|Home | About | Journals | Submit | Contact Us | Français|
The primary cilium has recently stepped into the spotlight, as a flood of data demonstrate that this organelle has crucial roles in vertebrate development and human genetic diseases. Cilia are required for the response to developmental signals, and evidence is accumulating that the primary cilium is specialized for Hedgehog (Hh) signal transduction. Formation of cilia, in turn, is regulated by other signaling pathways, possibly including the planar cell polarity pathway. The cilium therefore represents a nexus for signaling pathways during development. The connections between cilia and developmental signaling have begun to clarify the basis of human diseases associated with ciliary dysfunction.
The primary cilium, a slim microtubule-based organelle that projects from the surface of vertebrate cells, has been the focus of intensive studies transforming it from a poorly understood curiosity into a structure recognized for its importance in development, inherited human disease and cancer. Cilia and flagella are ancient structures present in organisms as diverse as single celled eukaryotes and humans. The evolutionarily conserved mechanism of Intraflagellar Transport (IFT), first described in the alga Chlamydomonas, is essential for the construction and maintenance of these structures in all species1, 2.
In the past decade, the function of mammalian primary cilia has been revealed by both developmental genetic analyses and human genetic studies. Disruptions of the primary cilium have been associated with the common disorder human cystic kidney disease3–6. In addition, rare recessive human disorders known as ciliopathies, with complex syndromes that include cystic kidneys, obesity, mental retardation, blindness and various developmental malformations, have been shown to be caused by mutations in proteins localized to cilia and ciliary basal bodies (reviewed in7–10). In parallel, genetic studies in the mouse demonstrated that cilia are essential for signaling through the Hh pathway, a crucial signaling pathway for organizing the body plan, organogenesis and tumorigenesis11.
The importance of primary cilia in vertebrate development was first revealed in genetic experiments that demonstrated that cilia are required for survival and patterning of the mouse embryo11. Phenotypic, genetic and biochemical analysis then showed that embryonic phenotypes of the cilia mutants were caused by disruption of Hh signal transduction. This unexpected finding raised many questions, including why the cilium is a good locale for signal transduction, why cilia are required for vertebrate but not invertebrate Hh signaling, and whether primary cilia are important in regulating other developmental signaling pathways.
Other recent experiments have suggested that additional developmental signaling pathways help regulate the formation of cilia. The most complete studies have implicated components of the planar cell polarity (PCP) pathway in the regulation of the position and formation of cilia. These processes, which could indirectly regulate the activity of Hh signaling, appear to be particularly important during organogenesis.
Here we review the relationships between primary cilia and signaling pathways during vertebrate embryonic development. After describing the evolutionarily conserved mechanism of IFT, we review the evidence that Hh signaling requires IFT and cilia. We then describe recent work suggesting that the primary cilium in vertebrate embryos is specialized such that Hh signaling is restricted to the cilium. After considering whether additional developmental signaling pathways require cilia, we discuss the evidence that other signaling pathways regulate ciliogenesis. We conclude with a discussion of how the findings on the relationship between cilia and developmental signals are beginning to explain the syndromes seen in cilia-related human diseases, focusing on the formation of kidney cysts, a hallmark of disorders caused by abnormal primary cilia.
The cilium is extended and maintained by the transport of particles along the axoneme mediated by the IFT machinery (Fig. 1). IFT trafficking from the base to the tip of the cilium depends on the microtubule plus-end directed motor Kinesin-2, which associates with two IFT protein complexes, IFT-A and IFT-B. IFT-B is essential for anterograde trafficking, while IFT-A and the minus-end directed motor cytoplasmic Dynein-2 (Dync2h1) are required for retrograde trafficking1. In all organisms studied, disruption of the Kinesin-2 motor or the IFT-B complex blocks cilia formation. Perturbation of retrograde trafficking by disruption of the Dynein or the IFT-A complex results in short, bulged cilia1, 12–15 (Table1).
Cilia are nucleated by the basal body, which is made up of the mother centriole and associated pericentriolar proteins. Some basal body proteins are required for cilia formation; evidence suggests some may recruit cargo from the Golgi to the nascent ciliary membrane and others may promote loading of cargo into the axoneme16, 17 (Fig. 1).
The first evidence that vertebrate Hh signaling depends on cilia came from a phenotype-based screen for mutations that alter patterning of the mouse embryo. This screen identified several mutants displaying morphological and patterning phenotypes consistent with altered Hh signaling, including loss of the ventral cell types in the neural tube specified by high levels of Sonic hedgehog (Shh)11. The genes disrupted in these mutants encode several components of the IFT machinery, including the IFT-B complex components Ift172 and Ift88, as well as Dync2h1, the IFT-dedicated retrograde motor11, 18, 19 (Fig. 1). Disruption of the Kinesin-2 motor in Kif3a null embryos also caused similar defects in Shh-dependent neural patterning (Fig. 2)5. Genetic studies showed that IFT proteins act at the heart of the Shh pathway, downstream of the membrane proteins Patched (Ptch) and Smoothened (Smo) and upstream of the Gli transcription factors that implement the pathway11, 18 (Fig. 3, Table 1).
The role of IFT proteins in Hh signaling is complex, partly because of the complex output of the Hh pathway. In the absence of Hh ligand, Gli transcription factors, which function as effectors of the pathway, are proteolytically processed to Gli repressor forms (GliR) that keep Hh target genes off (Fig. 3). In response to Hh ligand, processing of GliR is blocked and activated Gli transcription factors (GliA) activate the expression of Hh target genes. IFT is required both for production of GliA and Gli3R18–21; as a result, IFT mutants show loss of Hh phenotypes in some cell types and gain of Hh phenotypes in others. For example, GliA plays a central role in neural patterning, and IFT mutant embryos show a loss of Hh signaling in the neural tube. In contrast, GliR plays a central role in limb development and IFT mutants that survive to later stages of embryogenesis show pre-axial polydactyly, characteristic of loss of GliR (Fig. 2)18, 19, 21.
Recent experiments demonstrated that IFT is also required for Hh signaling in zebrafish. Zebrafish lacking both maternal and zygotic Ift88 exhibit Hh signaling defects in the neural tube and somites22, however the patterning defects caused by loss of IFT in zebrafish are slightly different than those seen in mammals. Mouse Ift88 mutants lack Shh-dependent ventral neural cell fates11, and zebrafish maternal/zygotic Ift88 mutants lack cell fates requiring the highest levels of Hh, such as V3 interneuron progenitors and muscle pioneer cells. However, cell types within both the neural tube and somites normally specified by lower levels of Hh are expanded22. This difference may be due to a different balance of Gli activators and repressors in the fish compared to the mouse22.
Additional evidence that cilia are required for Hh signaling came from analysis of basal body protein mutations. Dozens of proteins are localized to centrosomes and the pericentriolar material, and a subset of these proteins has been shown to be required for cilia formation (Table 1). In every case examined thus far, these proteins are required for Hh signaling. For example, the chick talpid3 mutation was first identified based on polydactyly23, 24 and causes developmental defects consistent with disrupted Hh signaling25, 26. Talpid3 mutant embryos fail to form cilia, and the affected gene was shown to encode a centrosomal protein27.
Mutations disrupting other basal body proteins, including OFD1, FTM, MKS1 and EVC, cause human ciliopathies and affect mammalian Hh signaling. Mice mutant for Ofd1, Ftm/Rpgrip1l, or Mks1 have abnormal or absent cilia, and exhibit Hh signaling defects corresponding to the severity of the cilia disruption28–32 (Table 1). EVC also localizes to the basal body and is mutated in the human skeletal disorder Ellis-van Creveld syndrome33, 34. Unlike other basal body proteins, expression of Evc in the mouse is limited to developing skeletal structures, and Evc does not appear to be required for the formation of cilia in chondrocytes. Nevertheless, Evc−/− mice exhibit reduced Indian hedgehog (Ihh) signaling specifically within skeletal structures. Thus Evc apparently does not affect ciliogenesis, but is required for Hh signaling in a specific cell type.
Based on the genetic studies associated Hh signaling with cilia, several groups have tested whether the proteins that mediate Hh signal transduction are localized to cilia. Remarkably, all the key components of the Hh pathway are enriched in cilia (Fig. 3b).
Two transmembrane proteins, Ptch1 (the Hh receptor) and Smo (which acts downstream of Ptch1), show dynamic, Hh-dependent trafficking in cilia35, 36. In the absence of Hh ligand, Ptch1 is localized to the base of the cilium and Smo is not associated with cilia. Upon exposure to ligand, Ptch1 exits and Smo moves into the cilium35, 36. Activating mutations in Smo, as well as pathway agonists, cause Smo to localize constitutively to the cilium36. Despite the enrichment of Smo in the cilium in response to Shh, Smo also accumulates in cilia in cells that lack the Dync2h1 retrograde IFT motor in the absence of ligand37, 38, suggesting that Smo trafficks through the cilium in the absence of ligand, and that Shh increases ciliary accumulation of Smo.
Gli transcription factors are also enriched in cilia. Both Gli2, which functions primarily as a transcriptional activator in mammalian Hh signaling, and Gli3, which can be processed into a repressor, localize to the tips of cilia39, and recent reports indicate that pathway activation increases the amount of Gli2 and Gli3 at the tips of cilia in fibroblasts37, 40. Ciliary enrichment of Gli2 depends on the presence of activated Smo37. Like Smo, Gli2 accumulates at high levels in Dync2h1 mutant cilia37, suggesting that Gli2 trafficks continuously through the cilium, and activated Smo increases the accumulation of Gli2 at the tip.
Sufu, an important negative regulator of mammalian Hh signaling, also localizes to the primary cilia tip39,40. Genetic and biochemical data have shown that Sufu can inhibit Hh signaling even in the absence of cilia,41, 42 however partial knockdown of Sufu results in pathway activation only if cilia are present, suggesting a complex role for Sufu in the cilium20 (M. Tuson and K. Anderson, unpub. data). While the relationship between Sufu and cilia remains to be defined, the data are consistent with a model in which Smo activates the pathway at the cilia tip by antagonizing the activity of Sufu, thereby promoting activation of Gli transcription factors40 (Fig. 3b). Thus, Gli2 activation requires cilia, however the precise mechanism by which this occurs is not defined. In addition to suppression of Sufu, it may require post-translational modifications to Gli2 and the presence of yet-unidentified Hh pathway components within cilia.
The finding that vertebrate Hh signaling requires primary cilia has raised the question of why this organelle is particularly suited to this critical pathway. The simplest explanation is that the cilium provides an environment where pathway components are enriched to facilitate their interactions. However, the dynamic relocalization of pathway components in response to ligand suggests that trafficking of Hh pathway proteins is crucial for pathway activation, and it is likely that IFT proteins are important in this trafficking.
IFT depends on two protein complexes, IFT-A and IFT-B, that form large platforms transporting cargo between the base and tip of the cilium13 (Fig. 1). Mutants lacking components of the IFT-B protein complex (Ift172, Ift88, Ift52 and Ift57)11, 21, 43 lack cilia and all response to Hh ligands, precluding analysis of the role of IFT-mediated transport within the cilium. In contrast, mutations in IFT-A proteins allow the formation of cilia (with abnormal morphology) and cause very different developmental phenotypes from mutants that prevent cilia formation: Hh signaling is activated rather than decreased (Fig. 2, Table 1). Studies in Chlamydomonas argue that the IFT-A complex cooperates with dynein to mediate retrograde transport1, as the rate of anterograde IFT is normal in these mutants, while retrograde trafficking is slowed12, 14, 44. Mutants in two mouse IFT-A complex proteins have been characterized, THM-1 (aka Ttc1b; IFT-139) and IFT-122. These mutants show an expansion of Hh-dependent neural cell types, as well as increased expression of direct Hh target genes15, 45, 46.
The opposing phenotypes of IFT-A and IFT-B mouse mutants are surprising, as IFT-A and -B were originally identified as subcomplexes of a single large complex47, and appear to move coordinately in the same particle48, 49. Both IFT-A and Dync2h1 are important for retrograde IFT, but mutations in IFT-A proteins increased Hh pathway activity15, 45 while mutations in Dync2h1 block the response to Hh ligands18, 19. These findings suggest that disruption of IFT-A may differentially disrupt trafficking of Hh pathway components, thereby causing phenotypes distinct from those observed in mutants in which cilia or absent or the dynein motor is disrupted. Recent data suggests that Smo may be trafficked laterally from the plasma membrane into the cilium, presumably an IFT-independent mechanism50. Given that the IFT machinery functions downstream of Smo but upstream of the Gli transcription factors, it will be particularly informative to examine trafficking of Smo and the Gli proteins in IFT-A mutants.
Despite the evolutionary conservation of the Hh pathway and the importance of primary cilia in vertebrate Hh signaling, cilia are not required for Hh signaling in Drosophila. This raises the question of why vertebrate Hh signaling is coupled to cilia. Recent data suggest that Kif7, a kinesin that is the vertebrate homolog of Drosophila Costal2 (Cos2), may tether the vertebrate Hh pathway to cilia.
Cos2, a key component of the Drosophila Hh pathway, is a kinesin-related protein that serves as a scaffold for Hh signaling complexes. Cos2 has dual functions in the pathway: it promotes formation of the repressor form of Ci (the Gli homolog) in the absence of Hh ligand by recruiting kinases that prime Ci for processing; and it permits high levels of pathway activation upon Hh stimulation by antagonizing Sufu51–53 (Fig. 3). Although Cos2 can bind microtubules, amino acids in its motor domain have diverged from those of other kinesins such that its ability to bind ATP is disrupted54.
Several recent papers demonstrated that zebrafish and mouse Kif7 proteins, like Drosophila Cos2, both positively and negatively regulate the Shh pathway40, 55–57. Unlike Cos2, the vertebrate Kif7 motor domain retains all the motifs typical of kinesin motors, suggesting that it should act as a motor protein. In the absence of ligand, Kif7 localizes to the base of the primary cilium and moves to the tip of the cilium in response to pathway activation40, 55 (Fig. 3). This translocation depends on the Kif7 motor domain, suggesting that Kif7, like Kinesin-II, acts as an anterograde motor within the cilium55.
Conventional kinesins, such as Kif7, carry cargo towards the plus end of microtubules, and the minus ends of both axonemal and cytoplasmic microtubules are located at the base of the cilium. Because Kif7-eGFP is enriched at the base of the cilium, we proposed that Kif7 may traffic Gli2 protein away from the cilium in the absence of ligand to prevent Gli activation55 (Fig. 3b). The positive role of Kif7 is presumably coupled to its movement to the cilia tip after pathway activation, where Sufu and the Gli transcription factors are enriched. Kif7 may promote Gli activation at the tip, perhaps by antagonizing the activity of Sufu55. Kif7 appears to be required for the increase of both Gli2 and Gli3 at the cilia tip after exposure to ligand40, suggesting both Gli proteins may be cargos of Kif7 within the cilium. Thus, the dual roles of Kif7 as a Shh pathway component and ciliary motor could explain why mammalian Shh signaling depends on the primary cilium (Fig. 3b).
Fused (Fu) is an important component of the Drosophila Hh pathway: it is a serine/threonine kinase that phosphorylates Cos2, Sufu and perhaps other components of the pathway. and is required for activation of Ci in response to Hh ligand53. Fu is also important for Hh signaling in zebrafish,58 but Hh signaling is normal in mice lacking Fu59,60. Recent work has shown that mammalian and zebrafish Fu are required for the construction of specialized motile cilia61, and Fu mutant mice die postnatally with hydrocephalus, presumably due to dysfunction of motile cilia in brain ventricles59,60. Thus Fu, like Kif7/Cos2, links Hh signaling with cilia, although this connection appears to have been lost in mammals. It has been proposed that another unidentified kinase may substitute for Fu in mammalian Hh signaling, and several human kinome screens have been undertaken to identify kinases required for Hh signaling62, 63. While the kinases identified in these screens have yet to be characterized in vivo, it will be interesting to determine whether a protein that is functionally homologous to Fu also links Shh signaling to cilia in mammals.
Recent work in planaria supports the view that some conserved components of the Hh pathway were associated with cilia before they were associated with Hh signaling. Planaria homologues of the Hh pathway components Kif7/Cos2, Fu, and Iguana are required in planaria for formation of motile cilia but not Hh signaling64, 65. Planaria represent a lineage of animals distinct from both that of insects and of vertebrates. Thus, the finding that Kif7 and Fu function in cilia in two independent metazoan lineages suggests the ancestral role of these proteins was in cilia. The requirement for these cilia-associated proteins in Drosophila Hh signaling suggests that Hh signaling was associated with cilia in the common ancestor of Drosophila and vertebrates.
Most cells in the mouse embryo have primary cilia, while a relatively small number of cells respond to Hh at any particular stage. This has raised interest in the possibility that other developmental signaling pathways may also depend on cilia. However, the disruption of other developmental signaling pathways, including canonical and non-canonical Wnt, TGFβ, Notch and FGF signaling, causes developmental abnormalities that do not overlap with the IFT mutant phenotypes. Nevertheless, cilia could have more subtle roles in other signaling pathways or might be important for signaling at later embryonic stages, after IFT mutants arrest.
Most attention has focused on the relationship between cilia and Wnt signaling. Several groups reported that knockdown of cilia-associated proteins in cultured cells or zebrafish embryos elevates canonical Wnt signaling and/or disrupts processes that depend on non-canonical Wnt signaling, such as convergent extension66–72. The primary cilium was therefore proposed to act as a switch between canonical and non-canonical Wnt signaling pathways68, 69.
However, this connection between cilia and Wnt signaling is controversial. Mouse IFT mutants do not show the phenotypes characteristic of Wnt pathway mutants. For example, reduced canonical Wnt signaling disrupts gastrulation and early patterning73 and inappropriate activation of the Wnt pathway can cause axis duplications and failure to form anterior structures74–77. Although mammalian non-canonical Wnt pathway mutants fail to close the entire neural tube caudal to the forebrain, neural patterning in these mutants is relatively normal78, 79. Similarly, zebrafish mutants that lack both maternal and zygotic activity of the Ift88 gene have defects in Hh signaling, but do not show the defects in convergent extension defects associated with disruption of non-canonical Wnt signaling22.
Recent work examined the expression of canonical Wnt reporters in vivo in Kif3a, Ift88, Ift172 and Dync2h1 mutant mice and failed to find any alteration in either the domain or levels of Wnt activity38. Similarly, mouse embryos homozygous for a mutation in the IFTA protein Tct21b (IFT-139) have a neural patterning phenotype consistent with activation of Hh signaling15, 46, but do not show altered canonical Wnt signaling46. Thus it appears that cilia are not required for canonical or noncanonical Wnt signaling in the first half of vertebrate embryogenesis.
Cilia have been found to be important for signaling by PDGF receptor alpha (Pdgfrα) in cultured fibroblasts80, as well as for PDGF-dependent directed cell migration in these cells81. Additionally, the receptor is localized to primary cilia in vivo in neural stem cells of the adult rat subventricular zone (SVZ)82. Loss of Pdgfrα signaling does not produce any striking phenotypes in early mouse embryos, but is critical for development of later tissues, including oligodendrocytes and neural crest-derived craniofacial structures83, 84. It will be important to test whether loss of cilia in the second half of embryogenesis affects Pdgfrα signaling in these cell types in vivo.
After birth, Shh signaling continues to play important roles in the growth of the brain and the maintenance of neural progenitors85. Conditional deletion of Ift88 or Kif3a within the brain results in severe hypoplasia of the cerebellum due to a failure of granule cell progenitor proliferation86, a process that depends on Shh signaling85, 87. Primary cilia are also required to modulate the Shh-dependent formation and maintenance of hippocampal granule neuron precursors, important for maintaining neurogenesis in the adult88. Based on the tight association between IFT/cilia mutants and specific defects on Hh signaling, we propose that cilia are essential for Hedgehog signaling in all cell types and that, at least in early development, primary cilia in vertebrate embryos are dedicated to Hh signal transduction. The data do not rule out the possibility that cilia may have important roles in other signaling pathways later in development or in specific cell types. For example, although a number of the defects observed in human ciliopathies can be attributed to abnormal Hh signaling, the molecular bases of other features of these diseases remain unknown (Box 1).
Because of the importance of primary cilia in embryonic patterning, there is considerable interest in identifying signaling pathways that regulate cilia formation. Several transcription factors are known to be required for formation of motile cilia and node cilia (reviewed by89). However, only recently has evidence emerged about signaling pathways that regulate the formation and position of primary cilia.
Recent evidence from zebrafish implicates FGF signaling in the regulation of cilia length. Knockdown of Fgfr1 or Fgf ligands results in shortened cilia in Kupffer's vesicle and randomized organ laterality90–92. The expression of ciliogenic transcription factors and Ift88 is reduced in these embryos 91. It will be interesting to test whether Fgf pathway mutations in the mouse also affect ciliogenesis, and if these effects on ciliogenesis alter Hh signal transduction.
Components of the phosphatidylinositol signaling cascade also appear to regulate cilia length. In zebrafish, morpholino knockdown of the inositol kinase Ipk1 reduces the frequency of cilia beating and decreased cilia length93. In humans, the inositol phosphatase INPP5E is mutated in one form of Joubert syndrome, a ciliopathy. INPP5E is enriched in the ciliary axoneme, and in fibroblasts from patients with Joubert syndrome, cilia were more labile than wild type94.
The mechanisms by which the Fgf and phosphatidylinositol pathways regulate cilia formation or maintenance remain to be elucidated. It will be informative to investigate whether these pathways have a general role in primary cilia formation or act in a subset of specialized cilia.
Recent experiments argue that there is a close connection between components of planar cell polarity (PCP) signaling and cilia positioning. An excellent example of this connection is found in the mechanosensory hair cells in the organ of Corti in the cochlea. The primary cilium of the hair cell, called the kinocilium, is always oriented on the lateral side of the developing cell. The position of the kinocilium determines the polarity of the chevron of stereocilia (actin-based sensory organelles) on the hair cell. Core components of the non-canonical Wnt pathway, a PCP pathway, are required for the polarity of these hair cells95–97,98.
When primary cilia are removed by conditional deletion of Ift88, the polarity of the hair cells is disrupted99, similar to the phenotype seen in non-canonical Wnt mutants97. This finding indicated that the presence of the kinocilium is important for the correct orientation of the stereocilia and the organization of the hair cells, and raised the possibility that the primary cilium might regulate the non-canonical Wnt pathway in this tissue; however the relationship between the kinocilium and planar polarity is more complex. As in Drosophila, components of the non-canonical Wnt pathway are planar polarized in hair cells, and that polarity is both required for PCP signaling and provides a readout of effective PCP signaling97, 100. In the hair cells of Ift88 conditional mutants, the polarity of PCP proteins is not disrupted. This indicates that IFT88, and presumably cilia, are not required for the activity of the core PCP pathway in this tissue, as in early embryos. Instead, it appears that one output of the non-canonical Wnt signaling is to control the position of the basal body and thereby cilia position99 (Fig. 4a). In addition, the findings indicate that IFT88 itself must be required to reposition the basal body to a polarized position. The mechanisms by which the position of the basal body is regulated by IFT88 are not known.
Studies on the motile cilia on the epidermis of Xenopus embryos support the hypothesis that components of the planar polarity pathway control polarized organization of cilia. These cells are multiciliated and the cilia on each cell share a common polarity101. Disruption of the activity of the PCP protein Disheveled (Dvl) disrupts the polarity of the cilia on these cells. Dvl localizes to the basal bodies of epidermal cells and Dvl morphants have a reduced number of short cilia101; this finding indicates that apical docking of basal bodies, and therefore the ability to form cilia, depends on Dvl and possibly other components of the PCP pathway (Fig. 4b).
While changes in the position of cilia are unlikely to influence their ability to transduce Hh signals, some PCP components do affect Hh signaling. Inturned and Fuzzy are downstream effectors of the non-canonical Wnt pathway in Drosophila, and morpholino knockdown of Xenopus Inturned or Fuzzy disrupts both the apical actin network and cilia formation102. These morphants fail to undergo normal convergent extension due to defects in PCP and also display defects consistent with a loss of Shh signaling102. Similarly, mouse Fuz and Inturned mutants have short cilia and disrupted Hh signaling103–105. Thus, components of the planar polarity pathway can be important for both formation and polarity of cilia (Fig. 4b). There is no evidence, however, that other components of the non-canonical Wnt pathway are required for ciliogenesis, suggesting that the roles of Inturned and Fuzzy in cilia formation may be unrelated to non-canonical Wnt signaling.
Numerous human disorders have now been linked to defects in cilia structure or in cilia localized proteins. These include autosomal dominant polycystic kidney disease (PKD), and recessive pleiotropic disorders such as Bardet-Biedl syndrome, Joubert syndrome, Meckel syndrome, and Ellis-van Creveld syndrome. Some aspects of these disorders, such as polydactyly and skeletal abnormalities, are likely due to misregulated Hh signaling, but the molecular basis of other defects, such as cystic kidneys, are not well understood (Box 1). The dysregulated Hh signaling associated with several types of human cancers also depends on cilia. Thus studies on the relationship between cilia and signaling during development have direct implications for human disease.
Inappropriate activation of Shh signaling can cause medulloblastoma and rhabdomyosarcoma, pediatric tumors of the cerebellum and muscle, and is found in all cases of basal cell carcinoma106–108. In addition, growing evidence indicates that Shh signals promote the growth of other types of tumors109, 110. Recent studies demonstrate that cilia regulate Hh signaling in tumors, and that the role of cilia in tumors depends on how the pathway is activated. Expression of activated Smo in the postnatal mouse brain can cause medulloblastoma, but removal of cilia prevents tumor formation111, consistent with the earlier genetic experiments indicating that cilia are required for activity of the pathway at a step downstream of Smo. Constitutively active Gli2 can also cause in medulloblastoma in mice, but only when cilia are removed111. This suggests that Gli2 alone cannot activate tumorigenesis in this cellular context when cilia-dependent Gli3 repressor is also present. Similar results were observed in basal cell carcinomas in mice harboring similar activating Hh pathway mutations: activated Smo caused tumors only in the presence of cilia whereas removal of cilia enhanced tumorigenesis caused by expression of activated Gli2112. Thus, in tumors as in development, cilia have both positive and negative effects on the Hh pathway.
A hallmark of many human ciliopathies is the formation of kidney cysts, which often begins during fetal life, and is a developmental, rather than physiological, defect3, 113. This raises the question of whether kidney cysts result from a disruption of cilia-dependent developmental signals. Hh signaling is required for normal kidney development114, but kidney cysts have not been reported in mutants that lack either positive or negative Hh regulators115, 116. Although a role for Hh signaling in PKD cannot be ruled out, some data suggest that cilia might modulate Wnt signaling in this tissue117.
The connection between cilia, cystic kidneys and Wnt signaling was first raised by analysis of the inversin (inv) gene. Inv binds microtubules, and localizes to the basal body118 and cilium119. Mutations in inv cause kidney disease in humans120, and renal cysts in mice118. Inv interacts with Dvl, targeting membrane-bound Dvl for destruction. Inv has been proposed to act as a switch between canonical and non-canonical Wnt signaling, based on cell culture and morpholino knock-down experiments68. However, altered canonical Wnt signaling has not been reported in the kidneys of inv−/− mice.
Mouse overexpression experiments show that increased Wnt signaling can cause kidney cysts, and increased nuclear β-catenin is observed in the cystic kidney tubules of mice in which cilia have been conditionally ablated121, 122. However, reduced Wnt signaling has been observed in mice lacking Jouberin (Jbn), a cilia-localized protein mutated in a form of Joubert syndrome: mice that lack Jbn (Ahi1−/−) have cystic kidneys with reduced expression of a Wnt reporter and reduced nuclear β-catenin123. Thus additional experiments that examine Wnt signaling during kidney development will be required to reconcile the conflicting data.
Recent theories of PKD have focused on the importance of the plane of cell division, under the control of PCP signaling, as a possible underlying defect in kidney cysts124. The elongation of kidney tubules is thought to depend on oriented cell divisions, and this is disrupted in kidney tubule cells of mice with cystic kidneys117, 124. Supporting this hypothesis, mice lacking the PCP protein Fat4 exhibit polycystic kidneys beginning at E16 associated with misorientation of mitotic spindles within the renal tubules125. This cyst formation is enhanced by removal of one or both copies of a core PCP pathway component Vangl2. Moreover, Fat4 localizes to cilia within the kidney, implicating the cilium in the modulation of the kidney PCP pathway125. Drosophila Fat, however, acts in a PCP pathway that does not depend on non-canonical Wnt signaling126. Moreover, mouse mutants of other PCP pathway components such as Vangl2 and Fuzzy have not been reported to have cystic kidneys, and the polarity of PCP component localization in kidney tubule cells has not been assessed in mice lacking renal cilia. Recent results suggest that misoriented cell division is neither necessary nor sufficient for the formation of kidney cysts113. Thus the pathway (or pathways) downstream of primary cilia that are misregulated to cause renal cysts remain to be determined.
Non-motile primary cilia play vital roles in vertebrate development from early stages of embryonic patterning, when they regulate the activity of the Hh pathway, to organogenesis, when they are important in the development and homeostasis of numerous tissues. The recent resurgence in interest in primary cilia has raised many new questions about the roles of cilia.
We know very little about the events of Hh signal transduction that occur within cilia. The mechanisms that traffic Smo to cilia, Ptch out of cilia, and that modulate the trafficking of Gli proteins within cilia in response to Hh pathway activation are all unknown. The opposing effects of different IFT components upon the regulation of the Hh pathway suggest that, in addition to providing a compartment where Hh pathway components are enriched, the IFT machinery plays a more complex role in regulating the pathway, but those roles have not been defined. In order to address these questions, it will be necessary to examine the trafficking of Hh pathway components in real time as well as probe their physical associations with the IFT machinery in wild-type cells as well as in cells mutant for various IFT components.
The dual roles of Kif7 in intraciliary trafficking and in the Hh pathway suggest a reason why vertebrate Hh signaling is tied to cilia, and other proteins may also have dual roles. For example: is there a mammalian kinase that performs functions analogous to those of Drosophila Fu and, if so, does it play roles in both ciliogenesis and Hh signaling? Do other components of the Hh signaling pathway affect the dynamics of ciliary trafficking?
Based on the phenotypes of the numerous IFT mutants characterized to date, it appears that during early vertebrate development, the cilium functions as a Hh-dedicated organelle. However, this does not preclude a requirement for cilia in modulating other signaling pathways in specific tissues later in development. These may include Pdgfrα signaling, signaling through G-protein coupled receptors in specific neurons127, and PCP signaling in the kidney. A particularly interesting question is whether, in specific cell types, cilia are sites where Hh and other signaling pathways are integrated. As complex cross-talk between pathways is vital in regulating cellular responses during development and in disease states like cancer, understanding the function of the cilium as a signaling center will be critical.
This work was funded by NIH grant NS044385 to KVA. SCG is an American Cancer Society postdoctoral fellow.