The Talpid3 gene (KIAA0586) encodes a centrosomal protein that is essential for primary cilia formation

The chicken talpid³ mutant has a complex phenotype,including polydactylous limbs with many unpatterned digits, vascular defects,hypoteleorism, abnormal dorsoventral patterning of the neural tube, loss of endochondral bone formation and embryonic lethality(Davey et al., 2007; Davey et al., 2006; Ede and Kelly, 1964a; Ede and Kelly, 1964b). Development of all the regions affected in the talpid³mutant embryo requires Hedgehog (Hh) signalling and analysis of the developing mutant limb bud and neural tube has shown that it is the response to Hh signalling that is defective (Lewis et al., 1999). Expression of some downstream Shh target genes in talpid³ mutant limb buds, head, neural tube and somites is lost, whereas other genes are expressed more widely, suggesting that there is failure both to activate gene expression in response to Shh and to repress gene expression in its absence (Buxton et al., 2004; Davey et al.,2006; Lewis et al.,1999). These opposing effects can be understood in terms of the bifunctionality of the Gli proteins, the transcriptional effectors of vertebrate Hedgehog (Hh) signalling, which function as activators (mainly Gli1, Gli2) or are processed to short forms (mainly Gli3) that function as repressors (Aza-Blanc et al.,2000; Bai et al.,2002; Marigo et al.,1996; Ruiz i Altaba,1999). Indeed, direct analysis of Gli3 proteins in talpid³ mutant tissues showed that Gli3 processing is abnormal, although translocation to the nucleus still occurs(Davey et al., 2006). We have identified KIAA0586 (Talpid3) as the gene affected in talpid³ mutant embryos(Davey et al., 2006). KIAA0586 is ubiquitously expressed and encodes a novel protein with no previously known function (Davey et al., 2006).

The phenotype of talpid³ mutant chicken embryos,including the inability to process Gli3, is strikingly similar to that of mouse embryos with mutations in genes encoding centrosomal or intraflagellar transport (IFT) proteins, such as Arl13b, OFD1, Polaris, IFT172, Kif3a,Dnchc2 and Ftm (Caspary et al., 2007; Ferrante et al.,2006; Haycraft et al.,2005; Huangfu et al.,2003; May et al.,2005; Vierkotten et al.,2007). These mouse mutants lack normal primary cilia, the site where cells receive Shh signals and other cell-cell signals(Corbit et al., 2005; Eggenschwiler and Anderson,2007; Rohatgi et al.,2007). An increasing number of human syndromes, collectively known as ciliopathies, which affect cilia formation and function, also have phenotypic features similar to talpid³ mutant chickens,such as polydactyly (Badano et al.,2006; Bisgrove and Yost,2006; Eley et al.,2005; Fliegauf et al.,2007; Pazour and Rosenbaum,2002; Tobin and Beales,2007). Interestingly, KIAA0586 was identified in a human centrosome proteome (Andersen et al.,2003). We have therefore investigated whether the mechanism underlying abnormal Hh signalling in the talpid³ mutant involves a failure of primary cilia formation and whether the Talpid3 protein is enriched in the ciliary apparatus.

Fertilised White Leghorn chicken eggs were obtained from H. Stewart(Lincolnshire) and incubated at 37°C and staged according to Hamburger and Hamilton (Hamburger and Hamilton,1992). Talpid³ carriers were maintained as described previously (Davey et al.,2006).

Embryos were fixed in 4% PFA for 2 hours at room temperature (RT) and genotyped as described previously (Davey et al., 2006). Selected embryos embedded in 10% sucrose, 7.5%gelatine were sectioned at 10 μm or 40 μm. Sections were stained as described previously (Das et al.,2006). Primary antibodies were as follows: for visualisation of primary cilia, rabbit anti-γ-tubulin 1:1000 (Sigma) and mouse anti-acetylated tubulin 1:1000 (Sigma); for microtubules, mouse antiα-tubulin 1:1000 (Sigma); mouse anti-Islet1 1:10 (Developmental Studies Hybridoma Bank; DSHB), mouse anti-NKX2.2 1:5 (DSHB), mouse anti-PAX6 1:2(DSHB), mouse anti-PAX7 1:10 (DSHB); and for filamentous actin, Alexa Fluor 546 phalloidin 1:100 (Molecular Probes). Secondary antibodies were:Alexa-Fluor-488 conjugated anti-mouse 1:500 (Molecular Probes),Alexa-Fluor-546 conjugated anti-rabbit 1:500 (Molecular Probes) and Alexa-Fluor-610 conjugated anti-rabbit 1:1000 (Molecular Probes). Sections counterstained with DAPI, viewed using Leica DMR compound microscope or Zeiss LSM510 or Nikon eC1 laser scanning confocal microscope.

Stage HH24 chicken embryos treated as described previously(Davey et al., 2007). For SEM,samples washed in 100% acetone for 10 minutes before critical point drying,flushed five times to ensure excess acetone was removed, dried, then placed on carbon mounts, coated with Au/Pd and imaged using an ESEM at 15 kV, 10 mm WD in high vacuum mode. For TEM sections cut between 70-100 nm and viewed with FEI Tecnai 12 transmission electron microscope.

Seven-day-old embryos fixed 4% PFA overnight, washed in PBS, dehydrated into 100% xylene, then wax embedded overnight. Sections were cut at 5 μm,dewaxed, stained with Hematoxylin and Eosin, and photographed using Leica DMR compound microscope.

The lumen of the neural tube of stage HH10-14 talpid³mutant flock embryos was injected with 1 μg/μl of construct encoding either full-length chicken Talpid3 (ggKIAA0586) or fragments of chicken Talpid3 in pCAGGs, plus 0.1 μg/μl pCAGGS-RFP and 0.02% Fast Green, and electroporated with square wave current of 25 mV applied five times for 15 mseconds (CUY21 electroporator, Protech International, TX, USA). Embryos were incubated for a further 48 hours, which is when mutant embryos can be identified, then fixed and sectioned.

Chick embryonic fibroblasts (CEFs) were generated from HH24 embryo bodies by cutting the tissue into small pieces before trypsinising for 1-2 minutes. Trypsin was inhibited with serum, tissue disaggregated, cells spun down and then resuspended in fresh DMEM/Ham's F12 (Gibco) plus 10% foetal calf serum(FCS). Cells were passaged twice before use. Primary cell cultures were generated from HH24 limb buds, by trypsinisation for 1-2 minutes, tissue disaggregated, then cells spun down and resuspended in 20 μl of DMEM and seeded onto fibronectin (SIGMA) coated 13 mm coverslips. Cells were left to adhere for 1 hour before flooding with DMEM/Ham's F12+10% FCS. HEK293T cells were maintained in DMEM+10% FCS under normal cell culture conditions.

In test for rescue of primary cilia, CEFs were seeded onto coated 13 mm coverslips at 1×10⁶/ml and transfected with a construct encoding full-length chicken Talpid3 in pCAGGS (ggKIAA0586) using Fugene6(Roche) for 48 hours. When near confluence, cells were synchronised in serum-free medium for 2 days, then fixed and stained.

For localisation of tagged Talpid3, HEK293T cells were seeded onto 13 mm coverslips at 1×10⁶/ml, then grown for 18-24 hours before transfection with 1 μg ggKIAA0586::Myc or Flag::ggKIAA0586. For localisation of GFP fusion construct, primary chicken cell cultures or HEK293T cells were transfected with 0.5 μg/μl ggKIAA0586ex11/12::GFP or hsKIAA0586ex11/12::GFP, respectively, in pCAGGS using Fugene6, then observed after 5-7 hours for GFP expression, fixed in 4% PFA 10 minutes, washed three times in PBS and stored for 24 hours at 4°C, then blocked in 5%FCS/PBS+0.5% Triton for 30 minutes.

Cells were washed in PBS, fixed in ice-cold methanol and acetone (50%) for 10 minutes, and blocked in PBS/0.2% Tween20/10% goat serum for 30 minutes.

An antibody was raised against C-terminal peptide (DSDSSGADTF) of chicken Talpid3 in rabbit. Serum from fifth bleed was used to detect overexpressed HA::ggKIAA0586 by western blot analysis, producing band of 200 kDa. Antibody was then affinity purified by coupling peptide to HiTrap NHS-activated HP column (Amersham) and tested by immunofluorescence in HEK293T cells transfected with HA::ggKIAA0586. All transfected cells were recognised by both anti-Talpid3 and anti-HA antibodies. Purified Talpid3 antibody was diluted 1:2 for cell immunofluorescence staining.

Primary antibodies used were as follows: for visualisation of primary cilia, rabbit anti-γ-tubulin 1:1000 (Sigma) and mouse anti-acetylated tubulin 1:1000 (Sigma); for actin and focal adhesions, anti-actin 1:1000(Sigma), Alexa-Fluor-488 phalloidin 1:1000 (Molecular probes), anti-vinculin 1:1000 (Sigma); for tagged Talpid3 constructs, monoclonal mouse anti-Myc 1:1000 (Sigma) and mouse anti-Flag antibody 1:1000 (Sigma); for centrosomes,rabbit anti-pericentrin 1:5000 (abcam); for transfected cells, mouse anti-HA 1:2000 (Sigma); and for microtubules, mouse anti-α-tubulin 1:1000(Sigma). All antibodies were applied for 1 hour at room temperature, then removed with three 5-minute washes in PBS/0.2% Tween20. Secondary antibodies were: Alexa-Fluor-488 conjugated anti-mouse 1:500 (Molecular Probes),Alexa-Fluor-546 conjugated anti-rabbit 1:500 (Molecular Probes) and Alexa-Fluor-610 conjugated anti-rabbit 1:1000 (Molecular Probes), and were incubated for 1 hour at room temperature. Samples were DAPI stained and mounted, and viewed on Zeiss LSM510 confocal microscope.

CEFs were seeded onto cover slips in DMEM/Ham's F12 at 0.75×10⁶/ml. 24 hours later treated with 25 μM nocodazol at 37°C for 1 hour. After nocodazol removal, cells were incubated for 0,10 or 60 minutes, then fixed in 50% methanol/50% acetone for 10 minutes, and stained with mouse anti-α-tubulin as above.

Nvtalpid3, the cnidarian homologue of Talpid3, was cloned from Nematostella vectensis by extending an EST sequence using ORF predictions available for Nematostella genome and by RACE PCR. Primer sequences are available upon request. Putative full-length clone 6058 bp(Accession Number FJ428244) encoding conceptually translated 1708 amino acid protein, was validated by RT-PCR and sequencing.

EMBOSS sequence analysis system (Rice et al., 2000) was used to extract amino acid sequences (SEQRET),to derive peptide statistics (PEPSTATS), to create plots of sequence conservation based on multiple sequence alignments (PLOTCON) and to produce helical wheel diagrams (PEPWHEEL) to visualise distribution of polar and non-polar residues in alpha helical regions. Multiple alignments of amino acid sequences were made using MUSCLE (Edgar,2004) and viewed using JALVIEW(Clamp et al., 2004). PFAMSCAN(Finn et al., 2006) and PSCAN(Gattiker et al., 2002) were used to scan amino acid sequences for sequence motifs. PCOILS was used to predict coiled-coil regions (Gruber et al., 2006). Secondary structure of primary protein sequences were predicted using SABLE (Porollo et al.,2004; Wagner et al.,2005); results were displayed using POLYVIEW(Porollo et al., 2004). DOMPRO(Yoo et al., 2008),SCOOBY-DOMAIN (Pang et al.,2008) and DOMAINATION (George and Heringa, 2002) were used to predict domain boundaries,globular domains and protein domains from local gapped alignments generated using PSI-BLAST, respectively. GLOBPLOT(Linding et al., 2003) was used as described previously (Davey et al., 2006).

Sections of stage HH24 wild-type chicken neural tube(Fig. 1A) and mesonephric duct(Fig. 1C) reveal well-developed ciliary axonemes projecting into the lumen (asterisk) from centrosomes (red;γ-tubulin). Cilia were also seen on limb bud mesenchyme cells(Fig. 1E); notochord cells(Fig. 1G); endothelial cells of both dorsal aorta and cardinal vein; epithelial and mesenchymal cells of somites; gut epithelia; neuroectoderm of the developing eye; endocardium; and extra-embryonic mesoderm (data not shown). By contrast, in sections of talpid³ mutant embryos, no ciliary axonemes could be seen projecting into the lumen of the neural tube, although centrosomes were clearly visible (Fig. 1B,D);nor could primary cilia be seen on cells in the limb bud(Fig. 1F), notochord(Fig. 1H) or any of the other tissues listed above (data not shown). Scanning electron microscopy of wild-type embryos also showed that primary cilia can readily be distinguished projecting from the centre of many cells of the outer periderm layer of the limb bud ectoderm (Fig. 1I,black circles; Fig.1I′)(41/111 cells examined) and in the neural tube, projecting from pits in the apical surface of cells lining the lumen(Fig. 1K, arrows). In talpid³ mutant embryos, no primary cilia could be seen projecting from periderm cells (Fig. 1J) (132 cells examined) nor neural tube cells(Fig. 1L).

In human ciliopathies and mouse mutants that lack primary cilia, a range of defects can occur in addition to those associated with abnormal Hh signalling,including polycystic kidneys (Bisgrove and Yost, 2006; Lehman et al.,2008). Mutant embryos from our current talpid³flock occasionally survive for up to 10 days, thus allowing examination of organs later in development. Histology of embryonic day 7 talpid³ mutant mesonephros (the functional embryonic kidney in chickens) revealed multiple large cysts(Fig. 1N, asterisks; compare with wild-type kidney, Fig. 1M). This is comparable with the pathology of the developing metanephric kidney seen in mice with abnormal ciliogenesis(Lehman et al., 2008).

To confirm that lack of primary cilia on talpid³ mutant cells is a consequence of a mutation in the Talpid3 gene(KIAA0586), we carried out rescue experiments in ovo by electroporating the neural tube of talpid³ mutant embryos with a construct encoding full-length chicken Talpid3 (ggKIAA0586). We have previously shown that electroporation of this construct restored wild-type dorsoventral patterning in the neural tube of mutant embryos(Davey et al., 2006). This construct also rescued primary cilia (Fig. 2A,A′,B). Furthermore, when mutant chicken embryonic fibroblasts in culture were transfected with the same expression construct,primary cilia were rescued (Fig. 2E) (n=5/5 transfected cells observed, compare with Fig. 2D) and axoneme formation was comparable with that in wild-type chicken fibroblasts(Fig. 2C).

Stages in formation of a primary cilium have been deduced from detailed analysis of fibroblasts and smooth muscle cells in chicken and mammalian tissues using transmission electronmicroscopy(Sorokin, 1962; Sorokin, 1968). We therefore examined the ultrastructure of talpid³ mutant cells to gain insights into why ciliogenesis fails.

In wild-type chicken neural tube, most cells had primary cilia projecting into the lumen (Fig. 3A,B). Each cilium emerged from a pit in the apical cell surface, and the axoneme was enclosed in a sheath of ciliary membrane and contained microtubules extending along its length (Fig. 3A,B). The basal body can be recognised by its appendages (structures including satellites and rootlets associated with the mature basal body)(Fig. 3A,B)(Sorokin, 1968). In some sections, the sister centriole located below the basal body could also be seen(Fig. 3B). By contrast, in cells of talpid³ mutant neural tube, no primary cilia projected from apical cell surfaces into the lumen(Fig. 3C-E) (3/3 mutant embryos examined). Basal bodies (Fig. 3C-E), however, were readily identified by the presence of associated appendages (Fig. 3C-E), and these, together with their sister centrioles(s), were seen in the apical region of the cells. In some cases, a few short microtubules were present distally on the basal body(Fig. 3C) but, in most cases,no trace of axoneme development was observed. Neither were any ciliary vesicles associated with the basal bodies(Fig. 3C,D) (55/58 basal bodies observed, two embryos examined) although in three cases, a vesicle was observed nearby (Fig. 3E,E′). Therefore in talpid³ mutant embryos, migration of the centrioles to the apical region of neural tube cells and maturation of the mother centriole into a basal body appeared unaffected,but docking, which involves fusion between the ciliary vesicle associated with the basal body and the apical cell membrane(Dawe et al., 2007; Sorokin, 1968), and subsequent axoneme formation did not occur. The TEM images of talpid³mutant cells also showed that basal bodies were frequently misorientated and did not lie perpendicular to the apical cell surface as in normal cells(Fig. 3C,D; see Fig. S1 in the supplementary material) [9/18 (50%) basal bodies misorientated in one talpid³ mutant embryo; 26/40 (65%) basal bodies misorientated in another].

It has been suggested that the actin cytoskeleton in the apical region of epithelial cells orientates the basal body, allowing microtubules to polymerise into the ciliary axoneme (Park et al., 2006), and that apical actin enrichment is required for ciliogenesis (Pan et al.,2007). Confocal microscopy of transverse sections of wild-type chicken neural tube showed an even continuous band of F-actin, stained with phalloidin at the apex of cells (Fig. 3F), whereas in the talpid³ mutant, even though there was strong staining at the cell apex, it was punctate and not continuous (Fig. 3G). Electroporation of ggKIAA0586, which rescued primary cilium formation (see Fig. 2A), also restored the continuous band of phallodin staining at the apex of talpid³ neural tube cells on the electroporated side(Fig. 3H). Abnormalities in microfilament organisation were also seen in cultured mutant cells. talpid³ mutant limb cells had stronger actin staining at the ruffled membrane and fewer stress fibres than did wild-type limb cells(compare Fig. 3I,K with Fig. 3J,L). In addition, there were many fine filopodia containing actin around the circumference of the mutant cells, which were not seen in wild-type cells (compare Fig. 3I,K with Fig. 3J,L), a feature previously observed in scanning electron microscopical studies on talpid³ mutant limb bud cells in vivo(Ede et al., 1974). Discrete vinculin-positive focal adhesions were also less well defined in talpid³ mutant limb cells than in wild-type cells (compare Fig. 3M with Fig. 3N).

To verify that the Talpid3 protein is present in the centrosome [as suggested by Andersen et al. (Andersen et al., 2003)], we raised an antibody against the C terminus of chicken Talpid3 and used this in double immunofluorescence staining withγ-tubulin as a centrosomal marker in serum-starved wild-type and talpid³ mutant chicken embryonic fibroblasts to determine the subcellular localisation of Talpid3. The talpid³mutation results in a premature stop codon and, even if a truncated protein was produced, this antibody would not recognise it. In wild-type fibroblasts,Talpid3 antibody staining colocalised with γ-tubulin(Fig. 4A-C), and was enriched in both centrioles (Fig. 4D-F),whereas, in talpid³ mutant fibroblasts, Talpid3 antibody staining could not be detected in the centrosome(Fig. 4G-I).

The detection of Talpid3 protein in the centrosome is consistent with a role in primary cilia formation but the centrosome also directs microtubule organisation, including the mitotic spindle. We therefore examined localisation of Myc- or Flag-tagged Talpid3 during the cell cycle in HEK293T cells. Tagged Talpid3 protein (stained with antibodies against either Myc or Flag) co-localised with Pericentrin (a centrosomal marker) during interphase(Fig. 4J-L) and metaphase(Fig. 4M-O). More diffuse staining of tagged Talpid3 protein was also seen throughout the cytoplasm during anaphase (Fig. 4P-R) and telophase (Fig. 4S-U). Despite the presence of Talpid3 at the centrosome in early phases of the cell cycle,no spindle defects were observed in talpid³ mutant cells(data not shown). In addition, there were no obvious differences in microtubule organisation between chicken embryonic fibroblasts from wild-type and talpid³ mutant embryos (compare Fig. 5A,B with Fig. 5C,D), although there was a delay in microtubule re-growth after nocodazol treatment (compare Fig. 5E,F,I,J,M,N with Fig. 5G,H,K,L,O,P).

In order to identify functional domains in the Talpid3 protein, we extended our previous bioinformatics analysis(Davey et al., 2006) using orthologous cDNA sequences of Nematostella vectensis and a predicted homolog in the genome sequence of Strongylocentrotus purpuratus(Putnam et al., 2007; Sodergren et al., 2006). Alignment of vertebrate Nematostella vectensis and Strongylocentrotus purpuratus Talpid3 sequences revealed a distinct highly conserved region between amino acids 498-585(Fig. 6A,B; see Figs S2,S3 in the supplementary material), which lies downstream of the Talpid3 mutation,which would truncate the protein at amino acid 366. This conserved region is predicted to contain a single coiled-coil domain between amino acids 498-529(Fig. 6A,B; see Figs S2-S5 in the supplementary material) and is encoded by exons 11 and 12(Fig. 6A).

We tested the function of this highly conserved region of the Talpid3 protein using complementation experiments in talpid³mutant neural tube, as before, and monitored both cilia formation and neural tube dorsoventral patterning. In talpid³ mutant embryos,the neural tube is dorsalised and expression of ventral markers such as Nkx2.2 and Islet1 is lost, whereas expression of dorsal markers such as Pax6 and Pax7 is expanded. We have shown previously that these patterns of expression could be normalised by electroporation of constructs encoding full-length chicken Talpid3 (Davey et al., 2006). Analysis of the neural tube from talpid³ mutant embryos electroporated with a series of constructs encoding different fragments of Talpid3 showed that the entire conserved region is essential for rescue(Fig. 7A-H). Thus, constructs containing the entire conserved region (construct D) rescued primary cilia formation (Fig. 7Q,R) and neural tube patterning (construct C, Fig. 7I-L), inducing expression of Nkx2.2 and Islet1(Fig. 7I,J; compare RFP electroporated side with non-electroporated side) and restricting Pax6 and Pax7 expression dorsally (Fig. 7K,L). Electroporation of a construct encoding just the conserved region (Fig. 7H), however, was not able to rescue neural tube patterning. Neither primary cilia formation nor neural tube pattern was rescued by constructs encoding C-terminal fragments lacking the coiled-coil domain (construct E, Fig. 7E), the adjacent highly conserved region, amino acids 529-585 (construct F, Fig. 7F) or the entire conserved region (constructs B and G, Fig. 7B,G). Fig. 7M-Pshows a talpid³ mutant neural tube electroporated with construct E. Despite substantial RFP expression, indicating successful transfection, expression patterns of Nkx2.2, Islet1, Pax6 and Pax7 were unchanged. Thus, the conserved region is required but not sufficient to rescue dorsoventral patterning of the neural tube. It should be noted that none of the constructs (A-H) altered neural tube patterning in wild-type embryos and therefore do not show any dominant-negative effects, including construct B,which encodes the fragment of Talpid3 protein predicted to be expressed in the mutant.

To determine the function of the conserved region in centrosomal localisation, we transfected a construct encoding this region alone from the human protein (KIAA0586) fused to GFP (hsKIAA0586ex11/12::GFP) into HEK293T cells. GFP expression was seen in the centrosome(Fig. 7T) (3/3 transfected cells observed) co-localising with γ-tubulin(Fig. 7S,U). Likewise, when a construct encoding the chicken Talpid3 conserved region fused to GFP(ggKIAA0586ex11/12::GFP) was transfected into chicken primary culture cells,co-localisation was also seen with γ-tubulin at the centrosome (data not shown). These data indicate that the conserved region is sufficient to target Talpid3 protein to the centrosome.

Here, we show that the talpid³ chicken mutant lacks primary cilia and demonstrate using rescue experiments that this is a direct result of loss of Talpid3 function. Lack of primary cilia in talpid³ chicken mutants provides an explanation for the Hh signalling defects, inability to process Gli3(Davey et al., 2006) and similarities with mouse mutants that lack cilia and were originally highlighted as being defective in Hh signalling(Haycraft et al., 2005; Huangfu and Anderson, 2005; Huangfu et al., 2003).

Primary cilia are absent on cells in all talpid³chicken mutant tissues studied, including those not known to be dependent on Hh signalling, such as the mesonephric duct. A growing number of human conditions known as ciliopathies, including syndromes such as primary cilia dyskinesia, Bardet-Biedl syndrome (BBS), Joubert syndrome and Meckel syndrome(Badano et al., 2006; Bisgrove and Yost, 2006; Fliegauf et al., 2007; Tobin and Beales, 2007), have a range of defects, including those associated with abnormal Hh signalling and also polycystic kidneys. A role for primary cilia in polycystic kidney disease was first suggested after it was discovered that the genes affected in mice with polycystic kidneys, e.g. the orpk and inv mice(Lehman et al., 2008; Moyer et al., 1994; Shiba et al., 2005; Siroky and Guay-Woodford,2006), encoded cilia associated proteins. Thus, our finding that kidneys of 7-day-old talpid³ chicken mutant embryos are cystic is consistent with the general inability of talpid³mutant cells to form primary cilia. Thus, we conclude that the talpid³ chicken mutant is a new example of a ciliopathy and a potential model for human disease. The range of defects in different human ciliopathies varies although the reasons for this are not clear. Some ciliopathies, such as BBS, Joubert syndrome, Meckel syndrome and oral facial digital syndrome (OFD) have features that one would specifically associate with Hh signalling defects, such as polydactyly, similar to the talpid³ mutants, whereas others, such as those caused by mutations in polycystin 1 and polycystin 2, have kidney defects. It will be interesting to define the precise spectrum of defects in talpid³ chicken mutants for comparison with the human syndromes.

Our ultrastructural studies of cells in the talpid³mutant neural tube suggest that ciliogenesis fails because basal bodies do not dock at the apical cell membrane. One possible reason why docking fails in the talpid³ mutant is that Talpid3 is involved in fusion of the ciliary vesicle to the basal body. Another centrosomal protein, BBS1,functions in this way by binding to Rabin8, a guanine nucleotide exchange factor, which activates Rab8, a Rab-GTPase that specifically traffics ciliary membrane to the base of the primary cilium(Nachury et al., 2007; Yoshimura et al., 2007). BBS1 is part of a complex of centrosomal proteins collectively termed the BBSome,which also includes BBS2, 4, 5, 7, 8 and 9. Whether the Talpid3 protein is also part of the BBSome or helps to traffic Rab8a to the cilium remains to be investigated. Another possible reason why ciliogenesis fails in the talpid³ chicken mutant is because Talpid3 is required for apical actin enrichment. In oviduct ciliated cells, the apical actin network is closely associated with basal body appendages(Chailley et al., 1989), and recent work on Xenopus laevis epidermal cells with motile cilia has shown that apical enrichment of actin is required for ciliogenesis(Park et al., 2006). Therefore, it is possible that the abnormal actin organisation in talpid³ mutant cells results in basal body misorientation,leading to failure of ciliogenesis. Apical actin enrichment is mediated by activation of RhoA (Pan et al.,2007) and it has been shown that RhoA is localised to the basal body in multiciliated cells (Park et al.,2008), thus providing an explanation for how a centrosomal protein such as Talpid3 could directly affect actin organisation. Furthermore, reduced numbers of stress fibres and focal adhesions in talpid³mutant cells suggests that RhoA activity is decreased(Nobes and Hall, 1999). Localisation of RhoA to the basal body in Xenopus laevis epidermal cells has been shown to be mediated by Inturned, an effector of the Wnt planar cell polarity signalling pathway (Park et al., 2008); interestingly, Inturned, together with Fuzzy, is required for normal cilium formation (Park et al., 2006).

We have shown that the Talpid3 protein is present in both centrioles of the centrosome. There are several mouse ciliopathy models with mutations in genes encoding centrosomal proteins, including BBS1, BBS2, BBS4, OFD1 and Ftm(Davis et al., 2007; Ferrante et al., 2006; Mykytyn et al., 2004; Nishimura et al., 2004; Vierkotten et al., 2007). BBS1, BBS2, and BBS4 mutant mice still form primary cilia, although they are abnormal or degenerate, whereas both OFD1 and Ftm mutant mice lack primary cilia and have a similar phenotype to talpid³ chicken mutants, including polydactyly and dorsalised neural tube. Furthermore, OFD1 has been shown to localise to both centrioles in human undifferentiated embryonic cells (Romio et al.,2004). Thus, the talpid³ chicken mutant most closely resembles OFD1 mutant mice. There are no ultrastructural studies, to date, of cells from either OFD1 or Ftm mutants, and therefore it is not clear whether ciliogenesis fails at the same stage in these mutants as in talpid³. Another centrosomal protein ODF2, has been suggested to be necessary for basal body docking. However, in Odf^-/- cells, unlike talpid³ mutant cells, basal bodies fail to mature and lack appendages(Ishikawa et al., 2005),suggesting that Talpid3 acts downstream of ODF2.

Other centrosomal proteins play roles in microtubule organisation but our observations suggest that this is not the case for Talpid3 as both the microtubule network in interphase cells and mitotic spindles appear normal in mutant cells. A change in microtubule dynamics, however, was observed in talpid³ mutant cells. Ede and Flint(Ede and Flint, 1975) found that talpid³ cells move slower than wild-type cells, but showed that this was not due to the talpid³ mutant cells moving intrinsically more slowly but instead spending more time at rest. This pausing might be explained by the slower rate of microtubule re-growth observed in talpid³ mutant cells.

We have identified a region of the Talpid3 protein that is conserved all the way down to Nematostella vectensis. This conserved region is sufficient for centrosomal localisation and interestingly there is significant distant homology between this region in Talpid3 and a region in another centrosomal protein, CCCAP (centrosomal colon cancer autoantigen protein)(Kenedy et al., 2003)(Fig. 6C) (PSI-Blast E value 1e-47 and sequence similarity 47%). Further structure/function analysis of the Talpid3 protein showed that the highly conserved region is required but not sufficient to rescue primary cilia formation, thus suggesting that other domains in the C terminus are also required. Rescue of neural tube patterning in the mutant provides a powerful assay with which to identify these domains. A deeper understanding of the Talpid3 protein will give new insights into mechanisms involved in normal ciliogenesis and may also shed light on the basis of human ciliopathies.