any interactions between these two pathways; i.e., whether regulators in the cell-specific pathway, egl-44 and egl-46, affect the HOB expression of the general cilium structure genes (osm-5 and osm-6), and whether ciliogenesis might be a prerequisite for execution of an HOB-specific program.

In wild-type males, OSM-5::GFP and OSM-6::GFP are expressed in the cell bodies and dendrites of HOA and HOB at the late L4 stage; their then expression decreases, which is coincident with the formation of ciliated sensory endings in these two neurons(Collet et al., 1998; Qin et al., 2001). Using an integrated osm-6::gfp line (mnIs17) and an extrachromosomal array carrying osm-5::gfp, we found that the HOB expression of these two GFPs at the L4 stage in egl-44(n1080) and egl-46(sy628)mutants was comparable with wild-type(Table 1; Fig. 1C1-C4). osm-5::gfp expression in HOA and HOB was also not affected by egl-46(n1127) (Table 1). In these egl-44 and egl-46 mutant males, the HOB dendritic process, visualized by osm-5::gfp or osm-6::gfp, was extended correctly into the male hook. Neither egl-44(n1080) nor egl-46(sy628) mutants had dye-filling defects (data not shown). We conclude that mutation of either egl-44or egl-46 impedes neither gross cell morphology nor the ultimate neuronal outgrowth and wiring of HOB.

Qin et al. (Qin et al.,2001) showed that an osm-5 mutation affects subcellular localization of LOV-1 and PKD-2, but not their expression. We found that ceh-26::gfp expression was not affected in osm-5(p813)animals. Therefore, it is unlikely that establishment of the HOB-specific program depends on the activities of downstream structure genes (such as OSM-5) in the ciliogenic pathway. The RFX transcription factor DAF-19 is a key upstream regulator of general ciliogenesis(Swoboda et al., 2000; Haycraft et al., 2001). In the male tail, we observed exclusively nuclear-localized GFP expression of daf-19 in male-specific ciliated sensory neurons, including the two hook neurons (Fig. 5A,B) and the 36 ray neurons. The fluorescence in HOA was usually fainter than in HOB. We observed no difference in the HOB expression of daf-19::gfp in egl-44 or egl-46 mutants compared with wild type(Fig. 5C,D). We then analyzed egl-44::yfp and egl-46::cfp in daf-19(m86) mutant males, and found that the timing and relative brightness of expression in HOB was similar to daf-19(+) animals. We infer that, during HOB differentiation, egl-44 and egl-46 are expressed independently of a general cilium formation pathway governed by daf-19.

We next examined the expression of three HOB-specific genes (ceh-26,pkd-2 and nlp-8) in daf-19 mutants. Swoboda et al.(Swoboda et al., 2000) have shown that daf-19 is required for general cilium formation, but not for cell-specific properties. Surprisingly, daf-19(m86) mutants lacked ceh-26 HOB expression (n=87)(Fig. 1A5). Non-sex-specific expression of ceh-26::gfp in some head neurons was also substantially reduced by the daf-19 mutation. All male-specific expression of pkd-2 was diminished in the daf-19 mutant background,including the four ciliated CEM neurons in the head, and the HOB and B-type ray neurons in the tail (n=97)(Fig. 1B5). Only the faint non-sex-specific pkd-2::gfp expression in a few neurons posterior to the nerve ring was retained in daf-19 mutant animals. Similarly,expression of nlp-8::gfp in daf-19(m86) males was only observed in the non-sex-specific PVT neuron and was totally absent in the HOB neuron (n=91). Therefore, complete execution of the HOB-specific program requires DAF-19 activity.

DAF-19 has been proposed to act on the X-box motifs in the cis-regulatory regions of downstream target genes to regulate their transcription (Swoboda et al.,2000). So far, 5′ regions of demonstrated DAF-19 target genes all harbor the X boxes in close proximity to the coding region (the typical spacing is within less than 200 nucleotides upstream). As expected from this hypothesis, expression of X-box-containing osm-6::gfp in the hook and ray neurons was not detected in daf-19 mutants(n=68) (Fig. 1C5). A single X-box sequence is located at about 1.3 kb upstream of the ATG start codon of egl-46. This relatively upstream X box in egl-46promoter was apparently not a functional target site, as egl-46::cfpexpression was not altered in daf-19 mutants. We found no matches to C. elegans X-box consensus sequences in the 5′ regions, introns and immediate 3′ regions of ceh-26, lov-1, pkd-2 and nlp-8. Regulation of ceh-26, pkd-2 and nlp-8 by daf-19 is thus likely to be indirect and mediated by some unknown factor(s), which is probably cell-type specific.

We have found that egl-46 is necessary for vulva location behavior, and for gene expression during HOB differentiation. HOB is a ciliated neuron required for C. elegans males to sense the vulva during mating (Sulston et al.,1980; Liu and Sternberg,1995; Barr and Sternberg,1999; Barr et al.,2001). The regulatory relationships among egl-46, another transcription factor, egl-44, HOB-specific genes and a ciliogenic pathway support a model involving coordinate contributions of general and cell-specific factors to specify a functional HOB sensory neuron(Fig. 6).

To fulfill its sensory function, HOB must build specific structures and express appropriate molecules to receive and transduce signals. In our model,the general cilium formation pathway governed by daf-19 programs HOB to have sensory cilia, and egl-46, partly with egl-44,regulates expression of genes in HOB involved in signal transduction cascades. These two pathways are distinct. Formation of the cilium structures is not necessary for HOB-specific gene expression, and regulators in the cell-specific pathway, egl-44 and egl-46, showed no obvious effect on the HOB expression of the cilium structure genes osm-5 and osm-6. However, these two pathways do interact: not only are they both necessary for HOB function; but the ciliogenic pathway regulator daf-19 has an effect on downstream components of the HOB-specific program without affecting egl-44 or egl-46 expression.

Previous studies suggested that daf-19 is only required for genes functioning in common aspects of cilium formation(Swoboda et al., 2000). We provide the first evidence that daf-19 is required for the expression of some cell-type-specific factors. We propose that daf-19 acts through some unknown factor(s) [which could be an X-box containing gene(s)] to modify HOB-specific gene expression. We observed stronger daf-19::gfpexpression in HOB than in HOA, but whether it is associated with additional daf-19 regulation of HOB-specific gene expression is not known. This daf-19 regulation is not limited to the HOB neuron as daf-19also affects pkd-2 expression in the ray neurons and CEM neurons,indicating some general features are common in this subtype of ciliated sensory neurons. Coupled regulation of general neuronal features and cell-specific identities by multiple transcriptional factors has been found in several different organisms, such as specification of the C. elegansAIY interneuron (Altun-Gultekin et al.,2001), C. elegans olfactory neurons(Troemel et al., 1997) and vertebrate motoneurons (Novitch et al.,2001; Zhou and Anderson,2002), and thus might be a general aspect of the logic of neuronal cell type specification.

Both male hook neurons, HOA and HOB, play a role in vulva location behavior. They both detect the presence of a hermaphrodite vulva, and then produce a distinctive output. This output causes the male to stop at the vulva and to proceed to the next step of mating(Liu and Sternberg, 1995) (M. M. Barr and P.W.S., unpublished). One possible explanation for the functional non-redundancy of HOA and HOB is that they possess different sensory specificity, and hence respond to different cues from the vulva. Another possibility is HOA and HOB might receive the same cues at different times. egl-44 is broadly expressed in many cells of the male tail, but its expression is almost undetectable in HOA. None of the other genes, including egl-46 and its downstream targets in the HOB-specific program described here, is expressed in HOA. The unequal expression of those genes in the two hook neurons provides molecular evidence supporting distinct roles for HOA and HOB in mating.

egl-46 mutations result in an extra cell division in the terminal differentiation of the C. elegans Q neuroblast lineage(Desai and Horvitz, 1989). Loss of either egl-44 or egl-46 function does not cause a cell division defect or a failure in establishment of primary ciliated neural fate during HOB specification. This was determined by anatomical examination and by expression of the cilium structure genes, osm-5 and osm-6. In the non-sex-specific FLP cells, it has been shown that egl-44 and egl-46 act as transcriptional repressors(Wu et al., 2001). They promote the correct subtype of mechanosensory neurons by suppressing expression of genes dedicated to another subtype. Possible positive roles in gene transcription are implicated for egl-44 and egl-46 in the HSN neurons, but no target has been identified(Desai and Horvitz, 1989; Wu et al., 2001). Our data suggest a positive effect of egl-44 and egl-46 on the expression of downstream HOB-specific genes. However, we have not ruled out that EGL-44 and EGL-46 activate gene expression in HOB by repression of a repressor of HOB-specific genes.

We propose that the sensory abilities of the HOB neuron are established by individual cell-specific components regulated by egl-44 and egl-46. One of these components, ceh-26, is the C. elegans ortholog of Drosophila prospero (pros) gene(Bürglin, 1994). pros is involved in the initiation of differentiation in specific neurons following asymmetric cell division(Hirata et al., 1995; Broadus et al., 1998; Manning and Doe, 1999). However, expression of ceh-26 in HOB is not coupled with cell division. Instead, it is expressed at a much later stage, after basic features of cell fate have been established. Similar to HOB, ray B neurons express both egl-44 and egl-46, but unlike HOB, these neurons do not express ceh-26::gfp. Therefore, we think that co-expression of egl-44 and egl-46 is not sufficient to activate ceh-26::gfp in HOB and additional co-factors are also required. The other downstream components, lov-1, pkd-2 and nlp-8, encode proteins that are probably involved in HOB sensory input and output. LOV-1 and PKD-2 accumulate in the sensory cilia and have been proposed to act in a complex; a working model is that LOV-1 is a sensory receptor and PKD-2 is a channel protein (Barr et al.,2001; Koulen et al.,2002). Neuropeptide-like protein NLP-8 might act as a neurotransmitter or neuromodulator released by HOB to mediate the response to the stimuli from the hermaphrodite vulva.

Potential mechanosensory and chemosensory interactions between the male and the hermaphrodite during mating is implied by the vulva location behavior itself, as well as by the requirement of functional ciliated sensory endings in the two hook neurons. Whether HOB is a mechanical sensor or a chemical sensor or both, as is the case for the polymodal ASH neuron(Kaplan and Horvitz, 1993), is not known. Because egl-44 and egl-46 distinguish between mechanosensory neuron subtypes during FLP fate specification, it is possible that these two genes regulate downstream targets that confer mechanosensory ability to the HOB neuron. If so, as members of TRP protein gene family, lov-1 and pkd-2 might be such targets. Known examples of TRP proteins that play a role in mechanotransduction include a C. elegansTRP protein OSM-9 and the Drosophila TRP-like NOMPC protein(Colbert et al., 1997; Walker et al., 2000). Both of these TRP proteins are expressed in mechanosensory neurons and are involved in mechanosensory response.

Human PKD1 and PKD2 were identified as two loci responsible for the autosomal dominant polycystic kidney disease (ADPKD), a genetic disorder that causes renal failure at various ages of adulthood (reviewed by Gabow, 1993; Wu, 2001). Relatively little is known about the regulation of these PKD genes and possible alterations during the disease process. In this work, we showed that expression of C. elegans PKD gene homologs, lov-1 and pkd-2, is affected by transcription factors egl-44 and egl-46. The mammalian TEF proteins, homologous to egl-44, have been implicated in multiple developmental processes (Chen et al.,1994; Jacquemin et al.,1996). Specific expression in kidney was reported for multiple members of TEF proteins (Jacquemin et al.,1996; Kaneko et al.,1997; Jacquemin et al.,1998). C. elegans EGL-46 belongs to a novel zinc-finger protein subfamily. Identified close mammalian homologs of egl-46includes insulinoma associated (IA) proteins, implicated in islet differentiation of the pancreas, and murine MLT 1 protein, silenced in the liver tumors (Goto et al.,1992; Tateno et al.,2001), but their possible roles in the kidney have not been investigated. Progressive cyst formation in ADPKD is not restricted to kidney:involvement of the liver and the pancreas occurs, indicating that those organs suffer similar pathogenesis during progression of the disease(Gabow, 1993; Chauveau et al., 2000). The demonstrated gene regulation network in HOB might reveal important insights into the regulation of human polycystin gene expression.

The dependence of ciliogenesis for the function of PKD-2 may be even more relevant to renal development in mammals. In C. elegans, the ARPKD homolog osm-5 is a direct target of the RFX factor DAF-19(Haycraft et al., 2001),making the requirement of DAF-19 activity for pkd-2 expression particularly interesting with regard to the link between ADPKD and ARPKD. Mammalian polycystins and the cilia of the kidney cells might participate in a common signaling pathway crucial for renal differentiation and function. This hypothesis implies that RFX factor(s) might play a role in the renal development.

We thank Ji Wu, Martin Chalfie, Peter Swoboda, Maureen Barr, Hillel Schwartz, H. Robert Horvitz, Ann Hart and Joan Collet for GFP, CFP, and YFP plasmids and strains. We thank Alan Coulson for providing cosmids, the Caenorhabditis Genetic Center for providing some strains used in this study,and the C. elegans Genome Sequencing Consortium and WormBase for sequence information. We thank L. Rene Garcia, Takao Inoue, Gary Schindelman,Allyson Whittaker and X. Z. Shawn Xu for helpful discussions and critical reading of this manuscript. This work was supported by USPHS grant#1P50DK57325 to G. Germino (co-PI, P. W. S.). P. W. S. is an Investigator with the HHMI. H. Y. was a recipient of a Gordon Ross graduate fellowship. R. F. P. and T. R. B. were supported by grants from the Swiss National Science Foundation. T. R. B. was recipient of a START Fellowship (NF. 3130-038786.93)and is now supported by the Swedish Foundation for Strategic Research(SSF).