Sensory signaling-dependent remodeling of olfactory cilia architecture in C. elegans

Mutations in chemosensory signal transduction molecules alter AWB olfactory neuron cilia structure

To determine whether signaling genes known to be required for chemosensory signal transduction play a role in regulating cilia structure, we examined cilia in animals mutant for each of these molecules. Mutations in the odr-3 Gα_i/o subunit gene required for chemosensory signal transduction were previously shown to affect cilia structures of the AWA and AWC olfactory neurons (Roayaie et al., 1998). However, cilia of the AWA and AWC neurons exhibit complex, three-dimensional morphologies which are challenging to visualize and measure (Evans et al., 2006; Perkins et al., 1986). We therefore chose to focus on the cilia of the AWB neurons which are relatively simpler structurally, and which we have studied previously (Mukhopadhyay et al., 2007).

The cilia of the AWB neurons can be visualized via GFP expression driven by the AWB-specific str-1 promoter (Mukhopadhyay et al., 2007; Troemel et al., 1997) (Figure 1A, B). Although all AWB cilia in wild-type animals exhibit the characteristic Y-shaped structure, we observed animal-to-animal variability in both the lengths of each cilia branch, as well as the area of a fan-shaped structure (henceforth referred to as a fan) occasionally present on either branch (Mukhopadhyay et al., 2007) (Table 1). Dramatic and highly penetrant defects in AWB cilia structure were observed in animals mutant for the odr-1 guanylyl cyclase, the tax-2/tax-4 cyclic nucleotide-gated channel subunits, and the grk-2 G protein-coupled receptor kinase genes (Figure 1C, D, Table 1). These molecules are expressed in, and required for the sensory functions of the AWB neurons (Coburn and Bargmann, 1996; Fukuto et al., 2004; Komatsu et al., 1996; L’Etoile and Bargmann, 2000). In tax-2/4 mutants, the length of each ciliary branch was truncated, whereas the area of the fan was significantly increased (Table 1, Figure 1C, C’). In odr-1 and grk-2 mutants, we observed a similar increase in the area of the fan in all examined AWB cilia, with less severe changes in cilia branch lengths (Table 1, Figure 1D, D’). However, the overall surface area of the AWB cilia is not significantly altered in signaling mutants [total surface area: wild-type - 56.8 ± 8.6 sq μm (n=5), odr-1 - 47 ± 5.9 sq μm (n=5), and tax-4 - 57.5 ± 10.1 sq μm (n=4)]. Similar phenotypes were observed in animals carrying different alleles of each gene (Table 1), as well as with gfp expressed under multiple AWB-selective promoters (data not shown). Expression of wild-type tax-2 and grk-2 cDNAs under the str-1 promoter was sufficient to restore wild-type cilia morphology in tax-2 and grk-2 mutants, respectively (Table 1), indicating that these genes act cell-autonomously to regulate AWB cilia structure. Although the ASI neurons also express these molecules (Komatsu et al., 1996; L’Etoile and Bargmann, 2000), we did not observe any ciliary defects in this neuron type (Table S1).

The ODR-3 Gα_i/o subunit has been suggested to transduce signals from ligand-bound receptors to the cGMP-mediated signaling pathway in C. elegans chemosensory neurons (Roayaie et al., 1998). However, loss-of-function (lf) mutations in odr-3 caused a loss of all fan-like structures in examined cilia (Table 1), whereas overexpression of a constitutively active ODR-3 [odr-3(Q206L)XS] protein caused a marked shortening of the cilia branches (Table 1). Cilia phenotypes of animals doubly mutant for odr-3(lf) and odr-1 or tax-4, and a tax-4; odr-3; odr-1 triple mutant strain were similar to those of odr-3(lf) alone (Table 1), suggesting that ODR-3 either acts downstream of these molecules, or in a parallel pathway. Taken together, these results indicate that mutations in a subset of AWB-expressed sensory signaling genes affect both AWB cilia branch length and shape.

AWB ciliary structures are altered in the absence of bacterial food

Altered AWB ciliary structures in signaling mutants could arise as a consequence of compromised sensory signaling, or alternatively, due to sensory signaling-independent roles of these molecules in the regulation of ciliary architecture. To distinguish between these possibilities, we investigated whether AWB cilia structures are similarly altered in wild-type animals grown in the absence of bacterial food.

Olfactory neurons in C. elegans respond to chemicals produced by their bacterial food source (Bargmann et al., 1993). The AWB neurons have previously been shown to respond to the chemicals 2-nonanone and serrawettin W2 produced by pathogenic bacteria, and likely also mediate responses to additional bacteria-derived cues (Pradel et al., 2007; Troemel et al., 1997). We reasoned that growing animals in chemically defined axenic media (CeMM) in the absence of bacteria may decrease sensory signaling while simultaneously permitting growth and development, albeit at slower rates (Szewczyk et al., 2003; Szewczyk et al., 2006).

We found that the AWB cilia of adult wild-type animals grown in either liquid or solid CeMM exhibited ciliary phenotypes similar to those observed in sensory signaling mutants (Fig. 1E, E’, Table 1, Table S2). The observed ciliary phenotypes were likely not due to the altered metabolic rates and caloric restriction of CeMM-grown animals (Szewczyk et al., 2006), since no AWB ciliary phenotypes were observed in eat-6(ad467) Na⁺/K⁺ ATPase mutants which also exhibit altered metabolic rates due to feeding defects (Avery, 1993; Davis et al., 1995; Lakowski and Hekimi, 1998) (data not shown). Ciliary structural defects were suppressed when the CeMM was supplemented with bacteria (Fig. 1F, F’, Table 1, Table S2). The AWB ciliary phenotypes of odr-1 and tax-4 mutants were not further altered upon cultivation in CeMM with or without bacteria (Table S2). Growth in CeMM also did not affect the ciliary structures of the ASH or ASI chemosensory neurons (Table S1). These results are consistent with the hypothesis that the AWB ciliary structures are altered to compensate for decreased sensory signaling, suggesting that the similar structural alterations observed in signaling mutants are due to altered sensory signal transduction and not solely due to signaling-independent roles of these molecules in other cellular processes. However, it remains possible that CeMM is instructive for the formation of altered AWB ciliary structures, or that the expression or localization of one or more signaling genes is reduced upon growth in CeMM.

Ultrastructure of the AWB cilia in signaling mutants

We previously described the ultrastructure of the wild-type AWB cilia, and showed that the far-distal segments do not appear to contain an axoneme (Mukhopadhyay et al., 2007). We wished to determine whether the fan-shaped structures observed in sensory signaling mutants also lacked an axoneme, and were perhaps membraneous extensions of the corresponding segments. Examination of serial sections of AWB cilia in odr-1 and tax-4 mutants via electron microscopy showed that similar to wild-type animals, the middle segments of these cilia contained singlet and rare doublet microtubules, whereas rare singlet microtubules were present more distally (Movies S1-3). The length of each cilia branch was truncated in tax-4 mutants. In both odr-1 and tax-4 mutants, we observed a highly flattened structure corresponding to the fan at the distal segments of both cilia branches (Figure 2C, D). Three-dimensional reconstructions of the serial sections indicated that the fan structures are asymmetric, extending from each cilia branch in the direction away from the midline (Figure 2A). Similar structures were not observed in wild-type cilia grown under standard conditions (Figure 2A, B). The fan contained few if any singlet microtubules, which were generally restricted to the area proximal to the midline. Thus, mutations in odr-1 and tax-4 result in an altered membraneous structure at the distal segments of the AWB cilia. Examination of the micrographs did not reveal any structural alterations in channel cilia in signaling mutants (Figure 2C, D, Movies S2, S3).

A subset of signal transduction molecules are localized to the fan-shaped membraneous areas

Signaling molecules such as receptors and channels are preferentially localized to the membranes of sensory cilia. We wished to determine whether the extended membraneous structures in CeMM-grown wild-type animals and in odr-1 and tax-4 signaling mutants also housed AWB-expressed transmembrane signal transduction molecules, or whether the composition of these areas was distinct.

To address this issue, we examined the localization of GFP-tagged ciliary transmembrane proteins including the AWB-expressed STR-1 and SRD-23 chemoreceptors, TAX-2 and ODR-3. We verified that the tagged TAX-2 and ODR-3 proteins were functional by rescuing the AWB cilia defects in tax-2 and odr-3 mutants (Table 1), although no str-1 or srd-23 mutants are currently available. In wild-type or rescued animals grown under standard conditions, all GFP-tagged fusion proteins with the exception of TAX-2 were localized throughout the AWB cilia, whereas TAX-2 was localized in a discrete proximal domain (Figure 3A-D). In CeMM-grown wild-type animals and in odr-1, tax-4 and grk-2 mutants, GFP-tagged STR-1, SRD-23, and ODR-3 fusion proteins were also present throughout the cilia, including in the fans (Figure 3A-C). We did not observe any GFP-tagged TAX-2 protein in either odr-1 or tax-4 mutants indicating that TAX-2 localization may depend on these genes (data not shown). Interestingly however, TAX-2::GFP remained restricted to a proximal zone in grk-2 mutants as well as in CeMM-grown wild-type animals, in a pattern similar to that in the cilia of wild-type animals grown in the presence of bacteria, and was excluded from the fan (Figure 3D). The localization of IFT particle and motor proteins such as OSM-6, OSM-3 and KAP-1 was also unaltered in signaling mutants, and these proteins were largely excluded from the membraneous expansions (Figure 3E and Figure S1). These results further indicate that the observed fans correspond to altered membrane structures, and that the overall distribution pattern of a subset of ciliary transmembrane signaling molecules is altered upon decreased sensory signaling (summarized in Figure 3F).

TAX-2 acts during postembryonic development to maintain AWB cilia structure

Cilia are formed during late embryonic development (Sulston et al., 1983). Sensory signaling may play a role in the generation or maintenance of these structures. To investigate the developmental stage at which sensory molecules are required to modulate AWB cilia architecture, we performed temperature-shift experiments using the temperature-sensitive tax-2(ks31) allele. The AWB cilia of tax-2(ks31) mutants exhibited wild-type morphology when animals were grown at the permissive temperature of 15°C, but exhibited large fan-like structures and truncated branch lengths when grown at 25°C (Figure 4A-D). We grew tax-2(ks31) animals at 15°C or 25°C, and transferred them to 25°C or 15°C respectively, at different larval stages. The structure of AWB cilia of adult animals moved from 25°C to 15°C at the L1 larval stage was similar to those of animals grown throughout their life cycle at 15°C, suggesting that the earlier embryonic phase of cilia development is not affected by mutations in tax-2. The AWB cilia of adult animals moved from 15°C to 25°C as late as the L4 larval stage also exhibited significantly aberrant morphology (Figure 4A-D), indicating that TAX-2 function is required at late larval stages to maintain cilia architecture. Conversely, AWB cilia defects were decreased in animals moved from 25°C to 15°C at L2 and later larval stages (Figure 4A-D). These results indicate that TAX-2 function is required during late larval stages to modulate AWB cilia morphology.

cGMP-mediated signaling modulates cilia morphology

During chemosensory signal transduction, increased levels of cGMP generated by the guanylyl cyclase ODR-1 is thought to gate the TAX-2/4 channel (L’Etoile and Bargmann, 2000). Thus, in odr-1 mutants, the TAX-2/4 channels may fail to open under appropriate conditions of sensory stimulation. If these molecules act similarly to regulate cilia architecture, then increasing cGMP levels in odr-1 but not tax-4 mutants may bypass the mutant phenotype. Addition of the membrane permeable cGMP analog 8-Br-cGMP has previously been shown to bypass the phenotypes of the daf-11 guanylyl cyclase mutant in the regulation of dauer formation (Birnby et al., 2000). We found that growing odr-1(n1936) but not tax-4(ks11) animals on plates containing 8-Br-cGMP partially suppressed the AWB cilia phenotypes (Figure 4E, 4F). Moreover, consistent with the finding that this pathway continues to act late in development to regulate cilia structure, addition of 8-Br-cGMP at later larval stages also resulted in significant rescue of the mutant phenotype (Figure 4E, 4F). Interestingly, addition of this cGMP analog also decreased both the percentage of cilia with fans, as well as overall fan area of wild-type AWB cilia (Figure 4E, 4F), suggesting that growth of the fan is inversely correlated with intracellular cGMP levels.

Increased calcium-mediated signaling can bypass the cilia defects of a subset of signaling mutants

Gating of TAX-2/4 channels by cGMP is predicted to result in influx of cations such as Ca²⁺ (Kaupp and Seifert, 2002; Kimura et al., 2004). Mutations in grk-2 have also been shown to decrease stimulus-evoked Ca²⁺ influx in the ASH sensory neurons (Fukuto et al., 2004). Thus, intracellular Ca²⁺ levels in the AWB neurons are predicted to be decreased in odr-1, tax-2/4 and grk-2 mutants. We wondered if the defects in cilia architecture in these signaling mutants could be bypassed by increasing intracellular Ca²⁺ levels.

The L-type voltage-gated Ca²⁺ channel egl-19 is broadly expressed, including in neurons (Lee et al., 1997; data not shown). gf mutations in egl-19 prolong the duration of depolarization (Avery, 1993; Lee et al., 1997; Raizen and Avery, 1994). egl-19(gf) mutations had minor effects on AWB cilia morphology on their own, but strongly suppressed the cilia defects and the transmembrane protein localization phenotype of both odr-1(n1936) and grk-2(gk268) mutants (Table S3; Figure 3A). lf mutations in egl-19 had minor effects on AWB cilia structures in either wild-type or mutant backgrounds (Table S3). lf mutations in the UNC-43 Ca²⁺/calmodulin-dependent protein kinase II (CaMKII) (Reiner et al., 1999) alone also resulted in minor effects on fan structure, but did not further modify the odr-1 cilia phenotype (Table S3). However, constitutively activated UNC-43 suppressed the odr-1(n1936) AWB cilia phenotype (Table S3). Neither egl-19(gf) or unc-43(gf) suppressed the cilia defects of tax-4(ks11) mutants (Table S3). These results suggest that increased Ca²⁺-mediated signaling is partly sufficient to bypass the requirement for a subset of signaling molecules, but not that of the cyclic nucleotide-gated channel, in the regulation of AWB cilia structure.

Sensory signaling-mediated remodeling of AWB ciliary architecture requires Kinesin-II

We next investigated the mechanisms required for the signaling-mediated structural alterations in the AWB cilia. Formation of the fans and localization of transmembrane proteins to these structures may occur via diffusion following vesicle fusion at the cilia base, or require active transport via IFT-dependent or IFT-independent mechanisms (Jenkins et al., 2006; Peden and Barr, 2005; Qin et al., 2005; Scholey and Anderson, 2006). We recently showed that the Kinesin-II and OSM-3 anterograde motors act redundantly and independently of each other to form the AWB cilia middle segments, whereas neither motor appears to be essential to form the distal segments (Mukhopadhyay et al., 2007). This is in contrast to the functions of these motors in channel cilia where both motors act cooperatively to form the middle segments, and OSM-3 alone acts to elongate the distal segments (Snow et al., 2004).

To determine whether IFT is altered in signaling mutants, we first investigated the movement of the IFT particle component OSM-6, and the KAP-1 Kinesin-II and the OSM-3 motor subunits in the AWB cilia (Mukhopadhyay et al., 2007) of odr-1, tax-4 and odr-3 mutants. Consistent with the localization of GFP-tagged IFT proteins and motors to the axoneme (Figure 3E), IFT could be observed only along the main cilia branches, and no IFT was observed in the fan-like structures (Figure S1; Movies S4-6). The velocities and patterns of movement of GFP-tagged OSM-6 and kinesin motors in odr-1, tax-4 and odr-3 mutants were similar to those observed in wild-type AWB cilia (Table S4; Mukhopadhyay et al., 2007), indicating that IFT was not grossly altered by these mutations.

We next determined whether formation of the fans requires IFT motor-based transport by examining the AWB cilia in CeMM-grown wild-type and signaling mutant animals lacking motor protein genes. Intriguingly, loss of function of either the kap-1 or klp-11 Kinesin-II subunits alone resulted in a smaller fan in wild-type animals grown under standard conditions, and strongly suppressed the increased fan phenotypes in CeMM-grown wild-type animals, as well as in odr-1, tax-4 and grk-2 mutants (Table 2). Correspondingly, the localization pattern of the STR-1::GFP fusion protein was also restored to the wild-type pattern (Figure 3A). This suppression was abolished upon expression of a wild-type kap-1 cDNA specifically in the AWB neurons (Table 2). However, the truncated cilia branch phenotype of tax-4 mutants was not suppressed by mutations in Kinesin-II subunits (Table 2), indicating that cilia membrane biogenesis and axoneme length may be regulated by distinct mechanisms. No suppression was observed in osm-3 mutants (Table 2, Figure 3A). These results indicate that formation of the fans and the correlated localization pattern of transmembrane proteins upon reduction of sensory signaling requires Kinesin-II, but not OSM-3-mediated transport.

Since mutations in sensory signaling genes lead to ciliary structural phenotypes, the sensory behavioral defects exhibited by these mutants could be due to these ciliary defects. However, klp-11; odr-1 double mutants failed to avoid the AWB-sensed volatile repellent 2-nonanone similar to odr-1 mutants despite exhibiting wild-type ciliary morphology (Figure S2). Previously, defects in AWA and AWC-mediated olfactory signaling in odr-3 mutants were also suggested not to be caused by their ciliary defects (Lans et al., 2004). We could not examine the behaviors of wild-type animals grown in CeMM since these animals exhibited locomotory defects (S.M. and P.S., unpublished observations). These observations suggest that the altered ciliary structures in signaling mutants may not directly contribute to sensory signaling defects, and that the behavioral defects of these mutants are likely due to their roles in sensory signal transduction. Interestingly, klp-11 mutants exhibited weak defects in 2-nonanone avoidance (Figure S2), suggesting that Kinesin-II may play a role in the transport of signaling molecules in the AWB cilia similar to previous observations in the AWC cilia (Evans et al., 2006) and in the flagella of Chlamydomonas (Pan and Snell, 2003).

BBS proteins and RAB-8/10 interact with the sensory signaling pathway to regulate AWB cilia structure

Recent work suggests that BBS proteins and RAB8 play a role in ciliary membrane biogenesis by facilitating docking and fusion of post-Golgi vesicles at the ciliary base (Moritz et al., 2001; Nachury et al., 2007). BBS proteins have also been suggested to couple the Kinesin-II and OSM-3 motors to regulate ciliogenesis in C. elegans (Ou et al., 2005). The formation of a membraneous fan containing a subset of cilia-targeted transmembrane proteins in response to reduced sensory signaling, and the requirement of Kinesin-II in this process, raised the possibility that BBS and/or RAB8 proteins may play a role in the formation of these altered structures. Homologs of many BBS proteins are encoded by the C. elegans genome (Blacque et al., 2004; Li et al., 2004; Ou et al., 2005). C. elegans RAB-8 and RAB-10 are related to mammalian RAB8, and play roles in vesicle trafficking and endocytic recycling respectively in non-neuronal tissues (Chen et al., 2006; Kamikura and Cooper, 2006). Both proteins are expressed broadly in C. elegans, including in neurons (Chen et al., 2006; S.M. and P.S. unpublished results).

Consistent with observations in channel cilia (Blacque et al., 2004), mutations in single bbs genes such as bbs-7 caused truncated AWB cilia at a low penetrance, although no gross defects were observed in bbs-8, bbs-1, rab-8 and rab-10 single mutants (Table 3 and Figure S3). However, mutations in all three bbs genes, as well as in both rab-8 and rab-10 suppressed the odr-1 and tax-4 phenotypes of the fan in the AWB cilia (Table 3, Figure S3). The area of STR-1::GFP localization corresponded to the area of the ciliary membrane in examined single and double mutants (Figure 3A). arl-3 also encodes a small GTPase and is expressed in ciliated neurons (Blacque et al., 2005). No suppression was observed upon loss of arl-3 (Table 3), indicating the specific requirement for RAB-8/10 in this process. Surprisingly, animals doubly mutant for odr-1 and all three bbs genes exhibited synergistic enhancement of the truncated AWB cilia phenotype (Table 3, Figure S3). Moreover, overexpression of a constitutively activated RAB-8 protein in the AWB neurons resulted in a loss of AWB ciliary branches (Table 3, Figure S3). These data imply that BBS and RAB-8-related proteins are required for the formation of the membraneous fans and localization of transmembrane proteins to the fans in sensory signaling mutants. Moreover, both BBS proteins and RAB-8 may regulate axoneme length.

Strain construction

Strains were obtained from the Caenorhabditis Genetics Center, the National Bioresource Project (Japan) and Cori Bargmann. Stably integrated strains used in this work were: kyIs104 (str-1p::gfp) (Troemel et al., 1997), odr-3(Q206L)XS (Roayaie et al., 1998), oyIs14 (sra-6p::gfp) (Troemel et al., 1997), and kyIs128 (str-3p::gfp) (Peckol et al., 1999). Double or triple mutant strains were constructed using standard methods.

Molecular biology

str-1 promoter driven cDNA constructs were generated by inserting 3 kb of str-1 upstream regulatory sequences together with the respective cDNAs into C. elegans expression vectors. A subset of cDNAs was tagged with GFP-encoding sequences at their C-terminal ends. Full-length gfp-tagged str-1 and srd-23 fusion constructs have been described previously (Colosimo et al., 2004; Dwyer et al., 2001). The gfp-tagged grk-2 construct and grk-2 cDNA was obtained from D. Ferkey (Fukuto et al., 2004). rab-8p::gfp and egl-19p::gfp fusion constructs were generated by the PCR-fusion method (Hobert, 2002). All plasmids were verified by sequencing.

Transgenic strains were generated by injecting plasmids at 30ng/μl except for str-1 promoter driven gfp-tagged tax-2 and odr-3 cDNAs, and full-length gfp-tagged str-1 and srd-23 constructs which were injected at 5 ng/μl. The str-1p::rab-8(Q67L XS) plasmid was injected at 100 ng/μl. unc-122::dsRed was used as the coinjection marker in all cases.

Microscopy

Images were acquired on a Zeiss Axioplan microscope equipped with a CCD camera (Hamamatsu). Fan area and cilia branch lengths were quantified using OpenLab 4.0 software (Improvision).

Confocal images were acquired using a Leica spectral confocal microscope equipped with 63X/1.4 NA and 100/1.4 NA objectives. Volumetric representations of cilia, and calculations of total membrane surface area were performed by three-dimensional rendering of stacks of confocal images using the Amira 4.1.1 graphics package.

Electron microscopy

Adult hermaphrodites were fixed, stained, embedded in resin, and serially sectioned using standard methods (Lundquist et al., 2001). Imaging was performed with an FEI Tecnai G2 Spirit BioTwin transmission electron microscope equipped with a Gatan 4K × 4K digital camera. Images of serial sections were processed, aligned and modeled as surface renderings using the IMOD 3.9.3 package (http://bio3d.colorado.edu/imod/) (Kremer et al., 1996).

Growth in CeMM

Animals were cultured in CeMM media (generous gift of Nathaniel Szewczyk and Lewis Jacobson and custom made by Mediatech) according to previous protocols (Szewczyk et al., 2003). In brief, gravid animals from 2-3 6 cm NGM plates seeded with HB101 bacteria were bleached and washed in M9 buffer to release eggs and kill bacteria. Growth-synchronized arrested L1 larvae were obtained by allowing these eggs to hatch by overnight incubation in M9 buffer. L1 larvae were then suspended in 1X CeMM liquid media, or cultured on CeMM plates containing 1.7% agar at 20°C in the dark. CeMM-cultured animals typically grew to the adult stage in 6-8 days.

Statistical analyses

Analyses were performed using the SPSS 13 statistical analyses package. Percentages were compared using a chi-square test. Fan areas were compared using a nonparametric independent sample Mann-Whitney U test for non-normal distributions; cilia lengths were compared using the independent sample t-test.