Functional Specialization of Sensory Cilia by an RFX Transcription Factor Isoform
Juan Wang, Hillel T. Schwartz, Maureen M. Barr


In animals, RFX transcription factors govern ciliogenesis by binding to an X-box motif in the promoters of ciliogenic genes. In Caenorhabditis elegans, the sole RFX transcription factor (TF) daf-19 null mutant lacks all sensory cilia, fails to express many ciliogenic genes, and is defective in many sensory behaviors, including male mating. The daf-19c isoform is expressed in all ciliated sensory neurons and is necessary and sufficient for activating X-box containing ciliogenesis genes. Here, we describe the daf-19(n4132) mutant that is defective in expression of the sensory polycystic kidney disease (PKD) gene battery and male mating behavior, without affecting expression of ciliogenic genes or ciliogenesis. daf-19(n4132) disrupts expression of a new isoform, daf-19m (for function in male mating). daf-19m is expressed in male-specific PKD and core IL2 neurons via internal promoters and remote enhancer elements located in introns of the daf-19 genomic locus. daf-19m genetically programs the sensory functions of a subset of ciliated neurons, independent of daf-19c. In the male-specific HOB neuron, DAF-19M acts downstream of the zinc finger TF EGL-46, indicating that a TF cascade controls the PKD gene battery in this cell-type specific context. We conclude that the RFX TF DAF-19 regulates ciliogenesis via X-box containing ciliogenic genes and controls ciliary specialization by regulating non-X-box containing sensory genes. This study reveals a more extensive role for RFX TFs in generating fully functional cilia.

CILIA and flagella are highly conserved structures that fulfill important functions in motility, development, and sensation (Marshall and Nonaka 2006; Scholey and Anderson 2006). All cilia and flagella share three general properties: a microtubule-based axoneme, a ciliary membrane containing receptors and channels, and a ciliogenic intraflagellar transport (IFT) machinery (Jekely and Arendt 2006; Satir and Christensen 2007; Sloboda and Rosenbaum 2007; Scholey 2008). Cavalier-Smith (1978) has proposed that the ancestral eukaryote was a ciliated unicellular organism (Cavalier-Smith 1978). The evolutionarily conserved cilium has undergone remarkable diversification in structure, length, morphology, molecular composition, and function (Silverman and Leroux 2009). The principles of ciliary specialization are not well understood.

Regulatory factor X (RFX) transcription factors (TFs) play a conserved role in ciliogenesis in the nematode, Drosophila, zebrafish, and mammals (Swoboda et al. 2000; Dubruille et al. 2002; Bonnafe et al. 2004; Liu et al. 2007; Thomas et al. 2010). RFX TFs share a characteristic winged helix DNA binding domain and bind to an X-box motif in the promoters of target genes (Reith et al. 1990, 1994; Emery et al. 1996; Gajiwala et al. 2000). Studies on Caenorhabditis elegans daf-19, the sole RFX TF in the worm, provided important insight into the role of RFX TFs in regulating ciliogenesis and ciliogenic gene batteries (Swoboda et al. 2000; Blacque et al. 2005; Chen et al. 2006; Senti and Swoboda 2008; Senti et al. 2009). daf-19 null mutants fail to express many ciliogenic genes, lack all sensory cilia, and are defective in many sensory behaviors (Perkins et al. 1986; Collet et al. 1998; Swoboda et al. 2000; Efimenko et al. 2005). daf-19 encodes three isoforms: DAF-19A/B functions in nonciliated neurons to maintain synaptic activity in the adult, while DAF-19C is expressed in all ciliated sensory neurons to regulate ciliogenesis (Senti and Swoboda 2008). The human genome encodes seven RFX TFs (Aftab et al. 2008). In mammals, RFX3 regulates genes involved in ciliary assembly and motility (Bonnafe et al. 2004; El Zein et al. 2009). RFX4 is also a ciliary TF (Ashique et al. 2009). In vertebrates, other ciliary TFs have been identified, including HNF1b, Foxj1, and Noto (Gresh et al. 2004; Beckers et al. 2007; Yu et al. 2008). The mechanisms of RFX target specificity and the relationships between ciliary TFs are not known.

The C. elegans nervous system is well endowed with sensory cilia located on the distal ends of sensory dendrites. Sixty sensory neurons are ciliated in the core nervous system (common in both sexes) (Ward et al. 1975; Ware et al. 1975; Perkins et al. 1986). The male has an additional 48 sex-specific ciliated sensory neurons required for mating behaviors (Sulston et al. 1980; Liu and Sternberg 1995; Barr and Sternberg 1999). These cilia may exhibit unique morphologies, express distinct repertoires of receptors and signaling molecules, and function in a variety of sensory modalities (Bae and Barr 2008). In core amphid sensilla, amphid channel cilia are simple single or biciliated rod-like structures that originate from transition zones to 9 + 0 arrangement of doublet microtubules in middle segments and 9 + 0 singlet microtubules in distal segments (Ward et al. 1975; Perkins et al. 1986). Conversely, the amphid wing neurons (AWA/B/C) possess elaborate branched structures (Ward et al. 1975; Perkins et al. 1986; Mukhopadhyay et al. 2007). The C. elegans Foxj1 homolog fkh-2 is required for AWB specification and is regulated by daf-19 (Mukhopadhyay et al. 2007).

General cilium structure mutants such as daf-19 are defective in multiple sensory behaviors, including male mating behaviors (Perkins et al. 1986; Barr and Sternberg 1999; Simon and Sternberg 2002; White et al. 2007). A subset of 21 male-specific ciliated neurons may be defined by their unique ultrastructural anatomy, functional properties, and gene expression profiles. Four CEM cephalic, one HOB hook, and 16 ray neurons (RnB where n = rays 1–9, but not ray 6) possess B-type cilia that are exposed to the environment and lie adjacent to A-type cilia embedded in the cuticle (Ward et al. 1975; Sulston et al. 1980; Jauregui et al. 2008). The CEM, HOB, and RnB neurons are implicated in chemotaxis to mates, response, and vulva location, respectively (Barr and Sternberg 1999; Chasnov et al. 2007; White et al. 2007; Jauregui et al. 2008). These male-specific neurons express the autosomal dominant polycystic kidney disease (ADPKD) gene homologs lov-1 and pkd-2, the kinesin-like protein gene klp-6, and five coexpressed with polycystin (cwp) genes, herein called the “PKD neurons” and “PKD gene battery,” respectively (Barr and Sternberg 1999; Barr et al. 2001; Portman and Emmons 2004; Peden and Barr 2005; Miller and Portman 2010). Only ray R6B does not possess an exposed cilium or express the PKD gene battery. Consistent with sensory function in RnB and HOB neurons, lov-1, pkd-2, and klp-6 mutant males are response (Rsp) and location of vulva (Lov) defective.

Six IL2 core neurons also express klp-6 (Peden and Barr 2005). The axonemes of the PKD and IL2 neurons are similar in that they possess singlet microtubules of varying numbers, which are distinct from axonemes of the core amphid and phasmid neurons. Although the PKD and IL2 neurons have several features in common, they are also individually specified via distinct lineage-driven mechanisms and express different sets of neurotransmitters and neuropeptides (Lints and Emmons 1999; Nathoo et al. 2001; Shaham and Bargmann 2002; Yu et al. 2003; Lints et al. 2004; Peden et al. 2007; Schwartz and Horvitz 2007). This raises the question of how the shared traits of PKD and IL2 neurons are patterned and how the PKD gene battery is regulated to generate functional sensory cilia.

Previous studies showed that a null allele of daf-19 disrupted pkd-2 expression, although the pkd-2 promoter does not contain an X box (Yu et al. 2003). Here, we identify a cis-regulatory mutation in the daf-19 locus that produces Rsp, Lov, and PKD gene battery expression defects without affecting ciliogenesis. daf-19(n4132) disrupts daf-19m, a daf-19 isoform required for male mating and that specifically acts in PKD and IL2 ciliated sensory neurons. In these neurons, DAF-19M is both necessary and sufficient for activating the PKD gene battery without affecting ciliogenic gene expression or IL2 ciliogenesis. These studies reveal how the complex genomic architecture of an RFX TF enables a single gene to encode a pan-ciliary isoform (DAF-19C) that regulates ciliogenesis and a tissue-specific isoform (DAF-19M) that controls cilia specialization.


Strains, plasmids, and PCR products:

Growth and culture of C. elegans strains were carried out as described (Brenner 1974). Male-enriched strain him-5(e1490)V (high incidence of males) was used as the wild-type reference strain for these studies (Hodgkin 1983). The daf-19 m86, sa190, sa232, m334, and m407 mutant alleles form dauers constitutively, but ∼20–30% are nondauers when raised at 15°. We maintained these daf-19 alleles at 15° and shifted them to 20° one day before the assay. All other strains were grown at 20° unless otherwise stated. Strains used for this study are listed in supporting information, Table S1 and File S1. daf-19 genomic DNA fragments were generated by PCR and described in Table S2. CWP-1∷GFP, Posm-9∷GFP, Punc-119∷daf-19mΔDIM∷GFP, Pdaf-19m∷GFP, and P1-5daf-19m∷GFP were made by PCR–splice by overlap extension (SOE) (Hobert 2002) or cloning into Fire lab vectors, as described in Table S2.

Determining gene expression pattern:

All expression analysis was carried out using transgenic GFP reporters. At least six stable lines were generated and scored for each transgene.

Behavioral assays:

Response and location-of-vulva efficiency assays were carried out according to Barr and Sternberg (1999). Briefly, 12 unc-31(e169) young adult hermaphrodites were placed on a 1-cm bacteria lawn on an NGM agar plate. Males were added to the mating plate and observed for 5 min. Response efficiency reflects the percentage of males that successfully responded to hermaphrodite contact within 10 min. An individual male's vulva-location ability was calculated as inverse of the number of times the male encountered the vulva before successfully locating it. In all experiments, at least 20 animals were scored per experimental trial. Triplicate trials were performed for each line to obtain statistical data. Mating efficiency assays were carried out as described by Hodgkin (1983). Mating efficiencies were calculated as the percentage of cross progeny divided by the number of total progeny.

Dye-filling assay:

A modified dye-filling method was used so that in addition to amphid and phasmid neurons, IL2 neurons were also stained (Burket et al. 2006). Stock solution (2.5 mg/ml) of 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI, Molecular Probe cat. no. D-282) in N,N-dimethylformamide was stored at 4°. Healthy worms were washed off plates with 50 mm calcium acetate, then washed with 50 mm calcium acetate two more times. Worms were immersed in 10 μg/ml DiI in 50 mm calcium acetate and rotated gently for 2 hr. Worms were washed with water twice and plated on fresh bacteria lawns for 30 min before scoring.

Mapping of n4132 and cloning of daf-19m:

Three-factor mapping placed n4132 on chromosome II to the right of unc-4. Deficiency strains were used to map n4132 to the 1.99–2.35 region. mnDf83 and mnDf29 failed to complement n4132, while mnDf71, mnDf25, mnDf28, mnDf12, mnDf22, and mnDf27 complemented n4132. PCR-amplified genomic DNA was used in n4132 rescue experiments. n4132 genomic DNA including 2 kb upstream and 1 kb downstream of the daf-19 gene was sequenced. A deletion flanked by the sequences “CACAAGCCACAAGCTA……GCCACCGCCGAGCCA” [F33H1 GenBank: Z48783.1 10,466–10,969 nucleotides (nt)] was identified in n4132.

daf-19m cDNA isolation:

mRNA was isolated from mixed staged him-5(e1490) worms using the Oligotex mRNA Mini Kit (Qiagen cat. no. 70022). The cDNA corresponding to daf-19m was generated from oligo dT primed first-strand cDNA by using SuperScript III First-Strand Synthesis System for RT–PCR (Invitrogen, Carlsbad, CA), followed by amplification with forward primer 5′-atgagaagagtgtacgaaacg-3′ and reverse primer 5′-gacctgcaggatgatgacga-3′. The daf-19m sequence is available at GenBank (EU812221.1).


Family Relationship II (Brown et al. 2005) was used to identify conserved DNA sequences among Caenorhabditis species.

Microscopy and image analysis:

Live worms were mounted on 2% agarose pads with 10 mm levamisol as described previously (Bae et al. 2006). Fluorescence images were obtained with an Axioplan 2 (Carl Zeiss MicroImaging, Oberkochen, Germany) microscope equipped with a digital CCD camera (Photometrics Cascade 512B; Roper Scientific), captured with Metamorph software (Universal Imaging, West Chester, PA), and then deconvolved with AutoDeblur 1.4.1 (Media Cybernetics). Photoshop (Adobe) was used for image rotation, cropping, and brightness/contrast adjustments.


The n4132 mutation disrupts sensory but not ciliogenic gene expression in male-specific PKD neurons and core IL2 neurons:

To understand how cilia are specialized for sensory functions, we characterized the n4132 mutant, which was isolated on the basis of the loss of Ppkd-2∷GFP expression in male-specific CEM head neurons. n4132 mutant males do not express lov-1, pkd-2, or klp-6 GFP reporters in the male-specific CEMs, HOB, and RnB neurons, herein referred to as the PKD neurons (Figure 1, C, D, G, and H; Table 1). n4132 also disrupts KLP-6∷GFP expression in the core IL2 neurons of males and hermaphrodites from embryogenesis through adulthood (Table 1). Five cwp (coexpressed with polycystin) genes share an identical expression pattern with lov-1 and pkd-2 (Portman and Emmons 2004; Miller and Portman 2010), with a full-length CWP-1∷GFP translational fusion also expressed in IL2 neurons (Table 1). n4132 abolishes CWP-1∷GFP expression in IL2 and PKD neurons (Table 1). The nlp-8 neuropeptide gene reporter is expressed in core neurons (PVT in the tail and some amphid neurons including ASK and ADL in the head) and in the male-specific HOB neuron (Nathoo et al. 2001). n4132 specifically disrupts Pnlp-8∷GFP in HOB without affecting expression in core neurons (Table 1). The TRPV channel osm-9 is expressed in the male-specific PKD neurons, core IL2 neurons, and core neurons in the amphid and phasmid sensilla (Colbert et al. 1997; Knobel et al. 2008). In n4132 animals, osm-9 expression is disrupted only in the IL2 and PKD neurons (compare n4132 in Figure 1, K and L to wild type in Figure 1, I and J; Table 1). In contrast, the daf-19(m86) null allele abolishes osm-9 expression in PKD and all core neurons [compare daf-19(m86) in Figure 1, M and N to wild type in Figure 1, I and J].

Figure 1.—

daf-19(n4132) disrupts expression of sensory but not ciliogenesis genes in core IL2 and male-specific PKD neurons. The cartoon shows the cell body positions of CEM, IL2, and amphid neurons in the male head, HOB, RnB, and phasmid in the male tail. (A, B, E, F) In wild-type males, LOV-1∷GFP and PKD-2∷GFP are expressed in four CEM, one HOB, and 16 RnB (n = 1–9 but not 6) neurons. (C, D, G, H) In n4132 males, LOV-1∷GFP and PKD-2∷GFP expression is completely abolished. Autofluorescence is observed in the intestine as well as the sclerotized hook and spicule structures of the male tail. (I and J) In wild type, osm-9 is expressed more broadly in ciliated sensory neurons, including the male-specific PKD neurons and core IL2, amphids, and phasmids (Colbert et al. 1997; Knobel et al. 2008). (K and L) In daf-19(n4132), osm-9 is not expressed in core IL2, male-specific CEM neurons, and ∼50% of HOB and RnB neurons. In those cells having osm-9 expression, the expression level is reduced (see also Table 1). (K and L) daf-19(n4132) does not affect osm-9 expression in core amphids or phasmids. (M and N) In daf-19(m86) null males, osm-9 expression is completely abolished. (O–R) In wild-type and n4132 males, the osm-6 ciliogenic gene is expressed in all ciliated neurons. Ciliary transition zones and ciliary axonemes are visible in wild-type and daf-19(n4132) males (arrowheads). In daf-19(m86) males, lov-1, pkd-2, and osm-6 expression is completely abolished (Swoboda et al. 2000; Yu et al. 2003). Bar, 10 μm.

View this table:

Comparison of reporter expression pattern in wild type and daf-19 mutant worms

To determine whether n4132 acts specifically in a subset of ciliated sensory neurons or plays a broad role in the ciliated nervous system, we examined a battery of ciliary GFP reporters. n4132 does not affect ciliogenic gene expression of intraflagellar transport (IFT) component OSM-6∷GFP in PKD, core IL2, or other ciliated sensory neurons (compare n4132 in Figure 1, Q and R to wild type in Figure 1, O and P; Table 1). n4132 does not affect ciliogenic gene expression of other IFT reporters including osm-5, bbs-1, bbs-2, bbs-5, and daf-10 in PKD, core IL2, and other ciliated sensory neurons (Table 1). The odorant receptor odr-10 and guanylate cyclase (gcy-5 and gcy-32) genes that are expressed in the core nervous system are not affected by the n4132 mutation (Table 1). We conclude that the n4132 mutation specifically disrupts expression of sensory signaling genes in IL2 and PKD neurons (pkd-2, lov-1, klp-6, cwp-1, nlp-8, and osm-9) without affecting genes required for ciliogenesis or acting in other ciliated sensory neurons.

Like pkd-2, lov-1, and klp-6 mutants, n4132 males exhibit response and Lov defects (Figure 2A). The response efficiency of n4132 males is not significantly different from the lov-1; pkd-2 double mutant (∼35% of mutant males respond to contact with a hermaphrodite compared to 95% of wild-type males, Figure 2A). However, n4132 males exhibit more severe Lov defects than lov-1; pkd-2 mutants (Figure 2A). n4132 mutant males are able to sire cross progeny in 24-hr mating efficiency assays, albeit at lower levels than lov-1; pkd-2 double mutant males (Figure 2B). We conclude that n4132 affects male mating behavior, consistent with a role in functional specialization of PKD ciliated sensory neurons.

Figure 2.—

daf-19(n4132) males are response, Lov (location of vulva), and mating efficiency defective. (A) Response and location of vulva efficiency was scored for each genotype. n4132 males exhibit behavioral efficiency defects when compared to wild type (P < 0.01 for both Response Efficiency and Location of vulva Efficiency assay). n4132 and lov-1;pkd-2 have comparable response efficiencies. n4132 males have a more severe Lov defect than lov-1; pkd-2 double mutants. (B) n4132 males are able to sire progeny, but mating efficiency is significantly lower than lov-1; pkd-2 double mutant (P < 0.01). Error bars indicate standard error of the mean. An asterisk indicates a significant difference between n4132 and lov-1; pkd-2 (*P < 0.01). NS, not significantly different. Statistical analyses were performed by nonparametric Mann–Whitney tests with a two-tailed P-value.

n4132 is a hypomorphic, recessive allele of daf-19:

We mapped n4132 to a small region near the daf-19 locus on chromosome II. Rescue of the PKD-2∷GFP expression, response, and Lov defects was obtained by transforming n4132 animals with cosmid F33H1 or ORF F33H1.1, which contains the RFX TF gene daf-19. To identify the lesion in daf-19, we sequenced n4132 genomic DNA including 2 kb upstream and 1 kb downstream noncoding regions. The n4132 allele contains a 504-bp deletion within the fifth intron of the daf-19 genomic clone (F33H1, GenBank: Z48783.1) deleted 10,466–10,969 nt, Figure 3A). In contrast, the daf-19(m86) null allele introduces a stop codon in exon 7 (F33H1, GenBank: Z48783.1; nt 9760 C to T) before all conserved domains, including the DNA binding domain (DBD) and DNA dimerization (DIM) domain (Figure 3A) (Swoboda et al. 2000).

Figure 3.—

n4132 is a hypomorphic allele of daf-19 that specifically disrupts DAF-19M, an isoform of the RFX TF required for male mating behaviors. (A) Genomic structure of daf-19 isoforms and the locus mutated in n4132. The daf-19(n4132) hypomorphic allele is a deletion in the fifth intron of the daf-19b predicted structure. White boxes are exons, light gray regions are untranslated regions (UTRs). (B) The daf-19m cDNA was amplified by RT–PCR from total mRNA of wild-type mixed-stage and mixed-sex cultures and sequenced (accession no. EU812221). In n4132 animals, the daf-19m but not daf-19a/b cDNA is absent. (C) Diagram of daf-19m genomic rescue of daf-19(n4132) defects in PKD-2∷GFP expression and male mating behavior (RE, response efficiency; LE, location of vulva efficiency). daf-19m cDNA structure compared to the daf-19b isoform. daf-19m uses an internal promoter (in the fifth predicted intron) and a distal upstream element (in the second predicted intron) but shares the DBD and DIM domain with daf-19a/b. daf-19 genomic fragments were scored for the ability to rescue n4132 PKD-2∷GFP expression and behavioral defects (restoration of RE and LE). Both PCR1 and PCR2 fully rescue daf-19(n4132) defects. PCR2 lacks the promoter and two exons of daf-19a/b and contains the daf-19c and daf-19m genomic regions. The shorter PCR3 fragment rescues PKD-2∷GFP expression in tail (HOB and RnB) but not head (CEM) neurons. PCR3 rescues n4132 Lov but not response defects (normal LE, abnormal RE). PCR4, made by introducing the n4132 molecular lesion to PCR3, abolishes rescue of n4132.

daf-19(m86) null animals resemble n4132 mutants in that pkd-2 expression in PKD neurons and nlp-8 expression in HOB is abolished (Yu et al. 2003). However, n4132 is distinct from other daf-19 alleles (m86, sa190, and sa232), which do not express ciliogenic genes like osm-6 and do not form cilia. The daf-19 alleles m86, sa190, sa232, m334, and m407 also result in constitutive formation of dauer larvae (Daf-c) (Malone and Thomas 1994; Swoboda et al. 2000), whereas the n4132 mutant is not Daf-c (Table 2). In daf-19(m86), IL2, amphid, and phasmid ciliated sensory neurons fail to fill with lipophilic fluorescent dyes (Perkins et al. 1986). This dye-filling defective (Dyf) phenotype is indicative of cilium structure defects. In n4132 animals, IL2, amphid, and phasmid cilia are intact, as judged by fluorescent dye uptake (200/200 animals normal in dye-filling assays) and visualization of an OSM-6∷GFP reporter in amphid, phasmid, and male-specific sensory cilia (Figure S1; Figure 1, compare panels O and P with Q and R). That the IL2 neurons of n4132 animals are not Dyf, do express ciliogenic reporters, but do not express sensory genes strongly suggests that daf-19(n4132) affects the function but not development of IL2 cilia.

View this table:

Complementation tests between n4132 and other daf-19 alleles

To genetically confirm that n4132 is an allele of daf-19, we performed complementation tests between n4132 and the daf-19 Daf-c alleles m86, sa190, sa232, m334, and m407 (Table 2). n4132 complements the Daf-c phenotypes of m86, sa190, sa232, m334, and m407. Conversely, the daf-19 alleles m86, sa190, and sa232 failed to complement n4132 PKD-2∷GFP expression defects (Table 2). The Tc1 transposon insertion alleles of daf-19 m334 and m407 map to the fifth intron of the daf-19 genomic clone (Figure 3A), are in opposite orientations (m334 is 5′ to 3′ while m407 is 3′ to 5′) (Swoboda et al. 2000), and exhibit varying defects in PKD-2∷GFP expression (Table 2). m334 is defective in PKD-2∷GFP expression in the CEMs but has normal expression in HOB and RnB tail neurons. The n4132/m334 heterozygote has nearly normal PKD-2 expression: 92% of males express PKD-2∷GFP in CEMs, 100% of males express PKD-2∷GFP in HOB, and 83% of males express PKD-2∷GFP in a subset of RnB neurons. The m407 allele does not affect PKD-2∷GFP and only partially complements n4132. n4132/m407 heterozygotes exhibit normal HOB expression but only express PKD-2∷GFP in only a subset of CEM and RnB neurons. These results indicate that pkd-2 expression in CEM, HOB, and RnB neurons is differentially regulated, that the directionality of the Tc1 insertion impacts PKD-2∷GFP expression, and that n4132 displays complex interactions with m334 and m407. We conclude that the n4132 mutation disrupts a particular aspect of daf-19 function by affecting sensory gene expression in PKD and IL2 neurons, but does not act globally in the ciliated nervous system.

n4132 specifically disrupts the daf-19m isoform:

The predicted daf-19 gene encodes three isoforms, DAF-19A, DAF-19B, and DAF-19C with alternative splicing and an internal promoter (Figure 3A) (Senti and Swoboda 2008). daf-19a/b are required for synaptic maintenance while daf-19c is necessary and sufficient for ciliogenesis (Senti and Swoboda 2008; Senti et al. 2009). We identified a fourth cDNA isoform by RT–PCR using mRNA from mixed-stage, mixed-sex culture (Figure 3B) (accession no. EU812221.1). We refer to this new isoform as DAF-19M for its apparent function in regulating mating behavior and gene expression in PKD neurons. daf-19m starts with an unique exon containing a 166 bp 5′-UTR and an 11-amino-acid (aa)-coding region not present in other isoforms, followed by exon 6 and the DNA sequence shared among all daf-19 isoforms (Figure 3C). Based on the cDNA sequence, daf-19m encodes a predicted 622-aa protein. All four daf-19 isoforms encode the same DNA binding and DNA dimerization domains. In the daf-19(n4132) background, the daf-19m but not daf-19a or daf-19b cDNA is absent as determined by RT–PCR (Figure 3B). From a mixture of cDNA clones, we obtained one copy of daf-19b and 29 daf-19a cDNAs, indicating the latter is more abundant. We infer that daf-19c is intact, because daf-19(n4132) animals form cilia as judged by dye filling, IFT reporters, and non-Daf-c phenotype. We conclude that daf-19(n4132) specifically disrupts daf-19m, a specific isoform of daf-19 required for functional specialization of a subset of ciliated neurons.

To determine how the n4132 lesion affects daf-19m, we identified a minimal-sized fragment capable of phenotypic rescue. PCR2, a PCR amplicon that lacks the predicted promoter and exons 1 to 2 of daf-19a/b (6826–17819 in F33H1 GenBank: Z48783.1), fully rescues the n4132 PKD-2∷GFP expression and mating behavior defects (Figure 3C). A shorter fragment (PCR3) lacking all coding and noncoding regions of daf-19a/b to intron five (6828–14,217 in F33H1 GenBank: Z48783.1) and exons 1 and 2 of daf-19c, rescues PKD-2∷GFP expression in tail but not head neurons. PCR3 also rescues n4132 Lov defects but not response defects. Introducing the n4132 molecular lesion into PCR3 to generate PCR4 (6828–14,217 in F33H1 GenBank: Z48783.1, with deletion 10,466 to 10,969) does not rescue any n4132 defects, indicating that those deleted elements are required to activate PKD-2∷GFP expression in the tail.

daf-19m is expressed in IL2 and PKD neurons via discrete regulatory elements:

To determine DAF-19M site of action, we examined daf-19m expression patterns by fusing the putative daf-19m promoter, a 1-kb region upstream of exon 6 to GFP, resulting in the P1daf-19m∷GFP reporter (Figure 4B). P1daf-19m∷GFP is specifically expressed in the male tail HOB and RnB neurons but not in male-specific CEM head neurons or other ciliated neurons. This result is consistent with PCR3 rescue of PKD-2∷GFP expression in tail HOB and RnB but not head CEM neurons.

Figure 4.—

daf-19m is exclusively expressed in male-specific PKD neurons and core IL2 neurons. (A) Discrete cis-regulatory elements regulate daf-19m expression in head (CEM/IL2) and tail (HOB/RnB) neurons. The CEM/IL2 and HOB/RnB elements are conserved among Caenorhabditis species C. elegans (Ce), C. briggsae (Cb), and C. remanei (Cr). (B) Diagram of daf-19m promoter∷GFP reporters and their relationship to daf-19 and the n4132 deletion. (C and F) In hermaphrodites, P4daf-19m∷GFP is expressed in IL2 neurons (C) and abolished in P5daf-19m∷GFP by deleting the HOB/RnB element (F). (D and E, G and H) In the male, P4daf-19m∷GFP is expressed in IL2 and CEM head neurons (D) and HOB and RnB tail neurons (E), and is completely abolished in P5daf-19m∷GFP, which removes the 13-bp HOB/RnB element (G and H). Bar, 10 μm.

To determine the element(s) responsible for daf-19m activity in core IL2 and male-specific CEM head neurons, we used comparative genomics. The genomes of three Caenorhabditis species (elegans, briggsae, and remanei) contain daf-19, lov-1, pkd-2, and klp-6, suggesting a possible conserved transcriptional regulatory pathway. Interspecific comparisons of noncoding regions enable identification of cis-acting regulatory elements that control gene expression and that are evolutionarily constrained. By comparing intronic daf-19 sequence of C. elegans, C. briggsae, and C. remanei, we identified a 22-bp conserved sequence in intron two of daf-19a/b and outside of the daf-19c regulatory and genomic regions, located −5572 bp from the daf-19m start codon (Figure 4A and Figure S2) (Senti and Swoboda 2008). Adding this 22-bp element to P1daf-19m∷GFP (generating P4daf-19m∷GFP) drives GFP expression in the male-specific PKD head and tail neurons and core IL2 neurons (Figure 4, C–E). We conclude that this 22-bp sequence acts remotely as a daf-19m CEM/IL2 element. The CEM and IL2 neurons are born embryonically, yet daf-19m is expressed in distinct temporal patterns. In core IL2s, P4daf-19m∷GFP is expressed throughout development in males and hermaphrodites. In male-specific CEMs, P4daf-19m∷GFP expression commences in the mid to late L4 male, a stage that coincides with pkd-2 expression and the onset of sexual maturity. daf-19(n4132) animals have a deletion in the HOB/RnB region but retain the CEM/IL2 element, yet do not express the PKD sensory genes in both head and tail neurons. These data indicate that the CEM/IL2 remote element depends upon the HOB/RnB promoter region.

To identify the essential HOB/RnB element, the 504-bp n4132 molecular lesion was introduced to P1daf-19m∷GFP1, generating P2daf-19m∷GFP. This lesion abolished HOB and RnB expression (P2daf-19m∷GFP in Figure 4B). To define the HOB/RnB element, we compared this 504-bp region among C. elegans, C. briggsae, and C. remanei and identified a 13-bp conserved element located −93 bp from the daf-19m start codon (Figure 4A). Deletion of this 13-bp element (P3daf-19m∷GFP) destroyed HOB and RnB expression of daf-19m (Figure 4B). Moreover, daf-19m expression from the CEM/IL2 element depends on this 13-bp element (P5daf-19m∷GFP in Figure 4, F–H). In summary, we identified two elements that confer precise spatial and temporal regulation to daf-19m: the HOB/RnB element and the CEM/IL2 remote enhancer element.

DAF-19M is sufficient to drive PKD-2∷GFP expression and localization in core IL2 and male PKD neurons:

To test daf-19m function, we fused a daf-19m cDNA or genomic fragment to GFP reporter and scored for rescue of n4132 PKD-2∷GFP expression defects. DAF-19M is nuclear localized (Figure S3A), while PKD-2∷GFP is nonnuclear and localized to the cell body and cilia (Figure S3B). The distinct DAF-19M nuclear and PKD-2 cytoplasmic/ciliary subcellular distribution patterns enable visualization of both reporters (Figure 5, B–J; Figure S3C). Expression of the full length daf-19m cDNA using the daf-19m P1 promoter did not rescue PKD-2∷GFP expression in n4132 tail HOB/RnB or head CEM/IL2 neurons (data not shown). Swoboda et al. (2000) reported that a genomic segment (from 3 kb upstream of daf-19a/b start codon to the DNA binding domain coding region) fused to GFP (pJT1228) fully rescued daf-19(m86) ciliogenesis defects, indicating that the DIM domain is not required for transcription activation activity of daf-19 (Swoboda et al. 2000). We therefore generated a daf-19m genomic fragment containing the P1 promoter, 5′-UTR and regions up to and including the DNA binding domain-encoding region fused to GFP reporter (P1daf-19m∷DAF-19ΔDIM∷GFP). This construct rescued n4132 PKD-2∷GFP expression and localization in male tail HOB and RnB neurons. Including the 22-bp CEM/IL2 remote element in P4daf-19m∷DAF-19ΔDIM∷GFP induced PKD-2∷GFP expression and localization in CEM, HOB, and RnB neurons, similar to PKD-2∷GFP expression and localization in wild-type males. In PKD neurons, daf-19m is necessary and sufficient for functional specialization neurons, is not autoregulated, and does not require a DIM domain in this context. In IL2 neurons, P4daf-19m∷DAF-19ΔDIM∷GFP ectopically induced PKD-2∷GFP expression and ciliary localization in both male and hermaphrodite throughout embryogenesis, larval development, and adulthood in the n4132 background (Figure 5A). This result suggests that the DIM domain may be required for regulation of DAF-19M activity in IL2 neurons.

Figure 5.—

daf-19m function is restricted to IL2 and PKD neurons and is sufficient for PKD-2 expression. (A) Genomic structure and rescuing fragments of daf-19. daf-19 isoforms differ in the 5′ exons. Isoform-specific exons are labeled accordingly (a, b, c, or m); no label indicates an exon conserved among all four isoforms. P1daf-19m∷DAF-19MΔDIMGFP rescues PKD-2∷GFP expression in the tail but not head neurons of n4132 males. Including the 22-bp CEM and IL2 element in P4daf-19m∷DAF-19MΔDIM∷GFP rescues PKD-2∷GFP expression in head and tail neurons of n4132 animals. (B–J) Pan-neuronal expression of daf-19m (Punc-119∷DAF-19MΔDIM∷GFP) is sufficient to drive PKD-2∷GFP expression and localization in core IL2 and male-specific PKD neurons, but is insufficient to induce PKD-2∷GFP expression outside of the endogenous daf-19m expressing neurons. (B and C) In wild-type hermaphrodites and males, pan-neural expression of daf-19mΔDIM is sufficient to induce PKD-2∷GFP expression and ciliary localization in core IL2. (C and D) In wild-type males, PKD-2∷GFP expression remains restricted to IL2 and male-specific PKD neurons with Punc-119∷DAF-19MΔDIM∷GFP. (E–J) In daf-19(n4132) and daf-19(m86) backgrounds, pan-neuronal expression of daf-19mΔDIM rescues PKD-2∷GFP expression in male-specific PKD neurons and ectopically in core IL2 neurons, indicating that DAF-19M is necessary and sufficient for pkd-2 expression. PKD-2∷GFP is nonnuclear and localizes to cell bodies (small arrows) and cilia (arrowheads) or dendritic tips in daf-19(m86) (H–J). Bar, 10 μm.

To determine whether DAF-19M function is sufficient and independent of DAF-19C, we used a pan-neural promoter (Punc-119) to drive expression of the genomic DAF-19MΔDIM∷GFP fragment (Punc-119∷DAF-19MΔDIM∷GFP) in wild-type, daf-19(n4132), and daf-19(m86) backgrounds. In the daf-19(m86) null mutant, Punc-119∷DAF-19MΔDIM∷GFP does not rescue dye-filling phenotype in amphid, phasmid, or in IL2 neurons (0%, n = 100), indicating that daf-19m cannot substitute for daf-19c function in ciliogenesis. However, Punc-119∷DAF-19MΔDIM∷GFP does activate PKD-2∷GFP expression in male-specific PKD and core IL2 neurons of daf-19(m86) mutants, indicating that daf-19m is sufficient to drive pkd-2 expression independent of daf-19a/b/c (Figure 5, H–J).

In all three genetic backgrounds, Punc-119∷DAF-19MΔDIM∷GFP does not induce expression of PKD-2∷GFP in other ciliated neurons, with exception of IL2 neurons. This observation suggests that DAF-19M function is restricted to PKD and IL2 neurons by unidentified positive and/or negative regulators. Similarly, PKD-2∷GFP ciliary localization requires cell-type specific transporting components that are only present in PKD neurons (Bae et al. 2006), but have the potential to be ectopically expressed in IL2 neurons by DAF-19MΔDIM overexpression. In IL2 neurons overexpressing daf-19mΔDIM (Punc-119∷DAF-19MΔDIM∷GFP or P4daf-19m∷DAF-19MΔDIM∷GFP), PKD-2∷GFP is expressed and localized to cilia or distal dendrites in cilia-less daf-19(m86) animals, indicating that the cell type-specific transporting components are under the control of daf-19m.

egl-46 acts upstream of daf-19m in a cell type-specific manner:

Previous work of Sternberg and colleagues showed that the zinc finger TF EGL-46 is a cell-specific TF that regulates HOB differentiation, while daf-19 regulates lov-1 and pkd-2 expression in all PKD neurons (Yu et al. 2003). The GFP-tagged daf-19 genomic clone that rescues daf-19 ciliogenesis defects is expressed in all ciliated neurons in the hermaphrodite as well as in all 36 ray neurons (RnA and RnB) and both hook neurons (HOA and HOB) in the male tail (Swoboda et al. 2000; Yu et al. 2003). Yu et al. (2003) also showed that daf-19(m86) is not defective in egl-46 expression, and egl-46 is not required for daf-19 expression, leading to the proposal that daf-19 and egl-46 act in distinct pathways to regulate gene expression in the HOB neuron.

On the basis of work by the Swoboda lab (Senti and Swoboda 2008) and shown here, we now know that the GFP-tagged daf-19 genomic reporter (2.9 kb of the daf-19 promoter and ∼10 kb of daf-19 genomic sequence just downstream of the DBD domain fused to GFP) includes the promoters of all daf-19 isoforms: a, b, c, and m (Figure 3A). To determine the relationship of daf-19m and egl-46, we examined daf-19m expression in egl-46 mutant males. Surprisingly, we found that egl-46 is required for daf-19m expression only in HOB, but not in any other cell types (Figure 6). egl-46 hermaphrodites exhibit normal daf-19m expression in IL2 neurons (n = 20 animals, 120/120 cells; Figure 6, A and D). In egl-46 males, 20/22 males do not express daf-19m in HOB neuron (Figure 6F), while 22/22 males exhibit normal daf-19m expression in head CEM and IL2 neurons (Figure 6E) and tail RnB (Figure 6F). We conclude that that egl-46 acts upstream of daf-19m to regulate PKD gene expression specifically in the HOB neuron but not other PKD neurons.

Figure 6.—

egl-46 is required for daf-19m expression in the HOB neuron. (A and D) In wild-type and egl-46 hermaphrodites, daf-19m is expressed in IL2 neurons. (B and C) In the wild-type male, daf-19m is expressed in IL2 and CEM neurons in the head and the HOB and RnB neurons in the tail. (E) In the egl-46 male head, daf-19m is expressed in IL2 and CEM neurons. (F) In the egl-46 male tail, daf-19m expression is lost in the HOB neuron but intact in RnB neurons. Arrowheads point to rays 4 and 5.


RFX TFs perform an evolutionarily conserved role in ciliogenesis (Silverman and Leroux 2009; Thomas et al. 2010). Cilia are highly specialized for functions in signal transduction, development, or motility (Marshall and Nonaka 2006). On the other hand, all cilia and flagella are built by the evolutionarily conserved intraflagellar transport (IFT) machinery, which was first identified in the unicellular algae Chlamydomonas (Kozminski et al. 1995). By contrast, ciliogenesis in metazoa depends on RFX TFs (Chu et al. 2010). The C. elegans genome encodes a sole RFX TF DAF-19, in contrast to two in Drosophila melanogaster, nine in Danio rerio, and seven in the genomes of Mus musculus and Homo sapiens (Chu et al. 2010; Thomas et al. 2010). Swoboda and colleagues demonstrated that the daf-19 locus encoded three isoforms, with the daf-19c isoform being necessary and sufficient for ciliogenesis (Swoboda et al. 2000; Senti and Swoboda 2008; Senti et al. 2009). Our study provides the first evidence that a fourth tissue-specific isoform, daf-19m, regulates the functional specialization but not formation of cilia.

C. elegans diversifies gene function by internal promoters that generate spatial–temporal specific isoforms (Choi and Newman 2006). As the sole RFX TF in C. elegans, daf-19 encodes distinct isoforms that function in synaptic maintenance (DAF-19A/B), ciliogenesis (DAF-19C), and ciliary functional specialization (DAF-19M) (Senti and Swoboda 2008). Senti et al. (2009) showed that in some core ciliated neurons (ASER, ADF, ASH, AWC, PHA, and PHB) expression of the daf-19c isoform is necessary and sufficient to generate a fully functional cilium. However, in another ciliated neuron (ASJ), daf-19c expression is insufficient to rescue the ASJ dye-filling defect of daf-19(m86) null animals, suggesting additional function from the daf-19 locus is required to generate a fully functional cilium in some contexts. Our work shows that the daf-19m isoform is required for cilia specialization in the male-specific PKD neurons and core IL2 neurons. The daf-19m isoform is not required for dauer formation, ciliogenesis, or for expression of ciliogenic genes. Conversely, the daf-19 Tc1 insertion allele m407 is Daf-c without affecting PKD-2∷GFP expression, indicating that these daf-19 functions are genetically separable. Expression of daf-19m using a pan-neural promoter does not rescue daf-19(m86) dye-filling defects but is sufficient to activate PKD-2∷GFP expression and localization. These results indicate that daf-19m specifically regulates genes required for sensory signaling in a daf-19c-independent manner. We conclude that in some sensory neurons, daf-19c is sufficient for the development and functional specialization of a cilium, whereas in PKD and IL2 neurons, pan-ciliary daf-19c regulates generic ciliogenesis while the tissue-specific daf-19m isoform controls functional specialization.

This model raises the question of how similar and coexpressed RFX isoforms target different gene batteries (ciliogenic vs. sensory genes). In mammalian genomes, genes with alternative promoters are common (Davuluri et al. 2008). Mammalian RFXs have alternative isoforms, with RFX3 encoding eight (Aftab et al. 2008) and RFX4 encoding six (Zhang et al. 2007). Similar to the case for daf-19, RFX3 regulates two categories of genes: those involved in ciliary assembly and in ciliary motility (El Zein et al. 2009). Rfx4_v3 is a novel brain-specific isoform, which was identified on the basis of mutant phenotype of transgene inserted into an intron (Zhang et al. 2007), which is reminiscent of daf-19m identification, based on n4132 mutant phenotypes. RFX TFs bind to the X-box motif, which is common in the genome. In C. elegans, there are 700 predicted X-box genes (Efimenko et al. 2005; Chen et al. 2006). In human, the L1 regulatory motif that is similar to X box is most highly enriched regulatory motif in the genome (Xie et al. 2007). In addition to the canonical X-box promoter motif, genes expressed in ciliated cell types may have half an X box, a modified X box, or as of yet unidentified regulatory elements in their promoters (Piasecki et al. 2010). Given the multiplicity of X boxes, noncanonical X boxes, and RFX TFs, how is ciliary gene expression temporally and spatially regulated?

DAF-19C target genes contain authentic X-box motif in the proximal promoter region (GTHNYY AT RRNAAC within −250 bp from the start codon) (Efimenko et al. 2005; Senti and Swoboda 2008). However, DAF-19M isoform target genes do not have a canonical, half, or modified X box in their promoters. In a similar scenario, FKH-2, the C. elegans homolog of Foxj1, the master regulator of motile cilia in mammals, is required for the specialization of AWB cilia, regulated by daf-19, but does not contain an X box in its promoter (Mukhopadhyay et al. 2007). Hence, the daf-19 locus might regulate target genes in different ways: generic cilia development genes may be directly regulated by DAF-19C via binding to X-box motif, whereas the cilia specialization genes may be directly regulated by DAF-19 via unidentified promoter motifs or indirectly regulated by DAF-19 through TF cascades.

In the HOB neuron, the EGL-46 zinc finger TF acts upstream of daf-19m to control PKD gene battery expression. In other PKD neurons (IL2s, CEMs, and RnBs), egl-46 plays no apparent role in daf-19m or PKD gene battery regulation. EGL-46 shares sequence similarity with mouse Insm1, a zinc finger TF essential or pancreatic islet cell development (Gierl et al. 2006). Intriguingly, Rfx6 is coexpressed with Insm1 in mouse embryonic E15.5 pancreas (Soyer et al. 2010). In the mouse pancreas, Rfx6 is not required for ciliogenesis or expression of cilia genes (Smith et al. 2010). Rather, authors propose that Rfx3 and Rfx6 heterodimer or Rfx6 homodimer may regulate genes involved in islet development, but not ciliogenesis (Smith et al. 2010). The epistatic relationship between Rfx6 and Insm1 in mouse is not known. Given the relative simplicity of C. elegans, the identification of daf-19m direct targets, PKD gene battery cis-regulatory elements, and PKD gene battery direct regulators will reveal how ciliated cells are molecularly programmed to generate a functionally specialized cilium.

DAF-19M shares the conserved DBD and DIM domains with known DAF-19 isoforms. RFX homo- or heterodimer formation may transcriptionally activate or repress different sets of genes according to cell type-specific functional requirements. The RFX1 DIM domain may play a repressive function (Katan et al. 1997). For example, the Saccharomyces cerevisiae RFX CRT1 functions as a transcriptional repressor in a DNA damage and replication block checkpoint pathway (Huang et al. 1998). The DIM domain is not required for daf-19c or daf-19m function in ciliogenesis or sensory gene expression in PKD neurons, respectively. However, pan-neural expression of DAF-19MΔDIM ectopically activates PKD-2∷GFP expression and localization in IL2 neurons, suggesting the DIM domain may have repressor function in this neuronal context. The four isoforms of daf-19 use distinct promoters for distinct spatial and temporal regulation, yet differ only slightly in amino-terminal sequence. DAF-19A and DAF-19B differ in only 25 amino acids, although the function of the two has not been separated. DAF-19C and DAF-19M have only 22 and 11 unique amino acids, respectively. At present, we do not know the role of these short, unique N-terminal sequences in DAF-19 regulation or function, nor do we know the function of the daf-19m 5′-UTR on mRNA stability or translation.

The IL2 and CEM neurons are born during embryogenesis, whereas the tail HOB and RnB neurons are born during the fourth larval (L4) stage before adulthood. daf-19m and the PKD gene battery are not expressed in CEM neurons until the late L4 stage. Expression of the daf-19m isoform is spatially regulated by two hierarchal cis-regulatory elements. An internal promoter activates daf-19m expression in RnB and HOB neurons after ray and hook sensilla development. Adding a remote enhancer element to the RnB/HOB promoter drives daf-19m expression in IL2 and CEM head neurons. The CEM/IL2 enhancer sequences are AT enriched. An ARID family TF CFI-1(CEM neuron Fate Inhibition 1) binds to an AT rich region and is required for the CEM fate inhibition (Shaham and Bargmann 2002). However, cfi-1 is not required for pkd-2 expression in CEM neurons (Shaham and Bargmann 2002). Identifying the TFs that activate daf-19m, daf-19m direct targets, and PKD gene battery regulators will reveal how a cilium is specialized in form and function.


We thank H. R. Horvitz, in whose lab the n4132 mutant was isolated; P. Swoboda and G. Senti for sharing unpublished data and stimulating discussions; P. Anderson, J. Kimble, Q. Mitrovich, and P. W. Sternberg for critical intellectual input in the early stages of this work; Barr and Swoboda lab members for constructive criticism of the manuscript; A. Hart, M. Leroux, D. Riddle, P. W. Sternberg, and the Caenorhabditis Genetics Center, which is funded by the National Institutes of Health (NIH) National Center for Research Resources (NCRR) for strains. We also thank the two anonymous reviewers for constructive criticism and valuable experimental suggestions. This research was supported by grants from the Polycystic Kidney Disease Foundation (to J.W.) and NIH/National Institute of Diabetes and Digestive and Kidney Diseases (to M.M.B.).


  • Received September 2, 2010.
  • Accepted October 4, 2010.

Available freely online through the author-supported open access option.


View Abstract