Advertisement

Distinct IFT mechanisms contribute to the generation of ciliary structural diversity in C. elegans

Saikat Mukhopadhyay, Yun Lu, Hongmin Qin, Anne Lanjuin, Shai Shaham, Piali Sengupta

Author Affiliations

  1. Saikat Mukhopadhyay1,
  2. Yun Lu2,
  3. Hongmin Qin3,,
  4. Anne Lanjuin1,,
  5. Shai Shaham2 and
  6. Piali Sengupta*,1
  1. 1 Department of Biology and National Center for Behavioral Genomics, Brandeis University, Waltham, MA, USA
  2. 2 Laboratory of Developmental Genetics, The Rockefeller University, New York, NY, USA
  3. 3 Department of Molecular, Cellular and Developmental Biology, Yale University, New Haven, CT, USA
  1. *Corresponding author. Department of Biology and National Center for Behavioral Genomics, Brandeis University, 415 South Street, Waltham, MA 2454, USA. Tel.: +1 781 736 2686; Fax: +1 781 736 3107; E-mail: sengupta{at}brandeis.edu
  • Present address: Department of Biology, Texas A&M University, College Station, TX, USA

  • Present address: Department of Molecular and Cell Biology, Harvard University, Cambridge, MA 02138, USA

Abstract

Individual cell types can elaborate morphologically diverse cilia. Cilia are assembled via intraflagellar transport (IFT) of ciliary precursors; however, the mechanisms that generate ciliary diversity are unknown. Here, we examine IFT in the structurally distinct cilia of the ASH/ASI and the AWB chemosensory neurons in Caenorhabditis elegans, enabling us to compare IFT in specific cilia types. We show that unlike in the ASH/ASI cilia, the OSM‐3 kinesin moves independently of the kinesin‐II motor in the AWB cilia. Although OSM‐3 is essential to extend the distal segments of the ASH/ASI cilia, it is not required to build the AWB distal segments. Mutations in the fkh‐2 forkhead domain gene result in AWB‐specific defects in ciliary morphology, and FKH‐2 regulates kinesin‐II subunit gene expression specifically in AWB. Our results suggest that cell‐specific regulation of IFT contributes to the generation of ciliary diversity, and provide insights into the networks coupling the acquisition of ciliary specializations with other aspects of cell fate.

Introduction

Microtubule‐based ciliary structures play critical roles in mediating cellular functions such as motility, sensation of environmental cues and intercellular and intracellular signal transduction in both protozoa and metazoans (Cole, 2003; Satir and Christensen, 2007; Singla and Reiter, 2006; Wang et al, 2006). Intraflagellar transport (IFT) motors such as kinesin and dynein build and maintain cilia by transporting ciliary axoneme precursors and membrane proteins as part of large macromolecular IFT particles (Rosenbaum and Witman, 2002; Scholey, 2003). Remarkably, the molecular motors and core components of IFT particles are highly conserved from the unicellular green alga Chlamydomonas to humans (Rosenbaum and Witman, 2002; Scholey, 2003). Moreover, mutations in core ciliary genes have been shown to result in a plethora of diseases such as Bardet–Biedl syndrome and polycystic kidney disease (Pazour and Rosenbaum, 2002; Eley et al, 2005). Thus, it is important to further characterize IFT mechanisms and to identify regulators of ciliary structure and function.

Ciliary morphologies can be highly specialized and diverse. Vertebrate sensory neurons such as olfactory neurons have eight or more elongated cilia emanating from a dendritic knob, whereas the outer segments of rod and cone photoreceptors are unique ciliary structures required for efficient phototransduction (Reese, 1965; Rohlich, 1975). Despite the critical importance of these ciliary specializations for neuron‐specific sensory functions, the mechanisms required for the generation of these unique and diverse ciliary structures remain largely uncharacterized.

Caenorhabditis elegans is an excellent model system to investigate the mechanisms required for the generation of ciliary morphological diversity. In particular, 11 pairs of chemosensory neurons in the bilateral amphid organs of the head contain structurally diverse cilia. Eight of these sensory neuron types respond to aqueous attractants (Bargmann and Horvitz, 1991a, 1991b), and contain relatively simple cilia structures consisting of one or two branches extending from the transition zone (channel cilia; Figure 1A) (Ward et al, 1975; Ware et al, 1975; Perkins et al, 1986). The axonemes of these cilia are comprised of a middle segment containing nine doublet peripheral microtubules and a distal segment containing nine singlet microtubules (Ward et al, 1975; Perkins et al, 1986). Three additional neuron types, AWA, AWB and AWC, respond primarily to volatile odorants (Bargmann et al, 1993; Troemel et al, 1997; Chou et al, 2001) and contain cilia that are morphologically unique and more elaborate (wing cilia; Figure 1A) (Ward et al, 1975; Ware et al, 1975; Perkins et al, 1986). The cilia of the AWA olfactory neurons are highly branched, whereas the AWB olfactory neurons contain two ciliary branches of different lengths, the distal ends of which exhibit an irregular morphology and exhibit some animal‐to‐animal variability (Figure 1B) (Ward et al, 1975; Perkins et al, 1986). The cilia of the AWC neurons are fan shaped and extended (Ward et al, 1975; Perkins et al, 1986). Microtubule doublets and singlets are present throughout the length of the AWC cilium and these are not organized similarly to the axoneme of the channel cilia (Evans et al, 2006). The ultrastructures of the AWB and AWA cilia have not been reported in detail partly due to the complexity of their ciliary morphologies. Primary signal transduction proteins such as chemoreceptors and channels are localized to chemosensory neuron cilia (Troemel et al, 1995; Sengupta et al, 1996; Dwyer et al, 1998; Roayaie et al, 1998; Tobin et al, 2002), and animals with defective cilia exhibit highly compromised sensory responses (Perkins et al, 1986; Starich et al, 1995). The ability to identify and examine individual neuron types and their associated cilia provides a unique opportunity to define the developmental and cellular mechanisms required for the generation of ciliary structural diversity.

Figure 1.

Effect of mutations in IFT genes and IFT motors on AWB ciliary morphology. (A) Diagrammatic representation of channel and wing cilia morphologies (left) and their location in the C. elegans head (right). (Adapted from Ward et al, 1975). (B–G) AWB cilia visualized via expression of a str‐1p∷gfp transgene in wild‐type (panel B), osm‐5(p813) (panel C), che‐3(e1124) (panel D), osm‐3(p802) (panel E), kap‐1(ok676) (panel F) and kap‐1; osm‐3 (panel G) mutant animals. Zones corresponding to ultrastructurally defined segments (Figure 2) are indicated in panel B; M, middle segment; D, distal segment; FD, far‐distal segment. Arrowheads indicate accumulations at the cilia tips or base (panels C, D). Average lengths of the long and short ciliary branches were 8.0±1.3 and 6.2±1.3 μm (wild‐type; n=43), and 7.8±1.3 and 6.2±1.2 μm (osm‐3; n=50), respectively. (H) Expression of a functional osm‐3∷gfp fusion gene in an AWB neuron of an osm‐3 animal. AWB neurons were identified by filling with DiI. Images in panels B and E–G were acquired using a confocal microscope. Scale bars are as follows: 5 μm (panel A); 4 μm (panels B, E–G); 10 μm (panels C, D); 15 μm (panel H). Anterior is at left in all images.

Although genetic screens have led to the identification of many molecules required for ciliary assembly in C. elegans (reviewed in Rosenbaum and Witman, 2002; Scholey, 2003), the mechanisms required to generate ciliary morphological diversity in C. elegans remain unclear. IFT has largely been visualized and quantified in all cilia simultaneously, which obscures cilia‐to‐cilia differences in transport mechanisms. Thus, in the channel cilia, the heterotrimeric kinesin‐II and the homodimeric OSM‐3 motors have been shown to act cooperatively and redundantly to elongate the doublet microtubules of the middle segments, whereas OSM‐3 alone elongates the distal singlet microtubules (Snow et al, 2004; Ou et al, 2005; Pan et al, 2006). However, although both kinesin‐II and OSM‐3 are also required to build the entire AWC ciliary axoneme (Evans et al, 2006), no IFT was detected in the AWC neurons, and thus, it is unclear how these and other wing cilia are assembled. Little is also known about the developmental pathways that coordinate the generation of neuron‐specific ciliary morphology with additional aspects of cell fate.

Here, we compare and contrast ciliary morphogenesis specifically in the AWB wing, and the ASH/ASI channel cilia. We describe the ultrastructure of the AWB cilia and show that anterograde kinesin motors function in an unexpected and cell type‐specific manner to build the AWB cilia. We also identify the FKH‐2 forkhead domain transcription factor as being required for the generation of AWB‐specific ciliary structures and show that FKH‐2 regulates expression of a kinesin‐II motor subunit gene specifically in the AWB neurons. Our results provide insights into the molecules and mechanisms governing the generation of ciliary diversity and have implications on how similar specializations may be formed in other systems.

Results

Mutations in IFT particle and the dynein retrograde motor genes affect AWB and channel cilia in a similar manner

To begin to explore the roles of core ciliary genes in the formation of the AWB cilia, we first examined the AWB cilia in animals mutant for genes previously implicated in channel cilia formation. The AWB cilia were visualized by expressing gfp under the AWB‐specific str‐1 promoter (Troemel et al, 1997), which allows visualization of the soma, axonal and dendritic processes, as well as the branched AWB cilia (Figure 1B). IFT particles can be biochemically further subdivided into IFT‐A and IFT‐B complexes, each of which contains distinct proteins (Cole et al, 1998). Mutations in IFT‐A complex genes result in slightly shortened channel cilia, with accumulation at the distal tips, whereas mutations in IFT‐B complex genes result in severely shortened channel cilia, with accumulations at the base (Perkins et al, 1986; Cole et al, 1998). We found that mutations in the IFT‐A complex gene che‐11 and daf‐10, and the IFT‐B complex genes osm‐5 and osm‐1 (Lewis and Hodgkin, 1977; Perkins et al, 1986; Cole et al, 1998; Bell et al, 2006), resulted in AWB cilia phenotypes similar to those of channel cilia in these mutant backgrounds (Table I; Figure 1C). In addition, mutations in the dynein retrograde motor protein gene che‐3, and the dynein light intermediate chain gene xbx‐1, also resulted in AWB ciliary phenotypes that resembled those exhibited by channel cilia mutants in these mutants (Albert et al, 1981; Schafer et al, 2003) (Table I; Figure 1D). Thus, the examined IFT particle component gene products and dynein appear to act via similar mechanisms in channel and AWB cilia.

View this table:
Table 1. AWB ciliary defects in core ciliary gene and fkh‐2 mutant animals

Kinesin‐II and OSM‐3 act partly redundantly to build the AWB cilia

We next examined the roles of the anterograde heterotrimeric kinesin‐II and homodimeric OSM‐3 motors in the AWB cilia (Tabish et al, 1995; Orozco et al, 1999; Signor et al, 1999b; Snow et al, 2004). Consistent with previous observations, we did not observe defects in the AWB cilia in osm‐3 single mutants (Perkins et al, 1986) (Figure 1E; Table I). However, we observed minor defects in kap‐1 and klp‐11 kinesin‐II subunit single mutants. Approximately 10% of kap‐1 or klp‐11 single mutants did not exhibit the characteristic forked structures, but instead had truncated ciliary branches, with occasional ectopic projections (Figure 1F; Table I). The phenotype was not further enhanced in kap‐1; klp‐11 double mutants (Table I). AWB cilia branches were severely truncated in kap‐1; osm‐3 double mutants (Figure 1G; Table I), indicating a hitherto unidentified role for OSM‐3 in the assembly of these cilia. Although previous reports suggested that OSM‐3 was not expressed in the AWB cilia (Tabish et al, 1995), we confirmed OSM‐3 expression in the AWB neurons by examining expression of a functional GFP‐tagged OSM‐3 protein (Supplementary Table 1) driven under its own promoter (Figure 1H). These results indicate that kinesin‐II and OSM‐3 function partly redundantly to build the AWB ciliary axoneme.

Ultrastructure of the AWB olfactory neurons

We examined the ultrastructure of the AWB ciliary axoneme in wild‐type and kinesin motor mutant backgrounds using electron microscopy. Unlike the axonemes of either the AWC or channel cilia (Perkins et al, 1986; Evans et al, 2006), the AWB axoneme in wild‐type animals could be subdivided into three segments based on microtubule organization (Figure 2A). In a short middle segment (∼2 μm), we noted the presence of singlet microtubules and occasional doublets. In contrast to the 9+0 arrangement of doublet microtubules in the channel cilia middle segments (Perkins et al, 1986), the microtubules in the AWB middle segment lacked an obvious organization. In a more elongated distal segment (∼5 μm), we noted only singlet microtubules, similar to channel cilia distal segments. Finally, in the far‐distal segment (∼1 μm), we did not detect any microtubules, suggesting that this region, which exhibits the most animal‐to‐animal variability, may lack an axoneme. No gross alterations in ciliary ultrastructure were observed in either kap‐1; klp‐11 double or in osm‐3 single mutant animals (Figure 2B and C). Thus, while the distal segments of channel cilia are absent in osm‐3 mutants (Perkins et al, 1986), formation of the distal segments of the AWB cilia does not require OSM‐3.

Figure 2.

Ultrastructure of the AWB cilia in wild‐type (A), kap‐1(ok676); klp‐11(tm324) (B) and osm‐3(p802) (C) mutant animals. The cilium is outlined by a dashed line. Images were acquired at 60,000‐fold magnification. Lengths of the middle, distal and far‐distal segments were approximately 2, 5 and 1 μm, respectively. Arrowheads indicate singlet microtubules, closed and open arrows indicate doublet microtubules in the AWB, and channel cilia, respectively. Note the absence of an axoneme in the far‐distal segment. Scale bars, 0.2 μm.

Kinesin motors function in a cell‐specific manner to build the AWB and ASH/ASI cilia

Kinesin‐II and OSM‐3 act together to build the middle segments of channel cilia (Snow et al, 2004; Ou et al, 2005). These motors are thought to compete with each other mechanically, resulting in an intermediate velocity of both motors in the middle segment (Snow et al, 2004; Pan et al, 2006). However, transport velocity in the distal segments is rapid due to action of the fast OSM‐3 motor alone. Thus, in kinesin‐II mutants, IFT velocity is accelerated in the middle segments due to the action of OSM‐3, whereas in osm‐3 mutants, particles are moved at the rate of the slow kinesin‐II motor movement in the middle segment (channel cilia distal segments are not formed in the absence of osm‐3) (Snow et al, 2004).

To directly compare the function of kinesin motors in the channel and the AWB wing cilia, we visualized and quantified the movement of GFP‐tagged KAP‐1 and OSM‐3 proteins specifically in the AWB and ASH/ASI neurons using the sra‐6 and str‐1 promoters (Troemel et al, 1995, 1997). We confirmed that both OSM‐3∷GFP and KAP‐1∷GFP fusion proteins were functional by rescuing the ciliary defects in the AWB neurons in osm‐3; kap‐1 double mutants (Supplementary Table 1). In the ASH/ASI channel, as well as in the AWB wing cilia middle segments, KAP‐1∷GFP moved at an intermediate velocity (Figure 3A and B; Table II; Supplementary Movie S1). This velocity was reduced in osm‐3 mutants in both cilia types, albeit less markedly in the AWB middle segments (Figure 3C and D; Table II; Supplementary Movie S2), suggesting, as described previously, that kinesin‐II and OSM‐3 likely act together in the middle segments of both the ASH/ASI channel and AWB wing cilia.

Figure 3.

Movement of GFP‐tagged KAP‐1 in the ASH/ASI (A, C) and AWB (B, D) cilia. For each panel, still images (left) and kymographs (right) from representative movies of KAP‐1∷GFP transport are shown. Histograms of KAP‐1∷GFP velocities are shown below in each panel. For each still image, the genetic background is indicated in the top left corner, the imaged cilia in the top right corner and the fusion protein examined in the bottom left corner. Dashed lines on the still images indicate the continuous lines used to generate the kymographs. The osm‐3(p802) allele was used. Scale bars in still images are 5 μm. Scale bars in kymographs are as follows: horizontal bar, 5 μm; vertical bar, 5 s. M, middle segment; D, distal segment. No ASH/ASI distal segments are formed in osm‐3 mutants. Statistical analyses are shown in Supplementary Tables 2 and 3.

View this table:
Table 2. Anterograde velocities of IFT particles and IFT motors in ASH/ASI and AWB cilia

As expected, in the ASH/ASI cilia, OSM‐3∷GFP moved at a slower intermediate velocity in the middle segments similar to that of KAP‐1∷GFP, and at a higher velocity in the distal segments (Figure 4A; Table II). Surprisingly however, OSM‐3∷GFP moved at a significantly higher average velocity than KAP‐1∷GFP in the middle segments of the AWB cilia (Figure 4B; Table II; Supplementary Movie S3), suggesting that a subset of OSM‐3∷GFP molecules moves independently of kinesin‐II. Consistent with this hypothesis, the velocity of OSM‐3∷GFP was increased by a lesser extent in kinesin‐II mutants in the AWB cilia middle segments as compared with ASH/ASI cilia middle segments (Figure 4C and D; Table II; Supplementary Movie S4). Moreover, in striking contrast to the observations in ASH/ASI cilia, movement of OSM‐3∷GFP appeared to be largely restricted to the AWB middle segments, such that OSM‐3 movement could be detected rarely in the distal segments of the AWB cilia in either wild‐type or kinesin‐II mutants (Figure 4B and D; Supplementary Movies S3 and S4). Taken together, these results suggest that unlike in the channel cilia, OSM‐3∷GFP moves partly independently of kinesin‐II to build the AWB cilia middle segments, and that transport via OSM‐3 may not be essential to build or maintain the AWB distal segments.

Figure 4.

Movement of GFP‐tagged OSM‐3 in the ASH/ASI (A, C) and AWB (B, D) cilia. For each panel, still images (left) and kymographs (right) from representative movies of OSM‐3∷GFP transport are shown. Histograms of OSM‐3∷GFP velocities are shown below in each panel. Velocities in the middle (M) and distal segments (D) are indicated by black and gray bars, respectively, in each histogram. Panels are labeled as in Figure 3. The kap‐1(ok676) allele was used. Scale bars in still images are 5 μm. Scale bars in kymographs are as follows: horizontal bar, 5 μm; vertical bar, 5 s. Statistical analyses are shown in Supplementary Tables 2 and 3.

IFT mechanisms are distinct in the AWB wing and the ASH/ASI channel cilia

We next compared the movement of IFT particles in the ASH/ASI and AWB cilia by expressing the gfp‐tagged IFT particle A and B genes daf‐10 and osm‐6 (Collet et al, 1998; Bell et al, 2006) under the sra‐6 and str‐1 promoters. We confirmed that both OSM‐6∷GFP and DAF‐10 fusion proteins were functional by rescuing the ciliary defects in the AWB neurons in osm‐6 and daf‐10 mutants (Supplementary Table 1).

OSM‐6∷GFP and DAF‐10∷GFP moved robustly in both the middle and distal segments of the ASH/ASI channel cilia (Figure 5A; Supplementary Figure 1; Supplementary Movie S5) (Orozco et al, 1999; Signor et al, 1999a; Snow et al, 2004; Ou et al, 2005). The velocities of transport of both fusion proteins in the ASH/ASI cilia in wild‐type and motor mutant backgrounds were consistent with these particles being transported by both kinesin‐II and OSM‐3 in the channel cilia middle segments, and by OSM‐3 alone in the distal segments (Figure 5A, C and E; Table II; Supplementary Figure 1; Supplementary Movie S6).

Figure 5.

Movement of GFP‐tagged OSM‐6 in the ASH/ASI (A, C, E) and AWB (B, D, F) cilia. For each panel, still images (left) and kymographs (right) from representative movies of OSM‐6∷GFP transport are shown. Histograms of OSM‐6∷GFP velocities are shown below in each panel. Panels are labeled as in Figure 3. Velocities in the middle (M) and distal (D) segments are indicated by black and gray bars, respectively, in each histogram. Scale bars in still images are 5 μm. Scale bars in kymographs are as follows: horizontal bar, 5 μm; vertical bar, 5 s. M, middle segment; D, distal segment. No ASH/ASI distal segments are formed in osm‐3 mutants. Statistical analyses are shown in Supplementary Tables 2 and 3.

We observed a distinct pattern of OSM‐6∷GFP movement in the AWB wing cilia. Although we observed robust OSM‐6∷GFP movement in the middle segments, we observed far fewer IFT events in the distal segments (Figure 5B; Table II; Supplementary Movie S7). DAF‐10∷GFP movement was also observed robustly in the middle segment but not in the distal segments (Supplementary Figure 1). Similar results were obtained with an OSM‐5∷GFP fusion protein, although movement of the OSM‐5 fusion protein was more difficult to image due to low levels of fluorescence (Supplementary Figure 1; Table II). In the middle segments, the velocities of IFT particle fusion proteins were similar to that of kinesin‐II in either wild‐type or osm‐3 mutants (Table II; Figure 5B and D; Supplementary Movie S8), although we also detected a small fraction of particles moving at the higher velocity of OSM‐3 alone (see histograms in Figure 5B; Supplementary Figure 1). The velocities were increased in the absence of kinesin‐II (Figure 5F; Table II; Supplementary Movie S9). These results suggest that as in channel cilia middle segments, the majority of IFT particles are transported by kinesin‐II and OSM‐3 acting together in the AWB cilia middle segments, whereas in the absence of kinesin‐II, these molecules are transported by OSM‐3 alone. OSM‐3 may also mediate IFT independently of kinesin‐II.

Since it was surprising that few IFT events were observed in the AWB cilia distal segments, we examined this transport in more detail. Although we could image only a small number of particles due to the relative rarity of IFT events in the AWB cilia distal segments in wild‐type animals, the velocity of OSM‐6∷GFP movement in the distal segments suggested that these particles were transported by OSM‐3 (Figure 5B; Table II). Consistent with this model, we could not detect IFT events in the distal segments in osm‐3 mutants, although these segments were present (Figure 5D; Supplementary Figure 1; Supplementary Movie S8). However, in kinesin‐II mutants, we observed robust movement of OSM‐6 in the AWB cilia distal segments (Table II; Figure 5F; Supplementary Movie S9). Thus, IFT does not appear to be essential to build and/or maintain the AWB distal segments in osm‐3 mutants, despite the presence of an apparently unaltered axoneme.

Mutations in the fkh‐2 forkhead domain transcription factor gene affect AWB ciliary and dendritic morphology

We next sought to identify molecules that regulate AWB cilia structure. Mutations in transcription factors such as cut and hamlet have been shown to alter neuron‐specific dendritic morphology in Drosophila, without affecting additional aspects of cell fate specification (Moore et al, 2002; Grueber et al, 2003). We reasoned that a transcription factor may act similarly in the AWB neurons to regulate the expression of genes required to generate AWB‐specific ciliary morphology. The fkh‐2 forkhead domain transcription factor gene was previously shown to be broadly expressed in the embryo (Molin et al, 1999). However, we noted that postembryonic expression of a functional gfp‐tagged fkh‐2 gene was restricted to the nuclei of the AWB neurons, with less consistent expression in additional chemosensory neurons (Figure 6A). fkh‐2 mutant animals were previously shown to not exhibit obvious phenotypic abnormalities, although animals doubly mutant for fkh‐2 and a second forkhead domain gene pes‐1 were embryonic lethal (Molin et al, 1999). Given the restricted postembryonic expression pattern of fkh‐2, we further examined the role of fkh‐2 in the determination of AWB cell fate and function.

Figure 6.

Mutations in fkh‐2 result in dendritic extension and ciliary defects in the AWB neurons. (A) A rescuing fkh‐2∷gfp fusion gene is expressed postembryonically in an AWB neuron (arrow) of a wild‐type animal. AWB was identified by filling with the lipophilic dye DiD. The genomic organization of fkh‐2 with the location of inserted gfp‐encoding sequences is shown at top. Arrows indicate the two predicted transcription initiation sites. Sequences encoding the forkhead domain are shaded. A dashed line indicates the extent of the ok683 deletion. (B) Shown is the cell body and dendrite of an AWB neuron of a wild‐type (upper panel) and fkh‐2 mutant (lower panel) adult animal expressing str‐1p∷gfp. Arrowheads indicate the dendrite ends. (C) Confocal images of AWB cilia from fkh‐2 mutant animals. Observed ciliary defects include truncated cilia (upper panel) or loss of a ciliary branch (lower panel). (D) Localization of a GFP‐tagged ODR‐10 protein to the AWB neuron cilia in wild‐type (upper panel) or fkh‐2 mutants (lower panel). AWB neurons are visualized via expression of an srd‐23p∷dsRed fusion gene. (E) AWB neurons (arrow) fill with DiI in a wild‐type pattern (upper panel), but fail to fill in a fraction of fkh‐2 mutants (lower panel). Scale bars are as follows: 15 μm (panels A, B, D, E); 4 μm (panel C). Anterior is at left in all images.

The fkh‐2 locus is predicted to encode two alternatively spliced transcripts (www.wormbase.org). Both transcripts are predicted to be affected in animals carrying the ok683 deletion allele (provided by the C. elegans Gene Knockout Consortium) (Figure 6A), suggesting that ok683 is a null allele. We first determined whether overall AWB cell fate specification was altered in fkh‐2 mutants by examining the expression of AWB‐specific markers. fkh‐2 mutants exhibited defects in the expression of two marker∷gfp fusion genes (Table III). These defects were markedly enhanced in animals doubly mutant for fkh‐2 and the ceh‐37 Otx homeobox gene, which has been previously implicated in regulating AWB fate specification (Lanjuin et al, 2003) (Table III). Thus, FKH‐2 likely acts in parallel to CEH‐37 to regulate AWB cell fate.

View this table:
Table 3. fkh‐2 regulates AWB cell fate specification

We next determined whether mutations in FKH‐2 alter AWB neuronal morphology. The dendrites of 14% of examined AWB neurons terminated prematurely in fkh‐2(ok683) animals (Figure 6B). Dendrites terminated at different locations along the heads of animals and did not possess the characteristic ciliary structures (Figure 6B). These defects were observed as early as the three‐fold embryonic stage. An additional ∼40% of AWB neurons did not exhibit obvious defects in dendritic extension, but instead exhibited a range of ciliary abnormalities including shortened forks, or lack of one or both forked ciliary structures (Figure 6C; Table I). Ciliary proteins such as GFP‐tagged ODR‐10 and STR‐1 chemoreceptors (Sengupta et al, 1996; Troemel et al, 1997) retained ciliary localization in a subset of AWB neurons exhibiting altered ciliary morphology (Figure 6D and data not shown), although no localization was observed to the ends of truncated AWB dendrites. Six pairs of amphid sensory neurons, including five channel cilia neurons and the AWB neurons, take up lipophilic dyes such as DiI (Perkins et al, 1986; Herman and Hedgecock, 1990) (Figure 6E). Dye uptake is severely compromised in animals with defects in dendritic extension and ciliary morphogenesis (Perkins et al, 1986). Consistent with the observed dendritic and ciliary extension defects, 36% of the AWB, but not other neuron types, failed to fill with DiI in fkh‐2 mutants (Figure 6E). The dendritic and ciliary defects were rescued by a genomic fragment containing fkh‐2 coding and regulatory sequences (Table I). These results indicate that mutations in fkh‐2 regulate both AWB dendritic extension as well as AWB ciliary morphogenesis.

FKH‐2 may act in the AWB neurons, or elsewhere, to regulate AWB morphology. To investigate the site of FKH‐2 action, we correlated the presence or absence of fkh‐2 genomic rescuing sequences in the AWB lineage with the ability of the AWB neurons to dye‐fill. The ability or failure of the AWB neurons to dye‐fill correlated more highly with the presence or absence, respectively, of fkh‐2 genomic rescuing sequences in the AWB and closely related lineages than in distant lineal relatives (Sulston et al, 1983), suggesting that fkh‐2 acts in the AWB lineage to regulate AWB neuron morphology (see Supplementary data).

IFT is altered in the AWB cilia in fkh‐2 mutants

To determine whether alterations in IFT underlie the ciliary defects observed in fkh‐2 mutants, we examined OSM‐6∷GFP movement in the AWB cilia of fkh‐2 mutants. Although we could not confidently image movement in severely truncated cilia, we found that in AWB cilia that could be imaged, OSM‐6∷GFP was transported robustly throughout the length of the cilia, whereas OSM‐3∷GFP movement was restricted to the AWB middle segments, similar to our observations in kap‐1 mutants (Figure 7A; Table II; Supplementary Movie S10). This result suggests that FKH‐2 may act in part by altering kap‐1 gene expression. However, the velocity of OSM‐6∷GFP movement in the AWB middle segments in fkh‐2 mutants was significantly different from that in kap‐1 mutant animals (Table II), suggesting that kap‐1 expression may not be fully lost in fkh‐2 mutants.

Figure 7.

FKH‐2 regulates kap‐1 expression in the AWB neurons. (A) Still image (left panel) and kymograph (right panel) from a representative movie of OSM‐6∷GFP in an AWB cilia of a fkh‐2(ok683) animal. Histograms of OSM‐6∷GFP velocities are shown below. Dashed lines on the still images indicate the continuous lines used to generate the kymographs. Velocities in the middle (M) and distal (D) segments are indicated in black and gray bars, respectively. Scale bar in the still image is 5 μm. Scale bars in kymographs are as follows: horizontal bar, 5 μm; vertical bar, 5 s. (B) Shown are the percentages of the indicated chemosensory neuron types expressing kap‐1p∷dsRed in wild‐type or fkh‐2(ok683) animals. Numbers shown are from two independent transgenic lines expressing kap‐1p∷dsRed from extrachromosomal arrays. Asterisk indicates values that are different from WT at P<0.001 using a chi‐square test. (C) A fkh‐2∷gfp (IeIs916) fusion construct is expressed in the AWB (arrowheads) and additional neurons in wild‐type animals (left panel) but not in daf‐19(m86) mutants (right panel). A total of 100% of daf‐19 mutants failed to express fkh‐2∷gfp (n=119) compared with 0% of wild‐type animals (n=31) in the L1 developmental stage. Scale bars, 15 μm. Anterior is to the left in both images.

To test this hypothesis, we examined the expression of kap‐1p∷dsRed in fkh‐2(ok683) mutants. We found that kap‐1 expression was significantly downregulated but not abolished in the AWB neurons of fkh‐2 mutants relative to wild‐type animals (Figure 7B). Expression in two additional neuron types (ASH and ASJ) was unaltered. We next correlated downregulation of kap‐1 expression with defects in ciliary and dendritic structures. A total of 87% of animals exhibiting dendritic and/or ciliary defects also exhibited loss of kap‐1p∷dsRed expression in the AWB neurons (n=126). However, we also observed wild‐type morphology in AWB neurons lacking kap‐1 expression. fkh‐2; kap‐1 double mutants exhibited cilia and dendrite defects similar to those of fkh‐2 mutants alone (data not shown). Taken together with the observation that kap‐1 mutants exhibit weaker AWB morphological defects than fkh‐2 mutants, these results suggest that kap‐1 represents one of a larger set of FKH‐2 target genes that regulate AWB dendritic and ciliary morphology.

The RFX transcription factor DAF‐19 regulates fkh‐2 expression postembryonically in the AWB neurons

The RFX transcription factor DAF‐19 has been shown to regulate ciliogenesis such that all ciliated neurons lack cilia in daf‐19 mutants (Perkins et al, 1986; Swoboda et al, 2000). We examined the regulatory relationship, if any, between DAF‐19 and FKH‐2. We observed a dramatic downregulation of postembryonic fkh‐2 expression in daf‐19 mutants (Figure 7C), although embryonic expression appeared grossly unaltered. DAF‐19 regulates the transcription of target genes via binding a conserved X box sequence (Swoboda et al, 2000). We identified a low‐probability X box element in the promoter of fkh‐2, which is also conserved in the fkh‐2 ortholog in the related nematode Caenorhabditis briggsae. However, mutating this sequence did not abolish fkh‐2 expression (data not shown), suggesting that DAF‐19 binds alternate sites or acts via intermediate molecules to regulate fkh‐2 expression. The AWB cilia of daf‐19; fkh‐2 mutant animals exhibited defects similar to those of daf‐19 mutants alone (Table I). However, unlike fkh‐2 mutants, we did not observe defects in AWB dendritic extension in daf‐19 mutants, and daf‐19; fkh‐2 double mutants had dendritic defects similar to fkh‐2 mutants (Table I). These results suggest that DAF‐19‐mediated regulation of FKH‐2 expression plays a role in AWB ciliogenesis, but that DAF‐19‐independent expression of FKH‐2 may regulate AWB dendritic morphology.

Discussion

Deployment of IFT mechanisms in a cell‐specific manner may generate ciliary structural diversity

Direct visualization of IFT and motor movement in cilia in vivo has allowed the identification and description of the dynamics of cilia formation and maintenance (Rosenbaum and Witman, 2002; Scholey, 2003). However, in C. elegans, IFT has previously been examined in multiple cilia at once (Orozco et al, 1999), precluding the identification of cell‐to‐cell differences in IFT mechanisms. Here, we examine and quantify IFT specifically in two cilia types, by driving expression of motor and IFT particle genes specifically in the ASH/ASI channel, and in the AWB wing cilia using cell‐specific promoters. This approach has allowed us, for the first time, to compare and contrast IFT mechanisms in cilia of two different morphologies, and to uncover cell‐specific mechanisms that may contribute to the generation of ciliary morphological diversity.

The ultrastructures of the channel cilia and the AWB wing cilia are grossly similar in that the middle segments contain doublet microtubules and the distal segments contain singlets (Perkins et al, 1986; this work). However, there are also differences in the axoneme structures of these cilia. Unlike in the channel cilia, the middle segment of the AWB cilium is relatively short and contains rare doublets as well as singlet microtubules that are not organized in the typical cylindrical structure. In addition, distal segments of the AWB cilium are elongated as compared with the distal segments of the channel cilia. Finally, the far‐distal region of the AWB cilium appears to lack a central axoneme, raising the possibility that this structure is simply a membraneous extension. The AWC ciliary axoneme was previously shown to contain both doublet and singlet microtubules throughout the length of the cilium (Evans et al, 2006). Thus, differences in axoneme structure underlie the observed morphological diversity in C. elegans sensory neuron cilia.

Modulation of IFT and motor protein function in a cell‐specific manner likely contributes to the differences in axoneme structure, and hence, morphological diversity. Previous studies have suggested that the conserved heterotrimeric kinesin‐II is the core ciliary anterograde motor, whereas the homodimeric OSM‐3 kinesin acts as an ‘accessory’ motor that can act with or substitute for kinesin‐2, and also elongates distal segments containing singlet microtubules (Snow et al, 2004; Evans et al, 2006). However, OSM‐3 appears to be deployed in a distinct manner in the AWB cilia. Although a subset of OSM‐3 molecules acts with kinesin‐II in the AWB cilia middle segments to mediate IFT, unlike in the channel cilia, OSM‐3 also appears to act in a kinesin‐II‐independent manner. Components of both IFT particle A and B complexes appear to be transported largely by kinesin‐II/OSM‐3 acting together, although a small subset may be transported by OSM‐3 alone (Figure 8A). OSM‐3 may also transport a partly distinct set of cargo proteins in the AWB cilia. The mammalian homolog of OSM‐3 is required for ciliary targeting of olfactory cyclic nucleotide‐gated channel subunits in mammalian cells (Jenkins et al, 2006), suggesting that OSM‐3 may be responsible for ciliary targeting of cell‐specific proteins. We note that OSM‐3 cannot fully substitute for loss of kinesin‐II function, such that a small fraction of kap‐1/klp‐11 mutants continue to exhibit severely truncated AWB cilia. This observation further supports the hypothesis that OSM‐3 acts as an accessory motor.

Figure 8.

Cell‐specific modulation of IFT may generate ciliary diversity. (A) Models comparing IFT in the channel and AWB cilia. Number of arrows are directly proportional to measured velocity. Note that distal segments are absent in channel but not in AWB cilia in osm‐3 mutants. In the AWB middle segments, OSM‐3 is proposed to function partly independently of kinesin‐II. We detected IFT events and motor movement in the AWB distal segments rarely in wild‐type or osm‐3 mutant cilia, suggesting that the majority of motor proteins and IFT particles may undergo turnaround (represented by a curved arrow) at the middle/distal segment border in AWB cilia. See text for additional details. (B) Model of FKH‐2 function in fate specification of the AWB neurons. FKH‐2 may link a ciliogenic module (left box) with an AWB‐specific developmental module (right box) to generate AWB‐specific morphology and AWB fate. Dashed lines indicate postulated regulatory relationships.

A particularly striking difference in the deployment of OSM‐3 in the AWB and channel cilia is the requirement of this motor in forming the distal ciliary segments. Although both the channel cilia and AWB cilia distal segments contain singlet microtubules, OSM‐3 is required to form these segments in the channel but not in the AWB cilia. Thus, in osm‐3 mutants, the distal segments of the channel cilia are absent (Perkins et al, 1986), whereas the distal segments of the AWB cilia are unaffected. Moreover, unlike in the channel cilia, we did not detect robust movement of either OSM‐3 or IFT particle proteins in the AWB cilia distal segments. This observation further corroborates the hypothesis that OSM‐3 may not be essential to form the AWB ciliary distal segments.

We envision several mutually non‐exclusive possibilities to account for our observations. The majority of motor molecules and IFT particles may undergo turnaround at the border between the middle and distal segments in the AWB cilia and are transported retrogradely to the ciliary base (Figure 8A), much like the case at the flagellar tip in Chlamydomonas (Sloboda, 2005; Pedersen et al, 2006). This would result in a decrease in the number of IFT particles and OSM‐3 motors in the distal segments. The IFT particle composition and cargoes may also undergo drastic alterations at the middle/distal segment border, such that the IFT components OSM‐5, OSM‐6 and DAF‐10 are largely excluded from the distal segments. It is also possible that the AWB distal segments are elongated in an IFT‐independent manner, or that IFT is required transiently early in ciliary assembly to generate the distal segments, but is not essential at other stages to maintain ciliary structure.

A puzzling observation was the robust transport of OSM‐6GFP in AWB cilia distal segments in kinesin‐II mutants, in the absence of corresponding robust movement of the OSM‐3 motor in this region. The IFT protein particle composition may be altered in kinesin‐II mutants, such that more IFT particle proteins are transported per OSM‐3 motor molecule in the distal segments, allowing us to visualize robust movement. It is also possible that this altered pattern may reflect the use of an alternate motor protein in the distal segments in a kinesin‐II mutant background. Proteins regulating kinesin motor functions have been identified in C. elegans (Blacque et al, 2004; Ou et al, 2005). Cell‐specific regulation of motor protein function is likely to underlie these cell‐specific differences in kinesin‐II and OSM‐3 function, and hence, ciliary structures.

The FKH‐2 transcription factor couples AWB cell fate specification with the acquisition of cell‐specific ciliary morphology

Our results suggest that the core ciliogenic program is likely to be modulated in a cell‐specific manner, so as to promote the generation of unique ciliary structures. In the AWB neurons, we have identified the FKH‐2 forkhead domain transcription factor as being an important link between the developmental programs that specify cell fate with the ciliogenic program that is common to all ciliated neurons (Figure 8B). FKH‐2 acts synergistically with the OTX homeodomain protein CEH‐37 (Lanjuin et al, 2003) to specify AWB cell fate, but in addition, regulates the expression of broadly expressed ciliary genes such as kap‐1 in an AWB‐specific manner. It is unlikely that kap‐1 is the sole target of FKH‐2 in the regulation of ciliogenesis in AWB, since downregulation of kap‐1 expression is not sufficient to account for all fkh‐2 ciliary phenotypes. We propose that FKH‐2 regulates additional genes in an AWB‐specific manner, a subset of which modulates IFT mechanisms. fkh‐2 expression, is in turn regulated by the DAF‐19 RFX transcription factor (Figure 8B). A similar developmental network has been described for the male‐specific HOB neurons in C. elegans, where an as yet unidentified factor has been proposed to couple the DAF‐19‐regulated ciliogenic program, with other aspects of HOB cell fate regulated by the transcription factors EGL‐44 and EGL‐46 (Yu et al, 2003). It is interesting to note that although fkh‐2 mutants also exhibit AWB dendritic morphological abnormalities, the functions of FKH‐2 in regulating dendritic structure are independent of DAF‐19. Thus, FKH‐2 acts in both DAF‐19‐dependent and independent pathways to regulate AWB morphology.

In mice, the forkhead domain transcription factor HFH‐4/FOXJ1 has been implicated in the generation of a subclass of motile cilia. hfh‐4 is expressed in epithelial cells in the embryonic node, which contain motile monocilia required for the generation of left–right asymmetry (Nonaka et al, 1998; Marszalek et al, 1999; Takeda et al, 1999), and also in ciliated epithelial cells in the airway and other tissues during ciliogenesis (Hackett et al, 1995; Lim et al, 1997; Blatt et al, 1999; Tichelaar et al, 1999). hfh‐4‐null mice lack motile airway cilia due to defective migration or anchoring of the basal bodies, and exhibit left–right asymmetry defects due to dysfunction of node cilia, perhaps in part via regulation of dynein (Chen et al, 1998; Brody et al, 2000; Gomperts et al, 2004). Thus, similar to FKH‐2, HFH‐4 also regulates the generation and function of specific ciliary structures. It remains to be determined whether other forkhead domain proteins play roles in similar aspects of cell‐specific ciliogenesis.

Diversity of ciliary structure ensures maximal optimization of cell functions. The precise topographic arrangement and compartmentalization of cell‐specific sensory receptors and signaling molecules in the cilia are essential for accurate and sensitive detection and signal transduction of external cues (Young and Droz, 1968; Kulaga et al, 2004; Nishimura et al, 2004; Corbit et al, 2005; Haycraft et al, 2005; Peden and Barr, 2005; Qin et al, 2005; Flannery et al, 2006; Wang et al, 2006). Results described here provide insights into the mechanisms by which the appropriate ciliary specializations may be generated via cell‐specific modulation of core ciliary proteins, and describe how the acquisition of ciliary structures may be coupled with additional aspects of cell fate specification.

Materials and methods

Nematode strains

Strains were obtained from the Caenorhabditis Genetics Center or the National Bioresource Project (Japan). fkh‐2(ok683) was outcrossed eight times before analysis and the extent of the deletion was verified by amplification. Stably integrated strains used in this work were: kyIs104 (str‐1p∷gfp) (Troemel et al, 1997), kyIs156 (str‐1p∷odr‐10cDNA∷gfp), oyIs44 (odr‐1p∷dsRed) (Lanjuin et al, 2003) and IeIs916 (fkh‐2∷gfp) (Molin et al, 2000; gift from I Hope). Double mutants were constructed using standard methods and the presence of individual mutations was verified by complementation and/or amplification, and sequencing of genomic sequences.

Microscopy

Images were acquired on a Zeiss Axioplan microscope using a × 40 (1.30 NA) or × 63 (1.25 NA) objective at room temperature and a CCD camera (Hamamatsu), and analyzed using Openlab 4.0 software (Improvision). Confocal images were acquired using a × 100 (1.4 NA) objective on a Leica spectral confocal microscope, and analyzed using Leica imaging software.

For quantification of IFT particle and motor movement, transgenic animals were anesthetized with 10 mM levamisole and mounted on agarose pads before imaging. Movies were obtained with a Nikon Eclipse TE2000 microscope equipped with a × 100 (1.4 NA) objective and a forced air‐cooled Photometrics Cascade 512B with a CCD87 CCD. All images for IFT particle and motor movement were collected at 200 ms frame−1 at room temperature. Kymographs and movies were generated using the Metamorph software package (Molecular Devices). Stacks of images for movies of movement in the AWB cilia shown in Figure 4 were further processed by unsharp masking, low‐pass filtering and edge detecting, and presented after autoscaling using Metamorph software.

Molecular biology and generation of transgenic strains

str‐1 or sra‐6 promoter driven gfp‐tagged osm‐6, osm‐5, daf‐10, osm‐3 (M02B7.3b), kap‐1 and fkh‐2 cDNA constructs were generated by fusing ∼3 kb of str‐1 or sra‐6 upstream promoter sequences with the respective gfp‐tagged cDNAs in a modified C. elegans expression vector. The srd‐23p∷dsRed plasmid was generated by fusing ∼3 kb of srd‐23 upstream promoter sequences to dsRed and gfp‐tagged fkh‐2‐gfp constructs were generated by cloning fkh‐2 promoter, coding and 3′‐UTR genomic sequences into C. elegans gfp expression vectors. kap‐1p∷dsRed was generated by PCR fusion (Hobert, 2002) of dsRed coding sequences with ∼212 bp of kap‐1 upstream sequences. A functional gfp‐tagged osm‐3 gene was generated by fusing gfp sequences in frame to osm‐3 genomic sequences including 1.9 kb upstream promoter sequences (Hobert, 2002). Plasmid constructs were verified by sequencing. C. elegans expression vectors were gifts from A Fire and were modified by M Colosimo. Transgenic strains were generated using unc‐122∷dsRed as the co‐injection marker.

Statistical analyses

Statistical analyses of IFT particle and motor protein velocities were performed using one‐way ANOVA and Scheffe's post hoc multiple comparison tests between all possible pairs in each data set using the SPSS 14 statistical analyses package.

Electron microscopy

Adult hermaphrodites were fixed in 3% gluteraldehyde on ice for 2 h, followed by a post fix in 1% OsO4 for 1 h at room temperature. Animals were then processed for electron microscopy as previously described (Bargmann et al, 1993). Sections measuring 50 nm were serially collected from the tip of the nose.

Supplementary data

Supplementary data are available at The EMBO Journal Online (http://www.embojournal.org).

Supplementary Information

Supplementary Data [emboj7601717-sup-0001.doc]

Supplementary Movie S1 [emboj7601717-sup-0002.mov]

Supplementary Movie S2 [emboj7601717-sup-0003.mov]

Supplementary Movie S3 [emboj7601717-sup-0004.mov]

Supplementary Movie S4 [emboj7601717-sup-0005.mov]

Supplementary Movie S5 [emboj7601717-sup-0006.mov]

Supplementary Movie S6 [emboj7601717-sup-0007.mov]

Supplementary Movie S7 [emboj7601717-sup-0008.mov]

Supplementary Movie S8 [emboj7601717-sup-0009.mov]

Supplementary Movie S9 [emboj7601717-sup-0010.mov]

Supplementary Movie S10 [emboj7601717-sup-0011.mov]

Supplementary Figure 1 [emboj7601717-sup-0012.pdf]

Acknowledgements

We are indebted to Joel Rosenbaum for his insights and enthusiasm, and for hosting SM in his laboratory. We are grateful to Ian Hope for directing our attention to the fkh‐2 expression pattern, Mayumi Shibuya and Caron Gauthier for technical assistance, David Hall and Cori Bargmann for interpretation of micrographs, Don Katz for help with statistics, Guangshuo Ou, Jonathan Scholey and the Sengupta laboratory for discussions, and Joel Rosenbaum and Paul Garrity for comments on the manuscript. We thank the Caenorhabditis Genetics Center, C. elegans Knockout Consortium and Shohei Mitani at the National Bioresource Project (Japan) for strains. This work was supported by the NIH (GM56223 – PS; S10RR16708 and P30NS45713) and the Klingenstein Fund for Neuroscience (SS).

References