General and cell-type specific mechanisms target TRPP2/PKD-2 to cilia

Autosomal dominant polycystic kidney disease (ADPKD) is the most common monogenic disease, affecting one in 800 individuals(Igarashi and Somlo, 2002). In individuals with ADPKD, the kidneys accumulate multiple cysts that ultimately cause end-stage renal disease. Mutation in the PKD1 or PKD2gene accounts for 95% of ADPKD cases(Hughes et al., 1995; Mochizuki et al., 1996). PKD1 and PKD2 encode a large 11-transmembrane spanning receptor (TRP polycystin 1, TRPP1) and transient receptor potential channel(TRP polycystin 2, TRPP2), respectively(Clapham, 2003). Mammalian TRPP1 and TRPP2 (PKD2 - Mouse Genome Informatics) localize to primary cilia of kidney epithelial cells (Pazour et al.,2002; Yoder et al.,2002), where they function in a mechanosensory capacity(Nauli et al., 2003). ADPKD is one of a number of human genetic diseases that are rooted in defects in cilia formation, maintenance, or function(Pazour, 2004; Pazour and Rosenbaum, 2002; Watnick and Germino, 2003). Understanding the mechanisms by which TRPP1 and TRPP2 are transported to and function in cilia is important from both the basic cell biology and human health perspectives.

The nematode Caenorhabditis elegans is a powerful model system with which to understand how cilia form and function. C. eleganspolycystins TRPP1/LOV-1 and TRPP2/PKD-2 localize on the ciliary membrane and are required for male sensory behaviors(Barr et al., 2001; Barr and Sternberg, 1999). Hence, the connection between the polycystins and cilia seems to be an ancient one. Moreover, many C. elegans proteins that are required for formation, maintenance and function of cilia are linked to human renal diseases, including ADPKD, Bardet-Biedl Syndrome (BBS) and nephronophthisis(Barr, 2005).

PKD-2 is a member of the evolutionarily conserved TRP ion channel superfamily and acts in the same genetic pathway with LOV-1. In male-specific ciliated sensory neurons, PKD-2 and LOV-1 are required for two mating behaviors - response to mate contact and location of the mate's vulva - and are postulated to sense cues from the mate. TRP channels function as sensory receptors of diverse stimuli (Clapham,2003). TRP vanilloid (TRPV) channels localize to cilia in mammals, Drosophila, and the nematode(Andrade et al., 2005; Colbert et al., 1997; Kim et al., 2003). In C. elegans, the TRPV channels OSM-9 and OCR-2 are mutually dependent on each other for ciliary localization and sensory function(Tobin et al., 2002). In Drosophila, the ciliary localization of TRPV hearing channels Nanchung (NAN) and Inactive (IAV) is also codependent(Gong et al., 2004). Human TRPP1 has been implicated in transporting TRPP2 from ER to plasma membrane(Hanaoka et al., 2000). A recent study found that TRPP2 localizes to cilia independently of TRPP1 in LLC-PK1 and MDCK epithelial cells (Geng et al., 2006), while another shows that TRPP2 ciliary localization is dependent of TRPP1 in collecting duct-derived epithelial cells(Nauli et al., 2003).

Intraflagellar transport (IFT) is a motility process required for the assembly and maintenance of cilia and flagella(Rosenbaum and Witman, 2002)and defects in IFT may result in PKD(Pazour and Rosenbaum, 2002). IFT is required for the ciliary movement of axonemal structural components,the IFT machinery itself, and the membrane-bound sensory receptors OSM-9 and OCR-2 (Kozminski et al., 1995; Kozminski et al., 1993; Orozco et al., 1999; Qin et al., 2005; Qin et al., 2004; Snow et al., 2004). PKD-2 motility in cilia is not detected, suggesting that PKD-2 may diffuse into the ciliary membrane, that PKD-2 may be physically restrained at the cilium, or that at least two mechanisms regulate ciliary protein localization(Peden and Barr, 2005; Qin et al., 2005). In C. elegans and mammals, IFT does not appear to be essential for membrane receptor transport into cilia but may regulate ciliary membrane protein abundance (Pazour et al.,2002; Qin et al.,2005; Qin et al.,2001).

Dynamic regulation of receptor localization is a common theme in signaling pathways. We previously reported that ciliary localization of C. elegans PKD-2 is modulated by its phosphorylation status(Hu et al., 2006). Casein kinase 2 (CK2) and the TAX-6 calcineurin phosphatase regulate PKD-2 function and ciliary abundance, but not the initial targeting of PKD-2 to cilia. PKD-2 ciliary abundance is regulated also by KLP-6, a cell-type specific kinesin 3(Peden and Barr, 2005). KLP-6 is not essential for ciliogenesis or for PKD-2 entrance to cilia. However,PKD-2::GFP accumulates in the ciliary base and cilium in a klp-6mutant. KLP-6 has been proposed to act as an anchor to tether PKD-2 between the ciliary membrane and microtubule axoneme or to act redundantly with the IFT machinery.

TRPP1 and TRPP2 must ultimately be localized to cilia in order to conduct the sensory function of the cell, whether it is a human renal epithelial cell or a worm sensory neuron. How TRPP1, TRPP2 and other ciliary proteins localize and gain access to the cilium, a spatially restricted organelle, is not well understood. After proper entrance to the cilium, regulation of PKD-2 ciliary abundance may be an equally important step for its sensory function. Given that it is prohibitively difficult to study TRPP2 ciliary localization in humans, we exploited the transparency of C. elegans and took a genetic approach to uncover the mechanisms governing PKD-2 localization in living animals. PKD-2 bidirectional motility rates in dendrites of male-specific sensory neurons is comparable with that of the olfactory G protein-coupled receptor ODR-10 in dendrites of chemosensory AWB amphid neurons (Dwyer et al., 2001; Sengupta et al., 1996),suggesting that a common dendritic transport machinery acts in diverse cell types. By analyzing candidate trafficking mutants, we show that PKD-2 subcellular localization requires two sorting steps. The somatodendritic sorting step acts between the neuronal cell body and dendrite, while the ciliary sorting step acts between the distal dendrite and cilium to regulate PKD-2 ciliary localization and abundance.

Nematodes were raised using standard conditions(Brenner, 1974). Strains used for injections were: `wild type', him-5(e1490)V(Hodgkin, 1983) and pha-1(e2123ts)III; him-5(e1490)V(Granato et al., 1994). A categorized C. elegans strain list follows.

CB1490, him-5(e1490)V; PS2172, pha-1(e2123ts)III; him-5(e1490)V;PT9, pkd-2(sy606)IV; him-5(e1490)V; PT572, myIs1[PKD-2::GFP;cc:GFP] pkd-2(sy606)IV; him-5(e1490)V; PT573, myIs1 IV;him-5(e1490)V; PT618, pkd-2(sy606)IV; myIs4[PKD-2::GFP; cc::GFP]him-5(e1490)V; PT621, myIs4 him-5(e1490)V.

PT624, unc-101(m1)I; pkd-2(sy606)IV; myIs4 him-5(e1490) V; PT657, lov-1(sy582)II; myIs4 him-5(e1490)V; PT658, lov-1(sy582)II;pkd-2(sy606)IV; myIs4 him-5(e1490)V; PT970, unc-101(m1)I;lov-1(sy582)II; myIs4 him-5(e1490)V.

PT628, osm-3(p802) pkd-2(sy606)IV; myIs4 him-5(e1490)V; PT629, pkd-2(sy606)IV; myIs4 him-5(e1490)V; osm-5(m184)X; PT630, che-13(e1805)I; pkd-2(sy606)IV; myIs4 him-5(e1490)V; PT631, daf-10(p821)IV; myIs4 him-5(e1490)V; PT677, che-13(e1805)I;daf-10(p821)IV; myIs4 him-5(e1490)V; PT678, daf-10(p821)IV; myIs4 him-5(e1490)V; osm-5(m184)X; PT632, che-3(e1124)I; myIs4 him-5(e1490)V; PT633, klp-11(tm324)IV; myIs4 him-5(e1490)V;PT868, kap-1(ok676) III; klp-11(tm324) IV; myIs4 him-5(e1490)V;PT870, kap-1(ok676) III; myIs4 him-5(e1490)V; PT1184, osm-12(n1606) III; myIs4 him-5(e1490)V.

PT665, pha-1(e2123ts)III; him-5(e1490) myEx342[Ppkd-2::GFP::KDEL;pBX]; PT977, pha-1(e2123ts)III; him-5(e1490) V myEx469[Ppkd-2::TRAM::GFP; pBX]; PT1126, pha-1(e2123ts)III; myIs4 him-5(e1490) V myEx500[Ppkd-2::TRAM::DsRed2; pBX].

PT1018, pha-1(e2123ts)III; him-5(e1490) V myEx481[Ppkd-2::TBB-4::DsRed2; pBX]; PT1020, pha-1(e2123ts)III;osm-3(p802) IV; myIs4 him-5(e1490)V myEx481;PT1022, pha-1(e2123ts)III; myIs4 him-5(e1490) V osm-5(m184)X myEx481;PT1024, che-3(e1184)I; pha-1(e2123ts)III; myIs4 him-5(e1490) V myEx481; PT1026, pha-1(e2123ts)III; klp-11(tm324)IV; myIs4 him-5(e1490)V myEx481.

PT1027, pha-1(e2123ts)III; him-5(e1490) V;myEx482[Punc-119::PKD-2::GFP; pBX]; PT1028, unc-101(m1)I;pha-1(e2123ts)III; him-5(e1490)V myEx482; PT1029, pha-1(e2123ts)III;lov-1(sy582)IV; him-5(e1490)V myEx482.

PT1030, pha-1(e2123ts)III; him-5(e1490) V myEx508[KAP-1::GFP;Ppkd-2::DsRed2; pBX].

The 5′ transcriptional initiation site of pkd-2 was identified using 5′ RACE (Roche). A full-length pkd-2 cDNA was generated by PCR, subcloned into the pCR-II TOPO vector (Invitrogen). The pkd-2 cDNA (2148 bp) encodes a protein of 715 amino acids. A 1.3 kb pkd-2 promoter (Ppkd-2) encoded by sequence directly upstream of the cDNA start site was amplified from a full length rescuing pkd-2 genomic PCR product (Barr et al., 2001) and subcloned into HindIII sites in the Fire Vectors (pPD95.75, pPD95.77 and pPD95.79) to generate the cloning vectors pKK64, pKK49 and pKK50. The plasmid pKK47 (Ppkd-2::GFP) is the PCR amplified promoter subcloned into HindIII and BamH1 sites of pPD95.75.

A Ppkd-2::PKD-2(cDNA)::GFP construct (pKK52 or PKD-2::GFP) was generated by subcloning the entire pkd-2 cDNA into pKK49. To target GFP to the ER (endoplasmic reticulum) of male-specific neurons (pYK4, Ppkd-2::GFP::KDEL), the pkd-2 promoter was isolated by digestion of pKK47 with HindIII and XhoI, and exchanged with the myo-3 promoter of Pmyo-3::GFP::KDEL(Labrousse et al., 1999). To generate a rough ER marker in pkd-2-expressing neurons, the TRAM cDNA from the Gateway ORFeome library (Reboul et al., 2003) was amplified and inserted into pKK47 between SalI and BamHI sites. To generate Punc-119::PKD-2::GFP, PKD-2 cDNA was subcloned into the PstI and BamHI sites in pPD95.77. The 2.4 kb unc-119 promoter(Maduro and Pilgrim, 1995)flanked by PstI sites was inserted into this construct.

Expression constructs were co-injected with pha-1(+) plasmid pBX1 into pha-1; him-5 (Granato et al., 1994). PKD-2::GFP and Pcoelomocyte(cc)::GFP,(Miyabayashi et al., 1999)were co-injected into pkd-2(sy606); him-5(Barr et al., 2001). PKD-2::GFP was integrated into the genome using gamma irradiation to generate two independently integrated lines, myIs1 and myIs4, which were outcrossed six times. myIs1 LG IV or myIs4LGV was crossed with the appropriate mutant strain.

Confocal images were collected using a 63× (NA 1.4) objective on a BioRad MRC 1024 laser-scanning microscope (Lasersharp2000™ software). Optical sections were collected between 0.4-1.0 μm and projected as z-series that were stored as .tif files and manipulated using Adobe PhotoShop. Time-lapse images were obtained with a Nikon Eclipse TE2000 microscope equipped with a 100×, 1.4 NA objective and a Photometrics Cascade 512B with a CCD87 CCD. All the images were collected at exposure time 200 mseconds per frame, no interval time between frames, with the staged worms anaesthetized with 1 mM levamisole, mounted on agarose pads and maintained at 25°C. Kymographs and movies were created using MetaMorph software. Time lapse movies are available upon request. Epifluorescence analysis was performed on a Zeiss Axioplan2 imaging system and Openlab software(Improvision). Each 12-bit grayscale image focuses on either nonoverlapping cilia or cell bodies. For intensity measurements, the ciliary or cell body region was selected to include the whole compartment. The average fluorescence intensity within cilia (F_cilia) and cell body (F_{cell body}) is normalized to the background intensity. All intensity comparisons were performed with images obtained using identical settings(exposure time, gain, offsets) avoiding saturation and with synchronized animals.

Male mating assays were performed as described previously(Barr and Sternberg, 1999),except the assays were run for 4 minutes as opposed to 10 minutes. At least 20 animals per genotype were scored per assay. Response efficiency reflects the percentage of the males that responded to the hermaphrodites within 4 minutes. Location of vulva efficiency is calculated by successful vulva location divided by the total number of vulva encounters for each male. Statistical analyses were performed by nonparametric Mann-Whitney tests with two-tailed P-value.

We generated integrated GFP-tagged PKD-2 (PKD-2::GFP) transgenes to visualize and study how PKD-2 is transported. PKD-2::GFP transgenes fully rescue the male mating defect of pkd-2(sy606) null mutants(Fig. 7), indicating that these transgenes are functional. PKD-2::GFP transgene expression patterns are consistent with previous anti-PKD-2 antibody staining(Barr et al., 2001), thereby providing an accurate readout of PKD-2 subcellular localization. pkd-2 is expressed in a subset of male-specific sensory neurons: four CEM neurons in the head, eight pairs of Ray B neurons (1-5 and 7-9, excluding R6B), and the HOB hook neuron in the tail(Fig. 1A)(Barr et al., 2001; Lints and Emmons, 2002). Each neuron that expresses PKD-2 has an exposed cilium at the distal ending of the dendrite (Sulston et al.,1980; Ward et al.,1975). In these neurons, PKD-2::GFP localizes predominantly to cell bodies and cilia, and is also observed in small dendritic and axonal puncta (Fig. 1B,C). Within the ciliary region, PKD-2::GFP is distributed throughout the cilium, but primarily concentrated at the ciliary base, which corresponds to the distal-most dendrite and transition zone regions (refer to cartoon in Table 3 and Fig. 6A)(Qin et al., 2005).

In the cell body, PKD-2::GFP is cytoplasmic and reticular, suggesting that PKD-2 may be present in the endoplasmic reticulum (ER). To visualize ER in PKD-2 neurons, we individually expressed TRAM::GFP, TRAM::DsRed2 and GFP::KDEL driven by a 1.3 kb pkd-2 promoter. TRAM is a component of ER translocon and localizes to rough ER and also post-ER compartments(Greenfield and High, 1999). Four amino acid KDEL (Lys-Asp-Glu-Leu) is an ER targeting and retention signal(Terasaki et al., 1996). TRAM::GFP and GFP::KDEL were primarily distributed in a tubular membrane structure within the cell body (Fig. 1D-G), but TRAM::GFP was also targeted to dendritic and axonal puncta and the ciliary base, but not the cilium proper(Fig. 1D,E). GFP::KDEL is predominantly located in ER lumen in the cell body(Fig. 1F,G). Double labeling with PKD-2::GFP and TRAM::DsRed2 indicated that the two share an overlapping localization pattern in ER of cell bodies(Fig. 1H). These data suggest that PKD-2 is synthesized in the ER and packaged into vesicles that are transported to the ciliary base and inserted onto the ciliary membrane, as shown for rhodopsin in photoreceptor cells(Deretic and Papermaster,1991).

Time-lapse fluorescence microscopy was used to determine if PKD-2::GFP particles move in dendrites of CEM and RnB ray neurons. PKD-2::GFP movement is bidirectional and saltatory; moving particles sometimes stop and then resume movement, occasionally reversing directions(Fig. 2A,B). Depending on directionality, PKD-2::GFP particles move at different rates. Anterograde movement (from cell body to ciliary base) averages 1.260±0.019μm/sec and retrograde movement (from ciliary base to cell body) averages 0.807±0.021 μm/sec (Fig. 2C and Table 1). The average velocities in CEM and RnB ray neurons are not statistically different (P>0.05), suggesting that the same anterograde and retrograde transport systems are responsible for PKD-2 movement in both cell types.

In dendrites of chemosensory amphid AWB neurons, the ODR-10 G-protein coupled receptor (GPCR) moves at rates similar to PKD-2(Dwyer et al., 2001). The ODR-10 and PKD-2 ciliary receptor proteins are expressed in different sensory neuron cell types, yet exhibit comparable velocities, with faster anterograde(∼1.3-1.4 μm/second) than retrograde (∼0.7-0.8 μm/second)movement (Table 2). The ciliary IFT machinery must also travel through dendrites to reach cilia. The rate of PKD-2 and ODR-10 dendritic movement is significantly different than published dendritic velocities of IFT components(Table 2)(Signor et al., 1999a),suggesting that the dendritic transport of ciliary receptors and the IFT machinery involves different motors. These data support the model that a common transport machinery operates in dendrites to move sensory receptors to and from ciliary bases.

The AP-1 μ1 clathrin adaptor UNC-101 is required for restricting ODR-10 and OSM-9 to dendrites (Dwyer et al.,2001). To determine whether PKD-2 also employs UNC-101, we examined PKD-2 localization in unc-101 mutants. In unc-101(m1) and unc-101(sy108) males, PKD-2::GFP is uniformly distributed throughout the entire neuron, including the axon,dendrite, cell body and cilium (Fig. 3D-F). There is no obvious difference in ER localization of PKD-2 in the unc-101 background. We could not detect individual PKD-2::GFP particle movement in unc-101 dendrites because the fluorescent protein is distributed evenly throughout the neuronal processes. These data indicate that UNC-101 acts at the somatodendritic sorting step, which is required for directing PKD-2 to dendrites and cilia.

The μ1 subunit of AP-1 recognizes cargo proteins via tyrosinebased[Yxxφ, φ: a bulky hydrophobic amino acid] or dileucine [D/ExxxLL/I or DxxLL] sorting signals in cytoplasmic domains of membrane proteins(Bonifacino and Traub, 2003). The N- and C-termini of PKD-2 contain one and four putative UNC-101 recognition motifs, respectively. If UNC-101 directly interacts with PKD-2,then mutation or deletion of these recognition signals should affect PKD-2 localization. Neither site-directed mutagenesis nor deletion of the N and C termini affects PKD-2 somatodendritic or ciliary localization(Hu et al., 2006) (K.M.K.,Y-K.B., J.H. and M.M.B., unpublished). These results indicate that the cytoplasmic domains of PKD-2 are not essential for somatodendritic or ciliary targeting in vivo.

We find that other adaptor protein (AP) complex components are not essential for regulating the subcellular localization of PKD-2. AP complexes mediate clathrin-coated vesicle (CCV) formation at various membrane structures in cells. Of these, AP-1 contains four subunits (γ, β1, μ1 andσ1) and localizes in the membrane of trans- Golgi network (TGN)and endosomes. apm-1 encodes C. elegans μ1-II subunit of AP-1 complex (Shim et al.,2000). In contrast to apm-1 RNAi L1 larval arrest(Shim et al., 2000), the of apm-1(cxP9732) reduction-of-function allele is viable(Martin et al., 2002). In apm-1(cxP9732) mutants, PKD-2::GFP localizes properly in cilia and cell bodies, suggesting an AP-1 complex with APM-1 as the medium subunit is not involved in PKD-2 targeting. Alternatively, residual apm-1function may be sufficient to ensure proper PKD-2 localization.

AP-2 and AP-3 are involved in endocytosis at plasma membrane and lysosomal-related organelle transport, respectively(Bonifacino and Traub, 2003). Mutations in the AP-2 and AP-3 complex [dpy-23(e840), apt-6 (ok429)and apt-7(tm920)] do not obviously affect PKD-2 dendritic or ciliary localization. UNC-11, the C. elegans homologue of a clathrin adaptor AP180, acts at pre- and postsynaptic regions to regulate clathrin-mediated endocytosis (Burbea et al.,2002; Nonet et al.,1999). We find that unc-11(e47) mutation does not alter PKD-2 localization to the dendrite or cilium. A GTPase dynamin is a central player in internalization of membrane proteins via clathrin-mediated endocytosis along with AP-2 complex(Brodsky et al., 2001). PKD-2::GFP dendritic and ciliary localization was not obviously altered in dyn-1(ky51) mutants. We conclude that UNC-101 acts at a somatodendritic sorting step to restrict PKD-2, along with other ciliary receptors, to the dendritic compartment.

PKD-2::GFP in the following mutants is not different from wild type:β-arrestin single mutant arr-1(ok401); kinesin and kinesinrelated single mutants unc-116 (e2310), unc-16 (ju146, e109),klp-10 (ok704, tm375), unc-104 (e1265) and klp-16 (ok1505);dynein single mutants dhc-1(or195ts) and dnc-1(or404ts);axonal outgrowth single mutants unc-33(e204) and unc-44(e362); and lipid raft single mutants cav-1(ok274) and cav-2(tm394).

Ion channels often require functional partner proteins for appropriate folding and targeting. To test whether PKD-2 requires its functional partner LOV-1 for ciliary targeting, we examined the localization of PKD-2::GFP in the null mutant lov-1(sy582). To quantify any changes in PKD-2::GFP abundance, we measured fluorescence intensity in cilia (F_cilia) and the cilium to cell body ratio (F_cilia/F_{cell body})(Fig. 4). The F_ciliareflects the absolute abundance of PKD-2::GFP in cilia and the F_cilia/F_{cell body} ratio measures the relative PKD-2::GFP level in the cilium to the cell body. In CEMs, PKD-2::GFP ciliary abundance in a lov-1 background is not significantly different from wild type(Fig. 4A). In rays, however,PKD-2 is often absent from RnB cilia (Fig. 4C). Moreover, the fluorescence intensity in those ray cilia with detectable PKD-2::GFP is largely reduced compared with wild type(Fig. 4D,E). We conclude that LOV-1 plays a tissue-specific role and is required for efficient somatodendritic and ciliary targeting of PKD-2 in ray but not CEM neurons.

In both the CEMs and RnBs of a lov-1 mutant, PKD-2::GFP forms one or two large aggregates in cell bodies(Fig. 3G-I). This aggregation phenotype is more severe in RnBs than CEMs, which may account for frequent loss of PKD-2 in ray cilia (Fig. 4C). PKD-2 may be retained within the ER in the absence of LOV-1. However, aggregate appearance does not resemble the normal ER structure(compare with Fig. 1D-H). The aggregates may represent an altered ER structure caused by misfolded PKD-2, or may result from the targeting of PKD-2 to an unidentified subcellular compartment. We conclude that LOV-1 plays a crucial role in PKD-2 localization within neuronal cell bodies.

We next examined PKD-2::GFP dendritic motility in the lov-1mutant. PKD-2::GFP particles exhibit similar bidirectional movement as wild type in the lov-1 dendrite. These data suggest that lov-1 is not required for PKD-2::GFP dendritic motility.

For proper somatodendritic sorting, PKD-2 requires UNC-101 and LOV-1. To determine the order of action, we examined PKD-2::GFP localization in the unc-101(m1); lov-1(sy582) double mutant. In the double mutant, as in unc-101 alone, PKD-2::GFP is distributed throughout the neuron and does not aggregate in cell bodies (data not shown). These data suggest that unc-101 acts at the somatodendritic sorting step before lov-1.

pkd-2 and lov-1 are specifically expressed in a subset of male sensory neurons and not observed in other ciliated cell types of males or hermaphrodites. To determine whether PKD-2 subcellular localization requires cell-type specific factors or simply uses a general transport machinery, we ectopically expressed PKD-2::GFP using the pan-neuronal unc-119promoter (Maduro and Pilgrim,1995). Punc-119::PKD-2::GFP is localized in the cell bodies of amphid and outer labial quadrant (OLQ) sensilla of the hermaphrodite head, but never observed in amphid or OLQ cilia(Fig. 5A,I). In the tail,Punc-119::PKD-2::GFP localizes to the cell bodies of the ciliated phasmid neurons but not their cilia (Fig. 5B). Another example comes from the R6B neurons, which do not normally express pkd-2 or lov-1. Punc-119::PKD-2::GFP is absent from R6B cilia; however, PKD-2::GFP is localized properly in R1-5B and R7-9B cilia(Fig. 5D). Similar to endogenous PKD-2, Punc-119::PKD-2::GFP localizes to cilia in greater than 92% of R3-5Bs. By contrast, Punc-119::PKD-2::GFP is found in cilia in less than 4% of R6B neurons (Fig. 5I). In the small number of animals with PKD-2::GFP in R6B cilia,PKD-2::GFP mislocalizes to distal dendrites, suggesting improper targeting. Alternatively, PKD-2 expression in R6B may an artifact of transgene overexpression. Hence, Punc-119::PKD-2::GFP localizes to cilia in only endogenous polycystin-expressing neurons and the PKD-2 somatodendritic sorting step requires cell-type specific factors.

To test the order of action between the cell-type specific factors and UNC-101, we examined Punc-119::PKD-2::GFP in the unc-101mutant. In unc-101 animals, Punc-119::PKD-2::GFP is uniformly distributed throughout only native polycystin-expressing neurons(Fig. 5E,F). In other cell types in the male and all neurons in the hermaphrodite,Punc-119::PKD-2::GFP is retained in cell bodies(Fig. 5E,F; data not shown). This data suggests mislocalization of PKD-2 to the entire neuron in unc-101 requires cell-type specific factors. We propose that cell-type specific factors associate with the UNC-101/AP-1 complex to restrict PKD-2 to the somatodendritic compartment.

One candidate cell-type specific factor is LOV-1, the partner of PKD-2. To determine whether LOV-1 is the sole cell-type specific factor,Punc-119::PKD-2::GFP was examined in the lov-1 mutant. In the lov-1 mutant, Punc-119::PKD-2::GFP fluorescence intensity in ray cilia is reduced compared with wild type, and RnB ciliary localization was often lost (Fig. 5H). In lov-1 mutants, Punc-119::PKD-2::GFP aggregates are only observed in RnB, HOB and CEM cell bodies(Fig. 5G,H). Hence, the absence of LOV-1 is not sufficient for PKD-2 aggregation in non-native neurons and additional cell-type specific factors act before lov-1 and unc-101 in the PKD-2 somatodendritic sorting step.

PKD-2 accumulates at the ciliary bases of osm-5/IFT88/Polarismutants (Qin et al., 2001). However, PKD-2 motility is not detected in wild-type cilia(Qin et al., 2005). To further explore the relationship between IFT and PKD-2 ciliary protein abundance, we characterized PKD-2::GFP distribution and dendritic motility in various IFT mutants, including complex A daf-10(p821)(Qin et al., 2001; Bell et al., 2006); complex B osm-5(m184), osm-5(mn397) and che-13(e1805)(Haycraft et al., 2003; Qin et al., 2001); and complex A/complex B double daf-10; osm-5 and daf-10; che-13 mutants. We also examined kinesin and dynein single mutants osm-3(p802)(Shakir et al., 1993), klp-11(tm324), kap-1(ok676) (Snow et al., 2004) and che-3(e1124)(Signor et al., 1999a; Wicks et al., 2000), and osm-3; kap-1 and klp-11; kap-1 double mutants.

In wild type, PKD-2 is enriched at the ciliary base and observed at lower and variable levels within the ciliary membrane(Fig. 6A). In a stark contrast to wild type, all IFT mutants examined abnormally accumulate PKD-2::GFP in the ciliary base and in the cilium for those mutants with ciliary axonemes(Fig. 3J-L; Fig. 4A,B; Fig. 6B-E). In RnB neurons of IFT mutants, PKD-2::GFP accumulation extends into the distal dendrites(Fig. 3L, thick arrow). In the CEMs, PKD-2::GFP levels in cilia are increased approximately threefold in both daf-10 complex A IFT122 and osm-5 complex B IFT88 mutants when compared with wild type (Fig. 4A,B). In osm-5 mutants, the ratio(F_cilia/F_{cell body}) is greater than daf-10because the F_{cell body} is reduced in osm-5 animals(Fig. 4B). These results suggest that, in the absence of IFT, PKD-2 is not properly removed from the ciliary compartment or the dendritic movement from cilia to cell body is compromised. Alternatively, PKD-2::GFP accumulation may be an indirect effect of abnormal cilium structure.

To discriminate between these possibilities, we started by examining CEM ciliary structure by labeling ciliary microtubules with DsRed2 tagged-β-tubulin (Ppkd-2::TBB-4::DsRed2, Fig. 6). tbb-4 is expressed in ciliated neurons (Portman and Emmons, 2004), including ray neurons and CEMs, the latter are shown in Fig. 6A-E. In wild type, the CEM ciliary axonemes are symmetrically arranged, with cilia bending outward with respect to the buccal cavity(Fig. 6A). PKD-2::GFP is enriched at the ciliary base and along the ciliary membrane(Qin et al., 2005). In most IFT mutants, CEM cilia defects are similar to those described in amphid cilia(Perkins et al., 1986). In osm-3 and osm-5 mutants, CEM cilia are stunted and often possess abnormal projections, and PKD-2 accumulates in ciliary bases and along the shortened cilia at variable levels(Fig. 6B,C). che-3cilia are longer than wild type, and abnormal PKD-2 accumulation is observed at the ciliary base and along the ciliary membrane (compare Fig. 6D with wild type in Fig. 6A). PKD-2::GFP similarly accumulates in klp-11 and kap-1, the two kinesin II mutants(Fig. 6E, only klp-11is shown). We conclude that in IFT mutants, PKD-2 is capable of entering the ciliary membrane, whether stunted or bent (see below), and accumulates both in the ciliary base and membrane. These data indicate that IFT is not essential for PKD-2 targeting to the ciliary base, but may play a role in regulating PKD-2 levels within the ciliary compartment. However, we cannot formally rule out indirect effects caused by abnormal structure.

The C. elegans BBS proteins BBS-7 and BBS-8 stabilize IFT particle complexes and are partially required for the formation of amphid distal cilia(Ansley et al., 2003; Blacque et al., 2004; Ou et al., 2005). In bbs-7(osm-12) mutants, PKD-2 subcellular localization and male mating is indistinguishable from wild type (Barr and Sternberg, 1999) (data not shown). These data indicate that BBS-7 function is not required for regulating PKD-2 ciliary localization or function.

We next determined whether IFT plays a role in PKD-2 dendritic transport. In addition to a role in cilia, kinesin 2 has been implicated in other microtubule based transport systems (Brown et al., 2005; Miller et al.,2005). IFT components such as OSM-6, OSM-5, OSM-3 and KAP-1 are also detected in dendrites of sensory neurons including male-specific neurons(Collet et al., 1998; Qin et al., 2001; Signor et al., 1999b)(Fig. 6F, arrow). To determine whether any IFT motors or polypeptides are essential for the dendritic transport of PKD-2, we examined PKD-2::GFP dendritic motility in IFT mutants. In all IFT mutants examined, PKD-2::GFP dendritic motility is still observed,confirming the primary role of IFT in the ciliary compartment. Combined, these data suggest that IFT may regulate PKD-2 ciliary abundance via removal from the cilium.

Ciliary accumulation of sensory receptors such as PKD-2 may result in profound physiological consequences. Many IFT mutants with general ciliogenesis defects exhibit male mating defects(Barr and Sternberg, 1999; Hodgkin, 1983; Perkins et al., 1986; Qin et al., 2001). In the IFT mutants examined previously, it was unclear whether PKD-2 accumulation in male-specific cilia or general sensory defects account for mating behavioral abnormalities.

Anterograde movement of IFT in C. elegans amphid cilia requires two types of Kinesin II motors: homodimeric OSM-3 and heterotrimeric kinesin II (KLP-11/KAP-1/KLP-20) (Snow et al.,2004). osm-3(p802) and kap-1(ok676) are probably null, and klp-11(tm324) may be a hypomorph(Snow et al., 2004). Similar to osm-3 and other IFT mutants, PKD-2::GFP accumulates at the ciliary base and in the cilium in klp-11, kap-1 and klp-11; kap-1single and double mutants (Fig. 6E, data not shown). However, the phenotype in kap-1 and klp-11 is distinct from other IFT mutants in two respects. First, klp-11 and kap-1 mutants develop full-length CEM cilia and amphid cilia but exhibit AWC-specific sensory defects(Evans et al., 2006). Second,CEM cilia adopt an inward trajectory, rather than an outward bend(Fig. 6E). This is the first report of morphology changes in cilia of kinesin II mutants.

To clarify the relationship between PKD-2 localization and sensory function, we performed male mating behavior assays in kinesin motor mutants(Fig. 7). osm-3(p802)animals have severely reduced response and vulva location efficiencies(Fig. 7). The Lov (location of vulva) defect of osm-3(p802) mutants is comparable with other IFT mutants with general ciliogenesis defects(Barr and Sternberg, 1999). However, kap-1 and klp-11 single mutants exhibit slight but significant reduction of response and vulva location efficiencies(Fig. 7). These data indicate that PKD-2::GFP ciliary accumulation in klp-11 and kap-1mutants compromises PKD-2 sensory function. The subtle ciliogenesis defects of kinesin II mutants strongly suggest a causal link between PKD-2 mediated sensation and the regulation of polycystin ciliary abundance by IFT. Alternative but not mutually exclusive explanations are that kinesin II is involved in transporting downstream signaling molecules or that subtle morphological changes in kinesin II mutant cilia account for behavioral defects.

Ciliary localization of polycystins is conserved from worm to human. In C. elegans sensory neurons, cilia are located at the distal-most somatodendritic compartment, while in mammalian kidney epithelial cells,primary cilia project off the apical surface. How are TRPP1 and TRPP2 targeted to cilia in these two different systems? Our data suggest that the somatodendritic and ciliary sorting steps control the subcellular localization of PKD-2 in C. elegans sensory neurons, and that each step involves both general and cell-type specific factors(Table 3). Dynamically moving PKD-2::GFP particles are detected in dendrites of male specific sensory neurons. The ciliary proteins PKD-2 and ODR-10 employ the same dendritic transport system (Dwyer et al.,2001).

UNC-101, the μ1 subunit of AP-1, is required for directing PKD-2, ODR-10 and OSM-9 to dendrites of sensory neurons(Dwyer et al., 2001). In the absence of UNC-101 and in its native environment, PKD-2::GFP is distributed throughout the neuron (Fig. 3D-F, Table 3). Hence, UNC-101 appears to play a general role in directing ciliary membrane receptors to the dendritic compartment in diverse cell types. Somatodendritic transport in neurons shares common features with the basolateral transport in epithelial cells (Bredt,1998). UNC-101 may be analogous to the μ1B subunit of the epithelial specific AP-1B complex, which mediates basolateral sorting. Is ciliary sorting in mammalian epithelial cells and in C. eleganssensory neuron comparable? Is apical sorting prerequisite for ciliary sorting in epithelia cells (or is basolateral sorting a prerequisite for ciliary sorting in sensory neurons)? In kidney epithelia, polycystins are found in the basolateral compartment in addition to cilia, yet have not been described on the apical surface (Pazour et al.,2002; Yoder et al.,2002), suggesting ciliary and apical sorting may involve distinct mechanisms.

Mammalian TRPP2 forms a complex with the γ subunit of AP-1 and with PACS-1 (phosphofurin acidic cluster sorting protein) and this association regulates TRPP2 ER versus plasma membrane localization(Kottgen et al., 2005). The mechanisms regulating mammalian TRPP2 ciliary localization were not explored in this study. There is a PACS-1-like protein (T18H9.7) encoded in the C. elegans genome. However, unlike the mammalian counterpart, C. elegans TRPP2/PKD-2 does not contain the acidic cluster to which PACS-1 binds, suggesting that PACS-1 may not play a role in PKD-2 localization in C. elegans sensory neurons.

In addition to unc-101, PKD-2 requires cell-type specific factors at the somatodendritic sorting step. When expressed pan-neuronally, PKD-2::GFP localizes to dendrites and cilia of only native neurons(Fig. 5). In non-native neurons, PKD-2 remains in the cell body(Fig. 5, Table 3). Similarly, the ODR-10 olfactory receptor also requires cell-type specific factors ODR-4 and ODR-8 for ciliary localization (Dwyer et al.,1998). Cell type-specific factors act to functionally specify a cilium and are predicted to possess a restricted expression pattern. Four novel cwp (co-expressed with polycystins) genes possess an expression pattern identical to pkd-2 and lov-1(Portman and Emmons, 2004). Future investigation on the cellular and molecular roles of the cwpgenes is warranted.

LOV-1/TRPP1, the functional partner of PKD-2/TRPP2, is an obvious example of a cell-type specific factor. In lov-1 mutants, PKD-2::GFP aggregates within cell bodies of both CEM and ray neurons and localizes to cilia at significantly reduced levels only in ray neurons. We propose that lov-1 may act at both somatodendritic and ciliary steps. At the somatodendritic sorting step, lov-1 may act after unc-101 to stabilize PKD-2 within the cell body. At the ciliary sorting step, lov-1 may stabilize PKD-2 in cilia in a cell-type specific way. In mammalian systems, the requirement of TRPP1 for TRPP2 ciliary targeting depends on the cell culture type (Geng et al., 2006; Nauli et al.,2003). Identifying the nature of the PKD-2 aggregates in the lov-1 mutant and determining whether this phenomenon occurs in ADPKD cells are important next steps in understanding TRPP2 trafficking.

The kinesin 3 KLP-6 is another cell-type specific factor regulating PKD-2 subcellular distribution (Peden and Barr,2005). Mammalian kinesin 2/KIF3 (IFT kinesins) and kinesin 3/KIF16B (KLP-6) have been implicated in transporting endocytic organelles(Brown et al., 2005; Hoepfner et al., 2005). Interestingly, post-ER vesicles are found in ciliary bases of pkd-2-expressing neurons (Fig. 1D,E). In human respiratory epithelial cells, a trans-Golgi network marker localizes to ciliary bases(Schermer et al., 2005),implicating this region as an active protein turnover and trafficking site.

In both klp-6 and IFT mutants, PKD-2 accumulates at the ciliary base and cilium, if formed. In contrast to IFT mutants, klp-6 animals have superficially normal cilia, indicating that PKD-2 accumulation phenotype is not merely a consequence of ciliogenesis defects. Additionally, our data show that dendritic motility of PKD-2::GFP particles is not abrogated in various IFT and klp-6 single mutants. Why then does PKD-2 accumulate in IFT and klp-6 mutants? IFT and KLP-6 may somehow regulate PKD-2 abundance in cilia. In Chlamydomonas, flagellar turnover products are removed from flagella via retrograde IFT(Qin et al., 2004). IFT may play a similar role in the recycling of ciliary membrane receptors and signaling molecules. Ciliary abundance of PKD-2 is regulated by phosphorylation status, which is controlled by calcineurin and CK2(Hu et al., 2006). Determining how the IFT process or IFT driven signaling molecules fit into this downregulation pathway will be informative.

Recent genomic and proteomic approaches have identified components required for formation and function of cilia and flagella(Avidor-Reiss et al., 2004; Blacque et al., 2005; Efimenko et al., 2005; Keller et al., 2005; Li et al., 2004; Ostrowski et al., 2002; Pazour et al., 2005). In contrast to ciliogenesis, very little is known regarding `sensorigenesis', the process by which a cilium is specialized for a particular function. Cilia often possess unique morphologies and express a distinct repertoire of sensory receptors and signaling molecules. TRPP1 and TRPP2 are required for the flow-induced mechanosensory properties of kidney cilia(Nauli et al., 2003; Praetorius and Spring, 2001; Praetorius and Spring, 2003a; Praetorius and Spring, 2003b),with defects resulting in ADPKD. How is this essential polycystin sensory complex regulated? How is the renal primary cilium specified for sensing urine flow? Identification of general and tissue-specific pathways mediating ciliary protein localization and function are of utmost importance from both cell biology and human health perspectives.

We thank Shohei Mitani, Laurent Segelat and the Caenorhabditis Genetics Center (CGC) for strains; Alexander Van der Bliek for the GFP::KDEL marker(Labrousse et al., 1999);Jonathan Scholey for KAP-1::GFP; Andy Fire for C. elegans vectors;Doug Portman for the tbb-4 cDNA; Erik Peden and Juan Wang for critical reading of this manuscript; John White for important experimental suggestions; Dave Saber for integration and behavioral assays; Doug Braun for excellent technical assistance; and the two anonymous reviewers for constructive criticism. This work was supported by the PKD Foundation and NIH.