The nephronophthisis-related gene ift-139 is required for ciliogenesis in Caenorhabditis elegans

Defects in cilia cause a spectrum of diseases known as ciliopathies. Nephronophthisis, a ciliopathy, is the most common genetic cause of renal disease. Here, I cloned and analysed a nephronophthisis-related gene ift-139 in Caenorhabditis elegans. ift-139 was exclusively expressed in ciliated neurons in C. elegans. Genetic and cellular analyses suggest that ift-139 plays a role in retrograde intraflagellar transport and is required for cilia formation. A homologous point mutation that causes ciliopathy disrupted the function of ift-139 in C. elegans. ift-139 is an orthologue of human TTC21B, mutations in which are known to cause nephronophthisis 12 and short-rib thoracic dysplasia 4. These results suggest that ift-139 is evolutionarily conserved and fundamental to the formation of cilia.

encodes the IFT139 subunit in humans 27 . IFT139 was originally identified as a component of the IFT complex in Chlamydomonas 10,28 . Because of the homology of the protein encoded by ZK328.7 with other IFT139 protein family members, the gene was renamed ift-139. Based on the predicted sequence, I performed polymerase chain reaction (PCR) from wild-type cDNA and cloned ift-139. While it is predicted that ift-139 encodes two isoforms, the longer isoform which encodes a 1324-amino acid protein was obtained by 5′ UTR and 3′ UTR primer pairs. IFT-139 has at least six TPR domains, as revealed with the SMART algorithm 29 . The feature is similar to other IFT139 family members.

IFT-139 is expressed in ciliated neurons and localises to basal bodies and cilia.
To assess the cellular expression of ift-139, an 800-bp fragment upstream of the start codon, spanning the putative promoter region, was cloned into a Green Fluorescent Protein (GFP) expression vector and injected into wild-type nematodes. Transgenic animals were established and observed under the fluorescent microscope. GFP expression was observed in ciliated sensory neurons in both head and tail regions (Fig. 1b).
Next, the open reading frame encoding ift-139 (1324 amino acids) was fused with GFP and expressed from the promoter described above (Fig. 1c,d). IFT-139::GFP was observed in ciliated sensory neurons. In the amphid neurons, the GFP signal was diffuse in dendrites and concentrated to the base of cilia (Fig. 1c,c'). Moreover, the cilia signal was stronger than dendrite. In phasmid PHA and PHB neurons, most of the GFP signals were observed at the base of cilia and in the cilia (Fig. 1d). Phasmid dendritic signals were relatively weak compared to amphid neurons. To observe the localisation of IFT-139 at single-cell resolution, IFT-139::GFP was expressed from the PHA neuron-specific promoter, srg-13 promoter (Fig. 1e). Once again, the IFT-139::GFP signal was localised to the base of cilia and in the cilia of PHA neuron (Fig. 1f). These data suggest that IFT-139 may play a role in cilia.
ift-139 is required for the formation and function of cilia in C. elegans. I searched WormBase (www.wormbase.org) and found that gk477 is a deletion mutant of ift-139 in which the first exon of the gene is deleted (Fig. 2a). Though it is predicted that the ift-139 genome encodes two isoforms, both isoforms share the promoter and the first exon, suggesting that both isoforms are deleted in gk477. To test whether gk477 affects the expression of ift-139, reverse transcription polymerase chain reaction (RT-PCR) was performed using primers that can detect both isoforms (Fig. 1a). Results of RT-PCR showed that the expression of ift-139 was under the detectable level in gk477 (Fig. 2b), suggesting that gk477 is a null mutant.
Cilia morphology was investigated in nematodes carrying the ift-139(gk477) mutation using the well-established and widely used cilia marker, mnIs17 30 . This marker stably expresses OSM-6::GFP in ciliated neurons under the osm-6 promoter. It was found that nematodes carrying the ift-139(gk477) mutation had abnormal cilia whose length were significantly shorter than wild type (Fig. 2c,d; wt: 6.4 ± 0.7 μ m; ift-139(gk477): 3.9 ± 0.7 μ m; mean ± standard deviation; n = 27 cilia; p < 0.01, t-test). Cilia were also more bulged in ift-139(gk477) than in wild-type animals (Fig. 2d,d'). To test that this phenotype was caused by the ift-139(gk477) mutation, and to determine whether the effect was cell autonomous, wild-type ift-139 cDNA was expressed specifically in ciliated neurons of the ift-139(gk477) mutant from the osm-6 promoter. The activity of both long and short isoforms of ift-139 was tested. As a result, the cilia morphology defect was rescued by expressing ift-139a (long isoform) but not ift-139b (short isoform) (Fig. 2e,f). Thus, I focused on the long isoform in this manuscript in the following experiments. To confirm that the morphological defect revealed by the GFP marker reflects abnormalities in vivo, phenotypes that are induced by ciliary defects were investigated. First, behaviour of ift-139(gk477) was observed. Single worms were placed on OP50 feeder bacteria and incubated overnight. While wild-type worms moved all over the feeder bacteria, ift-139(gk477) movement was restricted to a narrow area (Fig. 3a,b, 38% compared with wt). Nevertheless, unlike unc mutants, the speed of worm motility was not changed significantly when they were moving on the agarose plate. The phenotype is similar to those observed in ift-74 and ift-81 mutants, in which cilia are disrupted 31 . Next, dauer formation, dye-filling and mating behaviours were examined; these are known to require normal cilia function and morphology [32][33][34] . Consistent with the abnormal cilia morphology, defects in all of these processes were observed in ift-139(gk477) nematodes (Fig. 3d). In the ift-139-expressing strain (Fig. 2e), normal mating behaviour was recovered. Furthermore, previous studies have shown that C. elegans is attracted to diacetyl, a volatile substance, and that cilia are required for this chemotaxis 35 . We therefore calculated the chemotaxis index to diacetyl as described, comparing wild-type and ift-139(gk477) mutants 36 . As expected, chemotaxis was impaired, with the chemotaxis index of the ift-139(gk477) mutant significantly lower than wild type (Fig. 3e). Collectively, these phenotypes indicate that cilia morphology and function require ift-139.

Human TTC21B can partially rescue ift-139(gk477). ift-139 is the nematode orthologue of TTC21B.
To test whether the function of ift-139 is evolutionarily conserved, the human TTC21B gene was cloned and fused with the osm-6 promoter. Then, the vector was injected into ift-139(gk477) mutants. As a result, expressing human TTC21B could partially rescue cilia morphology; cilia length was longer in ift-139(gk477) mutants in the presence of human IFT139 than without, but shorter than wild-type cilia (Fig. 4a-c). The rescued cilia was more wild-type like morphology, but still more bulged than wild type. Statistical analysis supports the observations (Fig. 4e).
A ciliopathy mutation disrupts the function of ift-139. It has been shown that TTC21B mutations cause autosomal recessive ciliopathies such as NPHP12 and short-rib thoracic dysplasia 4 23,24,27 . Most TTC21B mutations causing ciliopathies are deletions 27 , although some point mutations have been described, including a point mutation leading to an L795P substitution in human TTC21B. The L795 residue is conserved in C. elegans IFT-139 (L810). We postulated that a similar substitution in this conserved amino acid would result in ciliary defects in C. elegans. Therefore, we introduced a transgene encoding ift-139 with the corresponding mutation, L810P, into C. elegans ift-139(gk477) and assessed its effect on the activity of nematode IFT-139. I tested the long isoform of ift-139 in this experiment. While expression of wild-type ift-139 could recover the cilia morphology of ift-139(gk477) mutants, expression of ift-139(L810P) could not (Fig. 4b,d,e). The morphology of cilia was still bulb like and could not be discriminated from ift-139(gk477) in IFT-139(L810P)-expressing worms. These rescue ). * p < 0.01, t-test, n = 10. Note that ift-139(gk477) mutants move in a narrow area compared with wild type. Bars, 1 cm. (d) Gross phenotypes in ift-139 mutants were compared with wild type, osm-3, and dyf-5 mutants. daf, dauer formation, dyf, dye filling and mating: male mating. wt means the phenotype was comparable to wt, -means the defects in the phenotype. (e) Chemotaxis to diacetyl was observed and chemotaxis index was calculated as described in the materials and methods. Mean ± S.D. n = 10 experiments. * p < 0.01, t-test. experiments by human TTC21B and IFT-139(L810P) suggest that the structure and function are well, but not perfectly, conserved between human TTC21B and nematode IFT-139.

IFT-139 movement in cilia.
The motility of IFT-139 was observed in phasmid cilia. The long isoform of IFT-139 was expressed as a GFP fusion as described above (Fig. 1c,d) and the signal was observed by spinning disk confocal microscopy. Many IFT-139::GFP-positive particles were bidirectionally moving in the cilia (Fig. 6a,b, and Supplemental Movie S1). The speed of anterograde particles was measured at 0.72 ± 0.11 μ m/s and 1.31 ± 0.11 μ m/s in middle and distal segments, respectively (Fig. 6a-c). The speed of retrograde transport was 1.21 ± 0.32 μ m/s and 1.11 ± 0.31 μ m/s in middle and distal segments, respectively (Fig. 6d). These biphasic and monophasic transport parameters in anterograde and retrograde transport are almost identical to those previously found for the IFT-particle::GFP 37,38,41 . To analyse the effect of disease-associated mutations on the motility of IFT-139, IFT-139(L810P) was fused with GFP and introduced into wild-type worms. Interestingly, in both anterograde and retrograde transport, the motility of IFT-139(L810P) was comparable to that of wild-type IFT-139 in wild-type worms (Fig. 6c,d). However, the mutant protein could not rescue ift-139(gk477), suggesting that IFT-139(L810P) is normally incorporated into the IFT complex but is not functional in vivo.

Discussion
The biochemistry of IFT is well studied in Chlamydomonas 1,10,11 , where the IFT complex is composed of at least 16 protein subunits. Most subunits are conserved among eukaryotes with cilia. The IFT139 subunit is important because mutations in TTC21B that encodes human IFT139 lead to ciliopathies such as NPHP12 and short-rib thoracic dysplasia 4 23,24,27 . While previous studies have identified and characterised most IFT subunits in the nematode, whether or not the C. elegans IFT complex contains and requires the IFT139 subunit remained uncertain. In this paper, I identified and characterised nematode ift-139 and showed that the IFT139 subunit is essential for the formation of normal cilia. The database predicted that ift-139 gene encodes two isoforms. However, only the long isoform (isoform a), but not the short isoform (isoform b), was essential for the intraflagellar transport (Fig. 2d-f). One possibility is that the short isoform is a prediction error. It is difficult to confirm the presence of the short isoform because of the gene structure (Fig. 2a). Another possibility is that the short isoform has roles that are different from the intraflagellar transport. Northern blotting to separate two isoforms and 3'-RACE experiments to determine the transcriptional termination would be needed to confirm the presence of the short isoform.
Genetic data suggest that IFT-139 works together with the products of ifta-1 and che-3. ifta-1 and che-3 encode IFT122 and cytoplasmic dynein-2 heavy chain, both proteins that are required for retrograde IFT 39,41 . Thus, genetic data suggest that IFT-139 is a component of the IFT-A subcomplex in C. elegans and required for retrograde transport. These genetic data are consistent with the biochemical data in Chlamydomonas that shows that IFT139 belongs to IFT-A 42 . Interestingly, the length of cilia is significantly longer in ift-139 mutants when a second mutation is introduced in klp-11 ( Fig. 5g). There are two possible explanations for how the klp-11 mutation recovers cilia length in ift-139 mutants. One possibility is that the short and bulged cilia are a result of a traffic jam induced by the ift-139 mutation. As klp-11 is an anterograde motor protein, loss of klp-11 reduces the amount of anterograde transport, leading to the reduction of the traffic jam. This is supported by the observation that the amount of OSM-6::GFP is reduced in klp-11 cilia 38 (Fig. 5c). This idea is also consistent with the genetic data suggesting that IFT-139 is involved in retrograde transport. While it is not exclusive, another possibility is that KLP-11 transports some factors that negatively regulate cilia length 43,44 . This hypothesis is supported by the observation that klp-11 mutant males have longer male-specific cilia 45 .
Disease-associated mutations often give valuable insights into the function of proteins. IFT-139(L810P) is the corresponding mutation to TTC21B(L795P) that causes ciliopathy. Previous studies in zebrafish have showed that TTC21B(L795P) is normally localised to the base of cilia 23 . We could observe the motility of this mutant protein in C. elegans and found that the IFT-139(L810P) protein is incorporated into the IFT particles and moves normally and bidirectionally in cilia, but cannot rescue the ift-139 mutant (Fig. 6). Together, these results suggest that this residue is not required for the binding of IFT139 to the IFT complex, but the activity of IFT139 is disrupted by this mutation. It is possible that this residue is required to bind to a more peripheral subunit, or IFT cargo molecules such as cilia tubulin 46 . Further investigations are required to fully analyse and test these molecular interactions; however, biochemical analysis of IFT has not been established in C. elegans. Nevertheless, our genetic data will inform future biochemical studies in Chlamydomonas and mammals to reveal both how IFT139 is incorporated into, and its role in, the IFT complex. Methods Strains and Genetics. Caenorhabditis elegans was cultured and maintained according to the standard protocol 12 . Genotypes were determined by PCR and genomic sequencing in deletion mutants and point mutations, respectively. Primer sets used to detect mutants were as follows: ift-139(gk477): 5′ -GGCAGTAGCATACGAAGTGTAAGGAG-3′ , 5′-CTCTCCAGTGCTCCACATCCTTGTG-3′ and 5′ -GCTGTATACACCACCTAAAGCCTACC-3′ . klp-11(tm324): 5′ -GCTCACACATTGACATAGGCCGTCG-3′ , 5′ -GAGACACCACTATCGGCACCAGATG-3′ and 5′ -GGCTTCTTATTGTTGAAAACTTCATCCCTCG-3′ . ifta-1(gk1004): 5′-GCACTGCTTCAACTCTTCATATGGG-3′ , 5′ -CCCTTCATCATCGTCATCATTTCGAAT C-3′ and 5′ -AGTGCCGTGCGGAACCCAGTAGATG-3′ . che-3: 5′ -TTTGCTTGGACTTGCTTGGAGCTTG-3′ and 5′-GTACTTCTGTTCAATGGATTCTTGCCTACC-3′ for PCR and 5′ -TCATATTTGGGCGATCCT TTGGTCTC-3′ for sequencing. Strains are described in Supplemental Table S1.
Construction of ift-139 promoter::gfp and cloning of ift-139 cDNA. The ift-139 promoter, defined as the region between the last exon of the next upstream gene and the start codon of ift-139, was amplified from genomic DNA extracted from wild-type worms Primer sets: 5′ -CTGCATGCTCCATACTCGAATACTGAG-3′ and 5′-atGGCGCGCCAAATTGAGATATCAGCAATAAAAATAT-3′ . SphI and AscI sites were added to ligate the PCR product to the pSM vector. A C. elegans cDNA library was a generous gift from K. Mizumoto 47 . ift-139 cDNA was amplified from the library by 2 step. Firstly, PCR was performed using 5′ UTR and 3′ UTR primer pairs and then second PCR was performed and cloned into the pSM vector using NheI and KpnI sites. PCR primers were as follows: PCR #1 was 5′ -GTCAAAGCAATATTTTTATTGCTGATATCTC5′ -TCACAAATGAAAGTATTCCCCATTTAC-3′ and -3′ . PCR #2 was 5′ -atGCTAGCaaaaATGGATTCCGAATCTGACGATAATCCAAATG-3′ and 5′ -atgcGGTACCcc AGTTCTGATTAGAGCTTTCGCTTTATCCAT-3′ .
Chemotaxis assay. Well-fed adult worms were washed four times with phosphate-buffered saline. A 10-cm Petri dish containing a standard assay surface (2% agar, 5 mM K 2 PO 4 , 1 mM CaCl 2 and 1 mM MgSO 4 ) was used for the assay. Two X marks 180° opposite each other were made near the edge of the plate. 1M NaN 3 was absorbed to these two points. After allowing the NaN 3 to dry, approximately 50 worms were pipetted onto the plate. Quickly, 1/100 diluted diacetyl or ethanol were placed on each X mark and the lid was closed. After 1 hour, the number of worms was counted and the chemotaxis index was calculated as follows: (number of worms in diacetyl zone − number of worms in control zone)/(total number of worms).
RT-PCR assay. Total RNA was extracted with TRIzol reagent (Thermo Fisher Scientific, MA, USA).
Male mating assay. Males were produced by heat shock procedure. 10-15 plates that have 10 L4 worms were placed in 30 °C for 8 hours and then transferred to 20 °C. 4 days later, males were picked up. Typically, more than 20 males were obtained from all genotypes. For mating assay, 3 wt hermaphrodites were mixed with 15 males that were labelled by mnIs17. Mating success was judged by the presence of mnIs17 in the F1 generation. When no F1 animals with the mnIs17 marker was found, I considered the strain exhibited defects in male mating. The phenotype was confirmed by repeating the same procedure. dauer-formation assay. Plates were starved for 1 week and then worms were observed using a Stemi 508 stereo microscope (Carl Zeiss, Jena, Germany). Three independent plates were prepared for each strains. Dauer was judged by the lack of pharyngeal pumping. When I could not find any dauer worms from 3 independent plates, I considered the strain dauer defective (daf). When the strain is judged as daf in the first assay, the phenotype was confirmed by repeating the same procedure. dye-filling assay. DiO (Thermo Fisher Scientific) was dissolved in DMSO (Sigma) at 2 mg/ml and stocked at − 30 °C. Well-fed worms were collected from a 60-mm plate with 1 ml M9 buffer and pelleted at 1400 g for 2 min. Worms were resuspended in 1 ml M9 buffer and DiO was added (final 0.01 mg/ml). The mixture was slowly shaked for 3 hours. Worms were washed with M9 buffer twice. Fluorescent signal was observed with an Axiovert A1 fluorescent microscope (Carl Zeiss).
Visualisation of cilia structure. Cilia were visualised using mnIs17 expressing osm-6::gfp. OSM-6::GFP is concentrated in cilia and the base of cilia. Phasmid PHA/PHB cilia were mainly observed because the structure was easily and reproducibly detected. Amphid cilia were clouded and seemed different depending on the body angle. They were not analysed in detail. An Axiovert microscope (Carl Zeiss, Jena, Germany) equipped with a LSM710 confocal system (Carl Zeiss) and a Plan Apochromat (x63, NA1.4) objective was used to detect fluorescent signals.
Visualisation of the IFT. Worms expressing IFT-139::GFP or IFT-139(P810L)::GFP were mounted on agarose pads and immobilised with 2 mM levamisole. Phasmid cilia were analysed on an IX81 microscope (Olympus, Tokyo, Japan) equipped with a 100× objective lens (NA1.35) and a CSU-X1 spinning disk confocal head (Yokogawa, Tokyo, Japan). The frame rate was 4 frames/s. Kymographs were made using Image J (National Institutes of Health) using the Multiple Kymograph plugin.