The cystic kidney diseases nephronophthisis (NPHP), Meckel–Gruber syndrome (MKS) and Joubert syndrome (JBTS) share an underlying etiology of dysfunctional cilia. Patients diagnosed with NPHP type II have mutations in the gene INVS (also known as NPHP2), which encodes inversin, a cilia localizing protein. Here, we show that the C. elegans inversin ortholog, NPHP-2, localizes to the middle segment of sensory cilia and that nphp-2 is partially redundant with nphp-1 and nphp-4 (orthologs of human NPHP1 and NPHP4, respectively) for cilia placement within the head and tail sensilla. nphp-2 also genetically interacts with MKS ciliopathy gene orthologs, including mks-1, mks-3, mks-6, mksr-1 and mksr-2, in a sensilla-dependent manner to control cilia formation and placement. However, nphp-2 is not required for correct localization of the NPHP- and MKS-encoded ciliary transition zone proteins or for intraflagellar transport (IFT). We conclude that INVS/NPHP2 is conserved in C. elegans and that nphp-2 plays an important role in C. elegans cilia by acting as a modifier of the NPHP and MKS pathways to control cilia formation and development.
Human ciliopathies, including nephronophthisis (NPHP), Meckel syndrome (MKS) and Joubert syndrome (JBTS), are a class of genetic disorders defined by defective cilia and cystic kidneys (Badano et al., 2006). The primary nephropathy is often comorbid with retinal degeneration, deafness, polydactyly, obesity and situs inversus (Hildebrandt and Zhou, 2007; Pazour and Rosenbaum, 2002; Wheatley, 1995). Many syndromic ciliopathies share loci; in particular, NPHP, MKS and JBTS have 13, 10 and 13 genetic loci identified, respectively, and exhibit considerable overlap (e.g. NPHP6/MKS4/JBTS5 and NPHP11/MKS3/JBTS6) such that there are currently known 25 distinct loci (supplementary material Table S1). Disease class overlap might be due partially to the underlying oligogenic nature of the disorders (Hoefele et al., 2007). Ciliopathy-associated genes encode cystoproteins and most of these gene products localize to the cilium. The wide array of symptoms associated with ciliopathies is due to the near-ubiquitous presence of cilia on mammalian cells.
Cilia are hair-like microtubule (MT)-based organelles that protrude from cell surfaces. Sensory cilia act to receive and transmit information from the extracellular environment and are integral to many signal transduction pathways (Goetz and Anderson, 2010). Eukaryotic cilia and structurally related eukaryotic flagella are constructed by intraflagellar transport (IFT), a MT motor-based transport system. The nematode Caenorhabditis elegans is a powerful system for studying fundamental questions in cilia biology and to model human ciliopathies. The primary sensory organs of C. elegans are the amphid and phasmid sensilla in the head and tail (Perkins et al., 1986). The cilia within these sensilla have a wide range of morphologies and functions, allowing for investigation of ciliary specializations and function (Bae and Barr, 2008; Jauregui et al., 2008). The MKS1-like genes B9D1 and B9D2, first characterized as ciliary proteins in C. elegans, are ciliopathic in humans, highlighting the power of this model (Bialas et al., 2009; Dowdle et al., 2011; Williams et al., 2008).
C. elegans cilia are located on the distal dendritic endings of sensory neurons, where a centriole-derived complex consisting principally of transition fibers (TF) precedes the microtubule-based axoneme (Dammermann et al., 2009; Perkins et al., 1986). The first segment of the axoneme is the transition zone (TZ), a region where many ciliary proteins congregate, either regulating ciliary traffic, being imported or exported into the cilium, or playing structural and functional roles essential for ciliogenesis and formation of the so-called ciliary gate (Rosenbaum and Witman, 2002). The axoneme extends from the TZ. In C. elegans amphid channel cilia, the axoneme comprises a MT doublet middle segment and a MT singlet distal segment (Perkins et al., 1986).
In C. elegans cilia, most ciliopathy-related proteins are either directly implicated in IFT function (e.g. BBS-1, BBS-7, BBS-8), or localize to the TZ–basal body complex (e.g. NPHP-1, NPHP-4, MKS-1, MKS-3, MKS-6) (Bialas et al., 2009; Blacque et al., 2004; Garcia-Gonzalo et al., 2011; Jauregui and Barr, 2005; Qin et al., 2001; Williams et al., 2008; Winkelbauer et al., 2005). This TZ complex is necessary for early ciliogenesis and the TZ is a region associated with traffic into and out of the cilium (Williams et al., 2011). Single mutations in genes encoding TZ complex members yield mild or nonobservable ciliogenic defects. However, in certain pairwise combinations, these mutations cause ciliogenesis defects that are attributable to anomalies in centriole and TZ membrane anchoring (Williams et al., 2008; Williams et al., 2010; Williams et al., 2011). Mammalian NPHP1 (also known as JBTS4) and NPHP4 physically interact to form a complex (Mollet et al., 2002). In C. elegans, nphp-1 and nphp-4 single and double mutants have superficially normal cilia, though transmission electron microscopy reveals subtle ultrastructural defects (Jauregui and Barr, 2005; Jauregui et al., 2008; Winkelbauer et al., 2005). Neither is required for localization of the ciliary receptors OSM-9 and PKD-2; however, mutations in nphp-4 indirectly affect the velocity of several IFT subcomponents (Jauregui et al., 2008). mks-1 (orthologous to MKS1/BBS13) encodes a B9 domain [a C2-like lipid and calcium binding domain (Zhang and Aravind, 2010)] containing protein in the same family and pathway as mksr-1 and mksr-2 (orthologous to B9D1 and B9D2, respectively). Mutations in any or all three of these genes produce no significant cilia defects (Bialas et al., 2009; Kyttälä et al., 2006; Williams et al., 2008). MKS-1, MKSR-1 and MKSR-2 are interdependent for proper localization and they function together with NPHP-1 and NPHP-4 to regulate ciliogenesis (Bialas et al., 2009; Williams et al., 2008). mks-3 (orthologous to MKS3/TMEM67/NPHP11) plays a minor role in sensation-mediated behaviors and genetically interacts with nphp-4 to yield dysfunctional cilia in the double mutant (Williams et al., 2010). Recently, mks-6 (a C2 domain-encoding ortholog of JBTS9/CC2D2A) was shown to function together with nphp-1 or nphp-4 to control centriole and TZ anchoring to the membrane, the initial step of ciliogenesis (Williams et al., 2011). These genetic interactions have provided evidence for two functional pathways, or modules: an ‘MKS pathway’ involving MKS-1, MKSR-1, MKSR-2, MKS-3 and MKS-6, and an ‘NPHP pathway’ comprising NPHP-1 and NPHP-4.
Further work in both C. elegans and mammalian systems has reinforced these findings and extended these modules. The Jackson group identified a NPHP5–NPHP6 module in addition to the MKS and NPHP1–NPHP4 modules. These modules are physically linked through mutual interactions with NPHP2 (Sang et al., 2011). In mice, the MKS module (comprising MKS1, TMEM216, TMEM67, CEP290, B9D1 and CCD2A) is essential for ciliogenesis in some tissues and is necessary for localization of several membrane-associated ciliary proteins, including Arl13b, Smoothened and Pkd2 (Garcia-Gonzalo et al., 2011). The MKS module also contains Tectonic1 (also known as TCTN1), a hedgehog signaling component, linking ciliogenesis, membrane composition and signaling (Garcia-Gonzalo et al., 2011). Several of these genetic and functional interactions between TZ-associated genes are conserved in C. elegans, providing a powerful tool for defining genetic pathways.
Mutations in human INVS (also known as NPHP2) are responsible for NPHP type II, an infantile autosomal recessive disease. The gene product inversin is named for the inversion of visceral asymmetry in the inv mouse, which is normally mediated by nodal monocilia (Morgan et al., 1998; Watanabe et al., 2003). Renal cysts of inv mice resemble those in infantile NPHP type II patients (Phillips et al., 2004). Interestingly, the N-terminal ankyrin repeat-containing region of inversin can localize to node cilia and rescue left–right defects, but not renal cysts, indicating that inversin is a multifunctional protein that acts in a cell-type-specific manner (Watanabe et al., 2003). In Xenopus and zebrafish, inversin acts as a molecular switch between canonical and noncanonical Wnt signaling by binding cytoplasmic disheveled (Dvl) and targeting it for degradation; disruption of this pathway interchange is concordant with disease expression (Bellavia et al., 2010; Benzing et al., 2007; Simons et al., 2005). In mammals, several isoforms of inversin localize to the proximal segment of sensory cilia, a region termed the ‘Inv compartment’. The zebrafish ortholog of MKSR-2, B9D2, physically interacts with and modulates ciliary localization of inversin, which in turn is necessary for anchoring the NPHP3 and NPHP9 gene products to the same compartment (Shiba et al., 2009; Shiba et al., 2010; Zhao and Malicki, 2011).
In this study, we show that the C. elegans ortholog of INVS, nphp-2, is expressed in ciliated sensory neurons and encodes two protein isoforms that localize to the ciliary middle segment or Inv compartment. We also demonstrate that nphp-2 is partially redundant with nphp-1 and nphp-4 in regulation of TZ placement, TZ orientation and IFT particle velocity, but not for localization of other TZ proteins. Cilia in the nphp-2 nphp-4 double mutant are severely compromised, although neither nphp-2 nor nphp-4 is necessary for localization of B9 domain proteins MKS-1, MKSR-1 and MKSR-2 (Williams et al., 2008). Genetic experiments on nphp-2 and MKS genes encoding TZ-localizing proteins demonstrate a complex interaction network, which varies in a sensillum-dependent manner. We conclude that nphp-2 plays an important role in C. elegans cilia and that nphp-2 acts as a sensillum-specific modifier of the previously described NPHP and MKS pathways.
Y32G9A.6 is the C. elegans ortholog of inversin
Inversin possess several domains, including protein-interacting ankyrin repeats, two nuclear localization signals, two calmodulin-binding IQ domains, a basic-residue-enriched region and a C-terminal ninein homology region (Fig. 1A) (Morgan et al., 2002). At least two of these regions, the ankyrin repeat and the ninein homology regions, are independently sufficient for ciliary localization of inversin (Shiba et al., 2009; Watanabe et al., 2003). Protein BLAST (Basic Local Alignment Search Tool) homology searches against the C. elegans genome yielded several candidates, including Y32G9A.6 (see Materials and Methods). The Y32G9A.6 promoter contains an X-box motif, found in many ciliary genes, and has been identified in at least two ciliary genomic screens, one of which demonstrated expression in ciliated neurons, and predicted Y32G9A.6 as ‘inversin-like’ (Blacque et al., 2005; Efimenko et al., 2005). Protein primary sequence analysis of Y32G9A.6 predicts a D-box ubiquitylation site, a bipartite nuclear localization signal and an EF-hand (Fig. 1A). All domains present in Y32G9A.6, with the exception of the EF hand, are conserved in inversin orthologs across several species (Fig. 1A) (Morgan et al., 2002). Rather than having a direct calcium-binding EF hand, inversin possesses an IQ domain, which binds the calcium sensor calmodulin. Amino acid sequence comparison of Y32G9A.6 and inversin shows 19.1% identity and 35.2% similarity, and only marginally less (16.9% identity and 32.6% similarity) when ankyrin repeats are omitted. This is less than other ciliary proteins, but is still significant (supplementary material Table S2). On the basis of BLAST results, presence of an X-box promoter motif, conserved protein motifs and domain organization similar to inversin, we characterized and showed that Y32G9A.6 is the C. elegans ortholog of inversin. Y32G9A.6 was coexpressed and genetically interacted with the other NPHP genes (see below). We hereafter refer to the sequence Y32G9A.6 as nphp-2.
nphp-2 is expressed in ciliated neurons and encodes two cilium-localized isoforms
The expression pattern of nphp-2 was determined through use of native promoter-driven GFP expression. Pnphp-2::GFP was expressed in the ciliated sensory nervous system throughout development (data not shown). In the adult, expression was evident in both hermaphrodite and male ciliated sensory neurons, including amphid, phasmid and IL2 neurons. nphp-2 was also expressed in the male-specific ciliated sensory nervous system, including the CEM, RnB and HOB neurons (Fig. 1B). Expression in the internal oxygen sensor neuron PQR was visible in the hermaphrodite, but could not be distinguished from the tail sensilla in the male.
RT-PCR with subsequent sequencing was used to confirm the predicted splicing pattern and cDNA sequence, and revealed two distinct isoforms. In addition to the predicted full-length transcript (long isoform, NPHP-2L), a more abundant (approximately tenfold) shorter isoform (NPHP-2S) was detectable. This latter isoform, spliced from within exon 10 to the start of exon 11, lacks a region encoding a 22 residue fragment (VLIARKNARALFRNYYHPGTEQ), which does not contain or lie within any identified domain. To ascertain the subcellular localization of each isoform, we used native promoter-driven expression of GFP-tagged splice form cDNA, e.g. the short form Pnphp-2::NPHP-2S::GFP and the long form Pnphp-2::NPHP-2L::GFP. Cilia in transgenic NPHP-2S::GFP and NPHP-2L::GFP animals formed properly in a wild-type background, as determined by dye filling (supplementary material Fig. S1). Both NPHP-2S and NPHP-2L transgenes were functional and rescued dye filling defects in the nphp-2 nphp-4 double mutant (supplementary material Fig. S2). To determine subciliary localization, NPHP-2S::GFP and NPHP-2L::GFP were coexpressed with Pnphp-1::NPHP-1::CFP, a transition zone (TZ) marker. NPHP-2S::GFP localized cytoplasmically throughout neurons, including dendrites, cell bodies and axons, and to the middle segment of amphid channel and phasmid cilia (Fig. 1C; supplementary material Fig. S3). NPHP-2S::GFP was excluded from the TZ and the nucleus (Fig. 1C,D; supplementary material Fig. S3). NPHP-2L::GFP also primarily localized to the middle segments of amphid channel and phasmid cilia; dim signals were observable in cell bodies and dendrites, though not in axons. In addition, several bright NPHP-2L::GFP puncta were observable within the cell body (Fig. 1C,D). Both NPHP-2S and NPHP-2L localized to the middle segment at all stages of development, from larval stage L1 to young adult (data not shown). The middle segment localization of NPHP-2S and NPHP-2L differs markedly from the TZ localization of most other cystoproteins, including NPHP-1, NPHP-4, MKS-1 and MKS-6 (Bialas et al., 2009; Jauregui and Barr, 2005; Williams et al., 2008; Williams et al., 2011; Winkelbauer et al., 2005). To confirm middle segment localization, the length of the NPHP-2S and NPHP-2L localizations was measured (2.5±0.59 μm) and was found to be similar to the length of the MS as measured by electron microscopy (2–3 μm) (David Hall, personal communication). GFP-tagged NPHP-2S and NPHP-2L ciliary motility was undetectable, indicating that NPHP-2 is, like the TZ proteins, unlikely to be a component of the IFT machinery (data not shown).
nphp-2 is necessary for proper placement of TZ and cilia
The available nphp-2 mutant gk653 contains a deletion from upstream of the start codon to within the first intron, removing the start ATG. A hidden Markov model splice site predictor correctly predicts the wild-type splice pattern, but predicts an alternate start site in gk653, which yields a transcript with a dysfunctional first exon followed by a continuation of the correct transcript (data not shown). RT-PCR of nphp-2 cDNA confirms the presence of a transcript in gk653 mutants (data not shown). We therefore generated (by Mos1 transposon mutagenesis) and examined two additional nphp-2 alleles, nx101 and nx102, which contain in-frame deletions over ankyrin repeats 5–11 and 5–8, respectively (Fig. 1A; supplementary material Fig. S4).
To characterize mutant ciliary phenotypes associated with these deletion alleles, we incubated animals with the lipophilic fluorescent dye DiI, which allows readout of gross cilium structure and dendrite extension (Perkins et al., 1986). The dye is taken up by a subset of amphid and phasmid ciliated neurons that are either directly exposed to the environment (e.g. ASK, PHA) or are embedded in the cuticle (e.g. AWB). Failure of a neuron to take up DiI might be indicative of cilia structural or dendritic extension defects. nphp-2(gk653) mutants had a moderate dye filling defect (Dyf) in phasmid neurons, whereas amphid neurons appeared normal (Fig. 2A). nphp-2(nx101) and nphp-2(nx102) mutants exhibited phasmid Dyf phenotypes that were weaker than those observed in gk653 animals (0.59 in gk653 versus 0.83 and 0.85 in nx101 and nx102, respectively). For this reason, we used nphp-2(gk653) in the remaining experiments. Although no loss-of-function or null alleles of nphp-2 exist, we predict that these would have stronger Dyf phenotypes than the moderate severity observed in gk653 animals.
To determine whether cilia displacement defects are phasmid sensilla-specific, as dye filling suggests, or general to core ciliated neurons, we directly examined cilia placement in both amphid and phasmid sensilla. To label amphid and phasmid cilia, we coexpressed NPHP-1::CFP, a TZ marker, and CHE-13::YFP, a TF/cilia axonemal marker. In wild-type amphid sensilla, TZs clustered closely together and cilia were tightly grouped (Fig. 2C, left). Whereas both NPHP-1::CFP and CHE-13::YFP localized properly in nphp-2 animals, amphid TZs were dispersed over a larger area and their associated cilia were no longer tightly bundled (Fig. 2C, right). We quantified TZ spread by measuring the distance between the anterior-most and posterior-most TZs (Williams et al., 2011). We found that nphp-2 TZs in the amphid sensilla were dispersed over a distance three times longer than in the wild-type (data not shown). In wild-type phasmid sensilla, cilia and TZs in PHA and PHB neurons align (Fig. 2C, left). In nphp-2 phasmid sensilla, cilia and TZs were no longer aligned (Fig. 2C, right). Use of another TZ marker, NPHP-4::YFP, yielded similar results: correct subcellular localization of the markers but disorganized TZ regions in nphp-2(gk653) animals (supplementary material Fig. S5). In general, amphid sensilla TZ–cilia displacement was less severe than in phasmid sensilla (approx. 1–2 μm versus 9 μm, respectively), which might explain the lack of an apparent amphid Dyf phenotype.
Next, we explored the possible correlation between the Dyf phenotype and ciliary displacement. nphp-2 mutants expressing CHE-11::GFP, which marks the ciliary TZ and axoneme (Williams et al., 2011), were incubated with DiI. Displaced cilia that fail to make contact with the phasmid pore are Dyf, whereas properly positioned cilia dye fill. This implies that the phasmid Dyf phenotype probably stems from an anterior shift in placement of phasmid TZ and cilia (Fig. 2B). Because we could not use TZ spread as a measure of cilia displacement in the phasmid sensilla as we did in the amphid cilia, we instead quantified the displacement by measuring the distance from the TZ to the cell body (dendritic length). In nphp-2 mutants, the distance between the cell body and the TZ was significantly shorter than in the wild type (Fig. 3B), indicating that cilia were displaced towards the cell body. However, nphp-2 displaced phasmid cilia frequently, but not always, had a concomitant increase in length, allowing for contact with the phasmid pore. We conclude that nphp-2 affects placement of TZ and cilia in both amphid and phasmid sensilla.
Genetic analysis reveals a complex interaction network between C. elegans ciliopathic orthologs
Given the oligogenic nature of nephronophthisis (Guay-Woodford et al., 2000; Kuida and Beier, 2000; Wolf and Hildebrandt, 2011), we explored genetic interactions of nphp-2 with ciliopathy disease gene orthologs with TZ-localized products. First, we measured amphid and phasmid dye filling. Sensilla of the single mutants nphp-1(ok500), nphp-4(tm1925), mks-1(tm2705), mksr-1(ok2092), mksr-2(tm2452), mks-3(tm2547) and mks-6(gk674) are non-Dyf with the exception of the nphp-2 phasmid Dyf phenotype (Fig. 2A) (Williams et al., 2011). However, many double mutant combinations showed a synthetic dye filling defect (SynDyf), exhibiting a wide range of dye filling phenotypes (Table 1; supplementary material Fig. S6). Double mutants between an MKS gene (mks-1, mks-3, mks-6) and nphp-1 or nphp-4 are severely Dyf in both amphids and phasmids, indicating two parallel redundant pathways (an MKS pathway and a NPHP pathway) that act similarly in amphid and phasmid sensilla (Williams et al., 2008; Williams et al., 2011).
Examination of nphp-2 double mutants reveals differences between ciliogenic roles in amphid and sensilla. In phasmid sensilla, nphp-2 was SynDyf with nphp-1, nphp-4, mks-3 and mks-6, indicating that nphp-2 acts as a modifier of both the MKS and NPHP pathways (Table 1, bottom). In the context of amphid sensilla, nphp-2 was SynDyf with nphp-1, nphp-4, mksr-1 and mksr-2 and was nonDyf in double mutant combinations with the MKS family members mks-1, mks-3 and mks-6 (Table 1, top; supplemental material Table S3).
nphp-2 single mutants showed an increase in amphid TZ spread, a shortening of phasmid dendrite length and an increase in phasmid cilia length. Therefore, we measured these features, in addition to amphid cilia length, in several SynDyf double mutants, including nphp-2 nphp-4, mks-6; nphp-2 and mks-6; nphp-4, and in the single mutants nphp-4 and mks-6. Dendrites, cilia and TZs were visualized using the IFT-A marker CHE-11::GFP. In the amphid sensilla, single mutants nphp-4 and mks-6 did not exhibit a significant increase in TZ spread. In the double mutants, TZ spread was increased compared with their respective single mutants, indicative of errors in cilia placement (Fig. 3A). Phasmid dendritic length, an indicator of TZ displacement in phasmid sensilla, was significantly decreased in single mutants (73% of the average length in the wild type), implying mild TZ displacement. Double mutants had dendritic lengths significantly shorter than any of the single mutants, implying severe TZ displacement (Fig. 3B).
The dye filling results suggest that nphp-2 has a sensillum-specific role, although nphp-2 interacts with nphp-1 and nphp-4 (in both amphid and phasmid sensilla). By contrast, nphp-2 genetically interacts with different members of the MKS family in a sensillum-specific manner. The synergistic defects seen in both amphid TZ spread and phasmid dendritic length suggest genetic redundancy between nphp-2 and the NPHP and MKS pathways. Cilia length measurements in both amphid and phasmid neurons also showed synthetic defects: double mutants had significantly shorter cilia than single mutants, with the single exception of the phasmid cilia length comparison between nphp-4 and nphp-2 nphp-4 mutants (Fig. 3C). Curiously, the mks-6; nphp-2 double mutant has synthetic defects in both amphid TZ spread and cilia length, but mutants are not amphid SynDyf. This might indicate either that the pathway governing dye filling is separate from the pathway governing cilia formation and placement, or that cilia morphology and dye uptake do not always correlate as previously reported, e.g. a retracted cilium might still contact the now elongated amphid pore and DiI solution. That nphp-2(gk653) is a hypomorph and not a null allele might contribute to this complexity. Identification and analysis of a nphp-2 null allele could clarify genetic relationships with other ciliopathic orthologs. Together with previous work, these data support two primary TZ genetic pathways, an NPHP pathway involving NPHP-1 and NPHP-4 and an MKS pathway, with a role for nphp-2 as a sensilla-specific modifier.
nphp-2 and nphp-4 act redundantly to regulate cilia placement and IFT
The genetic interactions between nphp-2 and TZ-associated genes imply redundant roles and functions. We began exploring these by focusing on the association between nphp-2 and nphp-4. In contrast to either single mutant, and unlike most other double mutants, the nphp-2 nphp-4 double mutant was completely Dyf in both amphid and phasmid sensilla (Table 1). We used IFT-B polypeptide OSM-5::GFP and IFT-A polypeptide CHE-11::GFP to visualize ciliary TF–axonemal structures. In amphid sensilla, wild-type animals exhibited a pyramidal amphid channel ciliary bundle, with all TFs within 5 μm of each other (Snow et al., 2004) (Fig. 4A,B). In nphp-2 nphp-4 double mutants, no cohesive bundle was evident, cilia orientation within the bundle was disrupted and a portion of the amphid channel cilia appeared to be elongated, curly or stunted (Fig. 4E,F). In the amphid sensilla, it was not possible to identify and associate defects with specific cilia. However, in phasmid sensilla, PHA and PHB exhibited both elongated and stunted phenotypes.
Defects in dye filling, cilia length and cilia placement might be associated with defects in IFT. In amphid channel and phasmid cilia, the IFT machinery is composed of two complexes, IFT-A and IFT-B, which are transported by two anterograde kinesin-2 motors (slow kinesin-II and fast OSM-3) towards the cilia tip (Snow et al., 2004). In wild-type amphid channel cilia, the IFT-A associated kinesin-II heterotrimeric motor traverses only the middle segment, whereas the IFT-B associated OSM-3 homodimeric motor travels along the entire length of the axoneme. As the IFT particles are linked by the BBS protein complex, when a particle is on the middle segment, both motors are engaged and move along at a speed of 0.7–0.8 μm/second, whereas on the distal segment, only OSM-3 is engaged and the particle travels at the speed of 1.0–1.2 μm/second (Ou et al., 2005; Snow et al., 2004). To probe IFT function, we measured the velocities of IFT-B polypeptide OSM-5::GFP, IFT-A polypeptide CHE-11::GFP and homodimeric kinesin-2 OSM-3::GFP. In nphp-2 mutants, GFP-tagged OSM-5, CHE-11 and OSM-3 velocities in middle and distal segments were similar to those in the wild type (Table 2). No abnormal accumulation of IFT reporters was seen at the ciliary base, indicating that particles were being loaded onto the axoneme correctly (Fig. 4C,D).
Amphid cilia disorganization in the nphp-2 nphp-4 double mutant made it difficult to define the middle and distal ciliary segments. In nphp-2 nphp-4 double mutants, both IFT-A component CHE-11::GFP (1.02±0.20 μm/second) and IFT-B component OSM-5::GFP (1.03±0.17 μm/second) moved at speeds comparable to the velocity of the IFT particle in the wild-type distal segment (1.13±0.15 μm/second and 1.06±0.16 μm/second, respectively) (Table 2). We also observed that the IFT-B associated motor OSM-3::GFP traveled along the entire cilium at a velocity of 1.12±0.24 μm/second, similar to the distal segment velocity of OSM-3 in the wild type. The velocities of the IFT particle observed in nphp-2 nphp-4 mutants were similar to IFT-A and IFT-B velocities in kinesin-II mutants, in which OSM-3 is the only motor associated with the IFT particle (Pan et al., 2006). Previously, it was reported that OSM-3 overexpression in nphp-4 mutants results dendritic accumulation of OSM-3 (Jauregui et al., 2008); we also observed this in nphp-2 nphp-4 animals overexpressing OSM-3. Surprisingly, OSM-3 overexpression suppressed the truncated cilia phenotype in the phasmid sensillum of nphp-2 nphp-4 double mutants (average length 7.40±0.62 μm). This rescue was not concordant with a rescue of the Dyf phenotype (data not shown). We conclude that within amphid cilia, nphp-2 does not modify IFT complex localization, velocity or loading onto the axoneme. However, in nphp-2 nphp-4 double mutants, OSM-3 is the primary motor driving IFT, perhaps due to the uncoupling of the heterotrimeric kinesin-II motor from the IFT complex.
nphp-2 is not required for proper localization of NPHP-1, NPHP-4 or B9 proteins
To first tested whether NPHP-2 might influence the localization of TZ proteins and observed that nphp-2 is not essential for NPHP-1 or NPHP-4 TZ localization (Fig. 2; supplementary material Fig. S4). To determine whether NPHP-2 is required for the localization of proteins belonging to the second TZ-associated gene pathway, we examined B9 protein localization in nphp-2 mutants. MKS-1::GFP and MKSR-2::GFP localized to the TZ in wild-type animals (Fig. 5A,G), nphp-2 single mutants (Fig. 5B,H) and nphp-2 nphp-4 double mutants (Fig. 5C,I), although several TZs were improperly positioned within the amphid sensillum. In wild-type animals, MKSR-1::GFP localized to both the dendritic tip and the TZ (Bialas et al., 2009; Williams et al., 2008) and was likewise unaffected in nphp-2 single and nphp-2 nphp-4 double mutants (Fig. 5D–F). We conclude that nphp-2 is not required for the proper TZ localization of NPHP pathway proteins (NPHP-1 and NPHP-4) or B9 proteins of the MKS pathway.
INVS (also known as NPHP2), which is mutated in infantile NPHP type II, localizes to the ciliary proximal segment or Inv compartment and is a genetic modifier of polycystic kidney disease. Here we show that the C. elegans ortholog of inversin, NPHP-2, localizes to the C. elegans ciliary middle segment, similar to the Inv compartment of mammalian cilia, and plays a role in TZ placement. We also find that the genetic interactions between nphp-2 and the NPHP and MKS pathways in C. elegans reflect the physical interaction between NPHP and MKS modules in mammalian systems (Sang et al., 2011).
Use of multiple bioinformatic analyses and examination of expression and localization patterns establishes C. elegans Y32G9A.6 (nphp-2) as the ortholog of INVS (Blacque et al., 2005). INVS encodes an IQ domain, which interacts with calmodulin in a Ca2+-dependent manner (Morgan et al., 2002). In contrast to inversin, NPHP-2 is predicted to contain a Ca2+-binding EF hand, which might function in a similar manner as the IQ domain. INVS encodes multiple isoforms (Nürnberger et al., 2002; Ward et al., 2004); likewise, we find that C. elegans nphp-2 encodes two isoforms. Both NPHP-2S::GFP and NPHP-2L::GFP localize to the middle segment, are excluded from the TZ and rescue the nphp-2 nphp-4 SynDyf defect, indicating that both isoforms function in TZ placement. The additional 22 amino acid residues in NPHP-2L appear to play a role in restricting NPHP-2L localization to cell-body puncta and the ciliary middle segment, but the functional difference between NPHP-2S and NPHP-2L remains unknown.
The defects present in nphp-2(gk653) animals might stem from one of several possibilities, including impingement on dendritic length, TZ placement or TZ protein localization. First, an alteration of dendritic length could lead to the observed TZ displacement. nphp-2 mutants have significantly shorter phasmid dendrites than wild-type animals, as do several TZ mutants. Double mutants have even shorter dendrites, which is further evidence of synergistic interactions between the genes. Though phasmid ciliogenesis is currently uncharacterized, these ciliary proteins seem to be necessary for phasmid dendritic extension, which might be mediated through ciliary attachment to the sheath and socket cells near the phasmid pore, as suggested by Williams and colleagues (Williams et al., 2010). Second, nphp-2 might affect TZ placement at the end of the dendrite. This placement might be mediated by the planar cell polarity (PCP) pathway, which, in zebrafish, is upregulated by inversin (Simons et al., 2005). The role of the PCP in the ciliated nervous system of C. elegans is unexplored, and is an exciting avenue for future research. Third, nphp-2 mutation might interfere with correct localization of TZ proteins. This does not seem to be the case for localization of NPHP-1, the B9 proteins, the IFT-A polypeptide CHE-11 and the IFT-B polypeptides CHE-13, OSM-3 and OSM-5. However, in mammalian cells inversin is important for localizing NPHP3 and NPHP9. Likewise, NPHP-2 might have a role in the localization of other ciliary proteins in C. elegans (Otto et al., 2003; Shiba et al., 2010).
A crucial question is raised by the middle segment localization of both isoforms of NPHP-2: how does a protein enriched in the middle segment and excluded from the TZ affect TZ placement? The middle segment forms after the TZ is placed at the tip of the dendrite during the first phase of ciliogenesis (Williams et al., 2011). Two distinct possibilities might account for this apparent paradox: first, that NPHP-2 might have a non-middle segment function and, second, that cilia placement might follow formation of the middle segment. In adult animals, NPHP-2S has a high non-middle segment concentration and is observed in the cell body, dendrite and dendritic tip; this population of NPHP-2S might modulate cilia placement. Alternatively, NPHP-2 localization could be dynamically regulated during ciliogenesis, with a transient non-middle segment population modulating cilia placement. The genetic interactions between nphp-2 and TZ-associated genes in C. elegans and the physical interactions between NPHP-2 and NPHP and MKS proteins in mammalian systems support this non-middle segment role for NPHP-2. If cilia begin forming before the dendrite has fully extended, the cilium could guide its own placement through interactions with the surrounding cellular matrix. This possibility is supported by evidence that TZ placement defects are linked to defects in attachment of the cilium to the sheath cell (Williams et al., 2010). In addition, SynDyf double mutants showed ectopic dye filling of the amphid and phasmid sheath cells, which has previously been linked to malformed cilia (Ohkura and Bürglin, 2011).
A handful of other proteins share an Inv compartment middle segment localization pattern with NPHP-2 in C. elegans: the membrane-associated GTPase ARL-13 (which modulates association of IFT-A and IFT-B), the histone/tubulin deacetylase HDAC-6 and the IFT-A-associated kinesin-II motor (Otto et al., 2003; Shiba et al., 2010). In zebrafish, inversin ciliary localization is mediated by Fleer (orthologous to C. elegans DYF-1) (Zhao and Malicki, 2011), an IFT-B polypeptide and a regulator of tubulin glutamylation (Ou et al., 2005; Pathak et al., 2007). ARL-13, HDAC-6 and Fleer might play a role in IFT regulation via possible tubulin post-translational modifications (Li et al., 2010). In addition, mutations of the gene encoding tubulin deglutamylase, ccpp-1, cause ciliary degeneration, as mirrored by a progressive Dyf phenotype (O'Hagan et al., 2011); OSM-3 appears to be the primary IFT motor functioning in ccpp-1 mutants. The signaling and tubulin-modifying roles of these components suggests that the Inv compartment might link cell signaling, tubulin modification and regulation of IFT. Combined, these observations lead to a quandary: does middle segment localization of these components regulate tubulin modification, or vice versa?
We found that amphid and phasmid cilia use modified gene networks (Fig. 6) (Perkins et al., 1986). Although the evolutionarily conserved IFT machinery is required for cilia formation, ciliary specialization can be shaped by sensillum- or tissue-specific modifications. Mammalian TCTN1 has tissue-specific roles: it is essential for axonemal extension in nodal and neural tube cilia but is only necessary for Arl13b localization in the cilia of the notochord and early gut epithelia (Garcia-Gonzalo et al., 2011). Mammalian MKS1 also acts in a tissue-specific manner and is essential for formation of nodal and neural tube cilia, but not bile duct and lung cilia (Weatherbee et al., 2009). Sensillum and cell-type specificity is also evident in C. elegans: IFT in the cilia of AWB neurons is modified such that the kinesin motor OSM-3 moves independently of kinesin-II and is not needed to build distal segments (Mukhopadhyay et al., 2007). Additionally, specifically in the cilia of male CEM neurons, the kinesin-3 motor KLP-6 regulates kinesin-2-driven IFT (Morsci and Barr, 2011). Here, we show that nphp-2 SynDyf phenotypes vary in a sensillum-specific manner. nphp-2 interacts with nphp-1 and nphp-4 in both amphid and phasmid neurons, but interacts with MKS family members in a sensillum-specific manner. We interpret this as evidence for a role for nphp-2 as a modifier of NPHP and MKS pathways in these sensilla.
Much recent work has been performed to elucidate the physical and genetic interactions between ciliopathy-related genes. A mammalian proteomics screen for physical interactors of cystoproteins revealed three distinct physically interacting modules: a NPHP1–NPHP4–NPHP8 module, a MKS1–MKS6–TCTN2 module and a NPHP5–NPHP6–ATXN10 module (Sang et al., 2011). The former two modules correspond to the C. elegans genetic interaction-based NPHP (involving NPHP-1 and NPHP-4) and MKS pathways, though the NPHP5–NPHP6–ATXN10 module does not appear to be conserved. NPHP-2 acts as a nexus between these three modules, which agrees with our findings that nphp-2 might act as a modifier of both the NPHP and MKS pathways (Sang et al., 2011). In addition to the genes examined here, Williams and co-workers showed that mks-5, a C2 domain-encoding gene orthologous to mammalian RPGRIP1L, was SynDyf with nphp-4, mks-6 and mksr-2 in both amphids and phasmids. This led them to link mks-5 to both NPHP and MKS pathways, similarly to the treatment of nphp-2 here (Williams et al., 2011).
In summary, we find that nphp-2 acts as a genetic modifier of the NPHP and MKS pathways in C. elegans and modulates sensillum-specific regulation of ciliary positioning. These results complement and extend our knowledge of the roles of ciliopathy genes in ciliogenesis and cilia placement, and integrate a wide range of ciliopathic genes into a single model.
Materials and Methods
General molecular biology methods
Standard protocols were used for all molecular biology procedures. PCR amplification was used for genotyping and building transgenic constructs using the following templates: C. elegans genomic DNA, cDNA or prebuilt constructs. High fidelity LA Taq (TaKaRa Bio Otsu, Shiga, Japan) or Phusion high fidelity DNA polymerase (ThermoFisher Scientific, Vantaa, Finland) were used for amplification of DNA for constructs. Sequencing reactions were performed on site and analyzed by the SEBS DNA sequencing facility (Rutgers University, Piscataway, NJ). PCR primer and construct sequences are available upon request.
DNA and protein sequence analysis
BLAST (Altschul et al., 1997) was used for identification of gene orthologs in C. elegans. Human protein sequence information was provided by NCBI, and C. elegans gene and protein sequence information was provided by Wormbase. SAGE data was provided by Wormbase (Release WS221). Structural and domain predictions of gene products were by MotifScan (Hau et al., 2007). Percent identity and percent similarity were computed by MatGat 2.02 software (Campanella et al., 2003). Splice predictions in gk653 deletion mutants were computed using Genemark.hmm (Lomsadze et al., 2005). ApE 1.17 was used for sequence manipulation.
C. elegans INVS homology search
BLAST (Altschul et al., 1997) homology searches of inversin (also known as nephrocystin-2) led to the identification of five possible C. elegans orthologs, unc-44, mlt-4, T28D6.4, F36H1.2 and Y32G9A.6 (in order of decreasing similarity). The unc-44 promoter does not contain a palindromic X-box sequence, which is common to many ciliary genes, and the UNC-44 protein contains a DEATH domain and a zona pellucida UNC-5 domain, neither of which are present in inversin (Blacque et al., 2005; Efimenko et al., 2005). SAGE tagging data also indicates that unc-44 transcripts are not enriched in ciliated neurons [Wormbase Release WS221 (Blacque et al., 2005)]. Likewise, the mlt-4 promoter does not contain an X-box, and the entire length of the MLT-4 peptide encodes only ankyrin repeats. T28D6.4 promoter does contain an X-box, but the gene product is predicted to contain only ankyrin repeats and an SH3 domain, not present in inversin. F36H1.2 has no X-box and the predicted gene product has no similarities to inversin beyond the common ankyrin repeat motif and a possible weak nuclear localization signal. As described in the results, Y32G9A.6 has an X-box and shares many predicted domains with inversin, including a bipartite nuclear localization signal and a D-box ubiquitylation site, with similar overall domain organization. Performing a reverse protein BLAST homology search on MLT-4 and Y32G9A.6 yielded a match for inversin, whereas UNC-44, T28D6.4 and F36H1.2 did not.
A similar in depth search was performed for orthologs of NPHP3 (also known as MKS7). Only C. elegans KLC-2 (kinesin light chain 2) had meaningful homology to NPHP3. KLC-2 has a C-terminal domain with medium homology to NPHP-3. klc-2 has a transcript enriched in ciliated neurons (Blacque et al., 2005), but does not possess an X-box (Efimenko et al., 2005). In addition, NPHP3 does not appear in the top results in a reciprocal BLAST search for KLC-2.
RT-PCR and transcript sequencing
Total RNA from animals was isolated using Trizol reagent (Invitrogen, Carlsbad, CA). Reverse transcriptase with oligo-dT was used to isolate stable mature mRNA. Transcripts were ligated into pGEM-T-Easy plasmid (Promega, Fitchburg, WI) and transformed into DH10-β competent cells. DNA from transformed colonies was amplified with Phusion, as above, using the following primer sets: 5′-ATGTCCCACACGCTGATCGAAGCATTAGACGATGAG-3′ with 5′-GTCGACGGTCTTCTTTGTTTCTTCTGTTCAGCTTTTAACTC-3′ to amplify the first half of the transcript and 5′-GTCGACTAGCGACGGAATTGTGGAAACCGAGA-3′ with 5′-AAGGAACAGGTGCCTATGCGTGTCAAGGCAATC-3′ to amplify the second half of the transcript. These amplicons were then sequenced.
Total RNA from animals was isolated using Trizol reagent. A 7900HT Sequence Detection system (Applied Biosystems, Foster City, CA) was used. Levels of nphp-2 mRNA were normalized to either actin or to the RNA polymerase large subunit. nphp-2 mRNA was detected using the following primer sets: 5′-TCGACTAGCGACGGAATTGTG-3′ with 5′-GTTTTGGAGCGTTTGAATCGG-3′ and 5′-AAAAAGGAGCTGGAGGCACTG-3′ with 5′-CAAATCGGAGATTGGCGGTA-3′. The actin gene, act-1, was used as a control and was detected using the following primer set: 5′-CCATTGTCGGAAGACCACGTC-3′ with 5′-AGTTGGTGACGATACCGTGCTC-3′. The DNA polymerase gene ama-1 was also used as a control and was detected using the following primer set: 5′-GGATGGAATGTGGGTTGAGA-3′ with 5′-CCGAGTAGTTTTTGCGAAGG-3′.
Strains and maintenance
All strains were cultured at 20°C, unless otherwise noted, under standard conditions (Brenner, 1974). Deletion alleles such as gk653 were outcrossed to him-5(e1490) at least four times. nphp-2(nx101) and nphp-2(nx102) alleles are deletions of ankyrin repeats 5–11 and 5–8, respectively, generated by imprecise Mos1 excision and outcrossed once to him-5(e1490). The mks-6 allele used, gk674, deletes portions of a second gene, xpa-1, but has no discernible effects on the function of XPA-1 (Williams et al., 2011). Strains used in this study appear in supplementary material Table S4, organized by the figure in which they first appear.
Worms were imaged using standard C. elegans slide mounts and either a Zeiss Plan-APOCHROMA 63× 1.4 oil DIC or 100× 1.4 oil DIC objectives on a Zeiss Axioplan 2 microscope (Zeiss, Oberkochen, Germany) with a Cascade 512B (Photometrics, Tucson, AZ) digital camera, or on a Zeiss Imager.D1M with a Retiga-SRV Fast 1394 digital camera (Q-Imaging, Surrey, BC, Canada). Fluorochromes were eGFP, VenusYFP, tdTomato, or CFP. IFT motility was recorded and kymographs and particle movement rate were measured using Metamorph software (version 184.108.40.206; MDS Analytical Technologies, Sunnyvale, CA). Strains were synchronized by picking L4 worms, culturing at 15°C overnight and imaging within 24 hours. Worms were anesthetized in a drop of M9 containing 50 mM muscimol, transferred to an agarose mount slide and imaged immediately. AutoDeblur software (version 1.4.1; Media Cybernetics, Bethesda, MD) was used for 3D blind deconvolution of image stacks. The amphid channel middle segment is the region of the cilium between the transition zone and the point where amphid cilia align; the distal segment is the region of the cilium past this point (Snow et al., 2004). Figures and diagrams were created with Adobe Photoshop CS3 (version 10.0; Adobe Systems, San Jose, CA) and Adobe Illustrator CS3 (version 13.0.0; Adobe Systems). Cilia length, dendrite length and TZ spread measurements were analyzed using ANOVA, followed by Tukey's post-hoc test.
Dye filling assays
For staining of cell bodies and cilia for evaluation of NPHP-2 localization, worms were incubated with 2.5 μg/ml DiO (2.5 mg/ml dimethyl formamide stock, diluted 1:1000 in M9) (Invitrogen, Carlsbad, CA) for 30 minutes, rinsed three times and then allowed to recover on a seeded plate for 1 hour before imaging.
For scoring dye uptake as an indicator of amphid and phasmid ciliary structural defects, we used a dye filling protocol modified from Tong and Bürglin (Tong and Bürglin, 2010). Staged 1-day-old young adult hermaphrodites were washed off plates in M9 and rinsed twice. Worms were then incubated in 40 μg/ml DiI (2.5 mg/ml dimethyl formamide stock, diluted 1:1000 in M9) (Invitrogen) for 1 hour in the dark, rinsed twice in M9 and allowed to recover on a seeded plate for 1 hour in the dark. Animals were anesthetized using 10 mM levamisole and mounted on a slide. Worms were scored as fraction of amphid or phasmid cell bodies stained by identifying and scoring individual neurons and averaging together the 12 amphid neurons (ASK, ADL, ASI, AWB, ASH and ASJ neurons on both left and right sides) or the four phasmid neurons (PHA and PHB neurons on both left and right). These were compared using the non-parametric Kruskal–Wallis test, followed by Dunn's multiple comparison test. GraphPad Prism (version 5.01; GraphPad Software, La Jolla, CA) was used for all statistical analysis and bar graph creation.
We thank the Barr Laboratory for constructive criticism on the manuscript and the Rutgers C. elegans community for ongoing intellectual input. Some strains were provided by Bradley Yoder (University of Alabama, Birmingham, AL), the C. elegans Genetics Center (funded by the National Institutes of Health), the Japanese National BioResources Project and the C. elegans Gene Knockout Consortium.
This work was supported by grants from the March of Dimes and the National Institutes of Health (NIH)/National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) [grant number DK074746] to M.M.B. S.R.F.W.-P. was supported by the Busch Graduate Fellowship Program. M.R.L. is funded by the March of Dimes, the Canadian Institutes of Health Research (CIHR) [grant number MOP-82870] and acknowledges a Michael Smith Foundation for Health Research (MSFHR) senior scholar award. Deposited in PMC for release after 12 months.