The Role of Primary Cilia-Associated Phosphoinositide Signaling in Development

Primary cilia are microtube-based organelles that extend from the cell surface and function as biochemical and mechanical extracellular signal sensors. Primary cilia coordinate a series of signaling pathways during development. Cilia dysfunction leads to a pleiotropic group of developmental disorders, termed ciliopathy. Phosphoinositides (PIs), a group of signaling phospholipids, play a crucial role in development and tissue homeostasis by regulating membrane trafficking, cytoskeleton reorganization, and organelle identity. Accumulating evidence implicates the involvement of PI species in ciliary defects and ciliopathies. The abundance and localization of PIs in the cell are tightly regulated by the opposing actions of kinases and phosphatases, some of which are recently discovered in the context of primary cilia. Here, we review several cilium-associated PI kinases and phosphatases, including their localization along cilia, function in regulating the ciliary biology under normal conditions, as well as the connection of their disease-associated mutations with ciliopathies.

The primary cilium is a sensory organelle presenting on the surface of most mammalian cell types [19]. Although the ciliary membrane is reported to be continuous with the plasma membrane, the lipid and protein composition in the ciliary membrane is maintained unique and distinguishable from the plasma membrane [20,21]. To fulfill its function as a signaling hub and cellular antenna, the trafficking of protein and membrane components in and out of cilia is highly active and tightly controlled, suggesting the potential involvement of PIs in cilia. Indeed, it was reported in 2009 that mutations in an inositol polyphosphate 5-phosphatase, INPP5E, are causal for multiple ciliopathies [22,23]. After Subsequently, more studies implicated the connection between PIs and primary cilia. PIs are reported to regulate the stability/disassembly of primary cilia [24,25], ciliogenesis [26], and ciliary trafficking [27,28]. The dysfunction of multiple PI enzymes, both phosphatases, and kinases, is connected to ciliary defects and ciliopathies [6]. The localization and function of PIs in cilia have been reviewed in several excellent recent reviews [6,29,30]. Here, we mainly focus on the role of several cilia-associated components of the PI signaling pathway in development, tissue homeostasis, and related diseases.

Structure of Primary Cilia
The primary cilium is a sensory organelle presenting on the surface of most quiescent cells in vertebrates [19,31]. The primary cilium is composed of a microtubule-based core structure called the axoneme, which is surrounded by a ciliary membrane that is continuously extended from the plasma membrane ( Figure 2) [32]. Different from motile cilia, which can generate a fluid flow on cells such as respiratory epithelial and ependymal cells, primary cilia only contain nine microtubule doublets without two centrally located singlet microtubules ( Figure 2) and are non-motile for lacking the axonemal dynein arms on the outer microtubules [33].
The basal body, transformed from the mother centriole upon serum starvation, nucleates the outgrowth of the axoneme and forms the base of primary cilia. The region between the axoneme and basal body is identified as the transition zone, which is characterized by Y-links connecting the axonemal microtubules to the ciliary membrane [34]. Transition fibers, which are transformed from distal appendages of the mother centriole, are located below the transition zone on the distal end of the basal body [35]. In addition to anchoring the basal body to the plasma membrane, transition fibers function together with the transition zone as the ciliary gate to regulate ciliogenesis and signaling by controlling the selective transport of specific ciliary proteins between the cell body and the cilium [35,36]. Intriguingly, many transition zone and transition fiber proteins are reported to contain phosphoinositide binding domains; however, the physiological significance of their PI binding potential remains unclear [37]. Structure of the primary cilium. A primary cilium is composed of a "9 + 0" microtubule axoneme and basal body complex (labeled in green). A motile cilium has two extra central singlet microtubules, generating a "9 + 2" arrangement (labeled in green). A motile cilium also contains dynein arms for ciliary movement and radial spoke to regulate the motility and motion pattern of motile cilia (labeled in black). Transition zone is characterized by the Y-links, which connect the proximal axoneme to the ciliary membrane. Transition fibers connect the distal end of the basal body to the ciliary membrane. Transition zone and transition fibers together coordinate the ciliary gate function (labeled in red) (created with BioRender.com, accessed on 18 November 2022).

Ciliopathy
Although largely considered a vestige before 1999, the primary cilium is now demonstrated as a signaling center to interpret both chemical and mechanical signals. Major components of multiple signaling pathways specifically localize in primary cilia [38]. These include pathways that are essential for the biogenesis and homeostasis of tissues and organs, such as the Hedgehog signaling [19], canonical Wnt signaling [39], Planar cell polarity [40], G protein-coupled receptor signaling [41], as well as tyrosine kinase receptor signaling [42] pathways. The activation and regulation of these pathways occur in cilia and depend on the normal structure and functionality of primary cilia [19].
Consistently, defects in primary cilia lead to a panel of genetic human diseases, collectively termed ciliopathies, including numerous seemingly unrelated syndromes, with involvement of the brain, eye, kidney, liver, pancreas, skeletal system, and some others in a growing list [43][44][45][46][47][48]. Common ciliopathies range from organ-specific disorders such as autosomal dominant polycystic kidney disease (ADPKD) and autosomal recessive PKD (ARPKD) to pleiotropic disorders such as Joubert syndrome (JBST), nephronophthisis (NPHP), Bardet-Biedl syndrome (BBS), and Meckel syndrome (MKS) [49,50]. Currently, more than 190 established and 250 candidate cilium-associated proteins are identified as ciliopathy proteins and over 35 ciliopathies are reported, emphasizing the significance of primary cilia in development and human disease [51,52]. Given the physiological significance of the primary cilium, the molecular mechanisms underlying its biogenesis, function, and related disease pathogenesis have just started to be explored in the recent two decades. Figure 2. Structure of the primary cilium. A primary cilium is composed of a "9 + 0" microtubule axoneme and basal body complex (labeled in green). A motile cilium has two extra central singlet microtubules, generating a "9 + 2" arrangement (labeled in green). A motile cilium also contains dynein arms for ciliary movement and radial spoke to regulate the motility and motion pattern of motile cilia (labeled in black). Transition zone is characterized by the Y-links, which connect the proximal axoneme to the ciliary membrane. Transition fibers connect the distal end of the basal body to the ciliary membrane. Transition zone and transition fibers together coordinate the ciliary gate function (labeled in red) (created with BioRender.com, accessed on 18 November 2022).

Ciliopathy
Although largely considered a vestige before 1999, the primary cilium is now demonstrated as a signaling center to interpret both chemical and mechanical signals. Major components of multiple signaling pathways specifically localize in primary cilia [38]. These include pathways that are essential for the biogenesis and homeostasis of tissues and organs, such as the Hedgehog signaling [19], canonical Wnt signaling [39], Planar cell polarity [40], G protein-coupled receptor signaling [41], as well as tyrosine kinase receptor signaling [42] pathways. The activation and regulation of these pathways occur in cilia and depend on the normal structure and functionality of primary cilia [19].
Consistently, defects in primary cilia lead to a panel of genetic human diseases, collectively termed ciliopathies, including numerous seemingly unrelated syndromes, with involvement of the brain, eye, kidney, liver, pancreas, skeletal system, and some others in a growing list [43][44][45][46][47][48]. Common ciliopathies range from organ-specific disorders such as autosomal dominant polycystic kidney disease (ADPKD) and autosomal recessive PKD (ARPKD) to pleiotropic disorders such as Joubert syndrome (JBST), nephronophthisis (NPHP), Bardet-Biedl syndrome (BBS), and Meckel syndrome (MKS) [49,50]. Currently, more than 190 established and 250 candidate cilium-associated proteins are identified as ciliopathy proteins and over 35 ciliopathies are reported, emphasizing the significance of primary cilia in development and human disease [51,52]. Given the physiological significance of the primary cilium, the molecular mechanisms underlying its biogenesis, function, and related disease pathogenesis have just started to be explored in the recent two decades.

PI-Binding Ciliary Proteins
It has been well documented that PIs are important regulators of membrane trafficking, protein transportation, and protein complex assembly in various subcellular locations [4,5,21], which are key events for the ciliary signaling. Similarly to in the cytoplasm, PIs exhibit different functions in cilia in different subdomains, and this spatiotemporal availability/functionality of specific PI species is achieved by the specific ciliary localization of relevant phosphatases, kinases, as well as PI-binding effectors. Among all the reported PI-binding domains, PH, C2, Tubby and B9 domains are found in primary cilium-associated proteins [28,37,[61][62][63].
TULP3 is a PtdIns(4,5)P 2 -binding protein with a conserved C-terminal Tubby domain and a conserved N-terminal helix that interacts with the intraflagellar transport complex IFT-A [67]. TULP3 is required for the ciliary localization of multiple membrane-associated proteins [67][68][69][70]. GPR161 is a Gαs-coupled receptor that suppresses the Hedgehog pathway by increasing cAMP and PKA activity [68,71]. In Inpp5e KO cells, accumulated PtdIns(4,5)P 2 in cilia attracts more TULP3 and subsequently more GPR161 in cilia for the repression of the Hedgehog signaling [27,28]. Reducing TULP3 levels restored the activation of Hedgehog pathway in Inpp5e KO cells, supporting that TULP3 is the key effector influenced by the availability of INPP5E and PtdIns4P [28]. On the other hand, TULP3 is also necessary for the ciliary localization of INPP5E, as well as other proteins bound to the ciliary membrane including the small G-protein ARL13B and ADPKD disease proteins polycystin 1 and polycystin 2 [69,70]. Current studies suggested that the IFT-A-binding motif instead of the PtdIns(4,5)P 2 -binding domain of TULP3 is necessary for the ciliary localization of those membrane proteins in normal conditions [70].
Many transition zone proteins contain C2 domains, such as AHI1, NPHP1, NPHP4, MKS5/RPGRIP1L/NPHP8 and CC2D2A/MKS6 [37,72], which have the potential to bind PIs. However, whether they indeed bind to PIs and/or regulated by PI binding have not been well characterized. MKS-5 acts as an assembly factor for the establishment of the transition zone [73,74]. In Caenorhabditis elegans, it was proposed that the C2 domains of MKS-5 interact with other transition zone proteins, which potentially forms a lipid gate to limit the abundance of PtdIns(4,5)P 2 within cilia [73]. This study showed that PtdIns(4,5)P 2 diffused from the transition zone to the ciliary proper in the mks-5 mutant [73]. The homeostasis of compartmentalized PtdIns(4,5)P 2 is critical for the stability and signaling of cilia [28,55]; thus, it is worth investigating whether and how PtdIns(4,5)P 2 localization is controlled by this MKS5-centered lipid barrier in vertebrates. B9 domains are specific ciliary C2 domains that are identified in MKS1, B9D1/MKS9, and B9D2/MKS10 [37,72,75]. These transition zone proteins form a linear MKS1-B9D2-B9D1 tripartite complex through the binding of B9 domain and are required for the diffusion-barrier function of the transition zone [63,76]. Similarly, the phosphoinositide binding property of these B9 domains is unclear.
Current knowledge from human patients and animal models supports that INPP5E is essential for development. Two human ciliopathy syndromes are associated with INPP5E mutations: Joubert syndrome (JBTS) and MORM syndrome (Tables 1 and 2) [80][81][82][83]. JBTS is an autosomal recessive neurodevelopmental disorder characterized by the appearance of a "molar tooth sign" on axial MRI, which results from the abnormal development of the cerebellar vermis and the brainstem [84]. The most common clinical features of JBTS include ataxia, hyperpnea, sleep apnea, ocular motor apraxia, hypotonia, and cystic dysplastic kidney (Figure 4 and Table 2) [85]. JBTS-associated INPP5E mutations are mainly missense and cluster in the catalytic domain (Table 1). In vitro experiments also confirmed that these muted INPP5E exhibit decreased phosphatase activity and cause impaired PI distribution in cilia [81,83,[85][86][87]. On the other hand, MORM syndrome is resulted from C-terminal deletions of INPP5E (Table 1), which leads to loss of the CAAX motif and exclusion of INPP5E from the ciliary membrane without affecting the catalytic activity [23,88]. Correspondingly, in addition to similar symptoms as JBTS including intellectual disability and retinal dystrophy, MORM syndrome shows phenotypes unseen in JBTS, including obesity and micropenis, as observed in other ciliopathies such as Bardet-Biedl syndrome and Cohen syndrome ( Figure 4 and Table 2) [23,88,89]. Compared with JBTS-associated INPP5E mutants which lead to a loss of INPP5E activity in the whole cell and whole body, MORM-associated INPP5E mutants only damage the INPP5E activity in cilia but may increase the INPP5E activity in other cellular locales due to mislocalization. In this context, the phenotypic differences between JBTS and MORM may be caused by the abnormal non-ciliary activity of INPP5E, which could be seen when other ciliary abnormalities disturb the ciliary targeting of INPP5E.
Results from animal models reinforce the conclusion in human patients that INPP5E function in primary cilia is required for the development of multiple organs [23,24,90]. Inpp5e D/D mice (deletion of exons 7 and 8) are embryonic and postnatal lethal with ciliopathy features, including bilateral anophthalmos, postaxial hexadactyly, renal cyst, skeletal abnormalities, as well as cerebral developmental defects, suggesting that INPP5E is essential for primary cilium-mediated functions [23]. Interestingly, the inactivation of Inpp5e in mice on the postnatal day-28 did not affect the survival of adult mice; however, still caused ciliopathy phenotypes such as higher body weight, retinal dystrophy, and cystic glomeruli [23]. This result supports the involvement of INPP5E in the cilium-dependent homeostasis of mature tissues/organs, whereas also indicates that the INPP5E-dependent ciliary function is more important for the embryonic development. Further studies using the Inpp5e D/D MEFs suggest that the Inpp5e deletion increases the ciliary level of the Hedgehog suppressor GPR161 via PtdIns(4,5)P 2 -dependent recruitment of TULP3, which promotes GLI3R formation and reduces Hedgehog signaling [28]. Consistent with the Inpp5e D/D mice, another mouse model of Inpp5e -/-(deletion of exons 2 to 6) also recapitulates JBTS features, including polycystic kidneys, cleft palate, polydactyly, edema, and ossification delays [23,24,90]. Inpp5e -/-MEFs exhibited the defective Hedgehog signaling with reduced ciliary accumulation of SMO and GLI2 when treated with the SMO agonist [24]. Expression of a constitutively active SMO mutant (SMOM2) in Inpp5e -/mice partially restored the associated embryonic development defects, emphasizing the involvement of Hedgehog signaling in INPP5E regulation in embryonic development [24]. However, whether the expression of SMOM2 can rescue the defects in Inpp5e -/adult mice is not determined in this study. Although most INPP5E-associated JBTS patients mainly show neurologic symptoms with rare kidney or hepatic features, both Inpp5e D/D and Inpp5e -/mice showed polycystic kidneys with a 100% penetrance [23,24]. Moreover, the renal-specific deletion of Inpp5e exons 2-6 in mice resulted in severe polycystic kidneys and renal failure, likely caused by the hyperactivation of PI3K/Akt and mTORC1 pathway [90]. Despite the possibility that INPP5E may carry slightly diverse functions in different species, that the PKD phenotype in Inpp5e-inactivated mice is absent in INPP5E-mutated human patients more likely reflects the dosage difference of functional INPP5E in both models. Mouse models suffer from a complete loss of INPP5E protein during the embryonic and/or postnatal development of kidneys [23,90], whereas in patients, INPP5E mutations are often hypomorphic with reduced protein level, weakened phosphatase activity, or defective ciliary localization [23,83], which may be sufficient to support the development and homeostasis of kidneys. In other animal model such as zebrafish, Inpp5e knockdown in morphants also impairs cilia formation and function in the Kupffer's vesicle and pronephric ducts, thus leads to ciliopathy-like phenotypes including body axis asymmetry, microphthalmia, pericardial edema, kinked tail, and pronephric cyst formation [91]. Moreover, expression of human INPP5E rescues the defects in the Inpp5e knockdown morphants [91], suggesting that the functionality of INPP5E is mostly conserved in vertebrate development.
Especially, INPP5E patient mutations exhibit reduced cilia stability compared with the wild-type protein [23]. Taken together, INPP5E is the most well-studied PI enzyme, which plays a critical role in cilia assembly, stability, and signaling pathways [6,29,92,93], highlighting the importance of INPP5E in development.

OCRL and INPP5B
OCRL and INPP5B are both members of the inositol polyphosphate 5-phosphatase family as INPP5E, but mainly hydrolyze PtdIns(4,5)P 2 to generate PtdIns4P [1]. OCRL localizes to the cilium and basal body, as well as the endocytic network, and functions in the assembly and maintenance of primary cilia (Figure 3) [94][95][96][97][98]. OCRL is a multi-domain protein including PH, 5-phosphatase, ASH (ASPM-SPD2-Hydin), and catalytically inactive RhoGAP (Rho GTPase-activating protein) domains [1]. At the early stage of ciliogenesis, OCRL is recruited by the small GTPase RAB8 to cilia through direct binding [95,99], which is essential for primary cilia assembly [94]. Mutations of OCRL in the 5-phosphatase domain interrupts its ciliary localization, whereas deletion of the RhoGAP domain eliminates OCRL in the ciliary proper and restrains it near the ciliary base [95]. This is consistent with the discovery that patients' fibroblasts with OCRL mutations exhibit defective ciliogenesis and shortened cilia [95,96], indicating the diverse disorders present in OCRL-mutated patients may be due to the dysregulation of primary cilia. As a paralog of OCRL, INPP5B shares similar structural domains [1] but contains a C-terminal CAAX prenylation domain, which is essential for the ciliary localization of INPP5B [100]. Knockdown of INPP5B results in a significant decrease of ciliogenesis and ciliary length in both cultured mammalian cells and zebrafish Kupffer's vesicle, but the detailed mechanism remains unclear [100].
OCRL mutations are identified as causative of two human diseases: oculocerebrorenal syndrome of Lowe (OCRL), also called Lowe syndrome, and Dent-2 disease (Tables 1 and 2) [101][102][103][104][105]. Both Lowe syndrome and Dent-2 disease are rare X-linked genetic disorders. Lowe syndrome patients exhibit defective cilium assembly and ciliopathy symptoms such as intellectual disability, congenital cataracts, and renal dysfunction ( Figure 4) [105]. Dent-2 is often described as a milder form of Lowe syndrome, as most patients only develop renal symptoms and a few have mild intellectual disability, hypotonia, cataracts, and rickets ( Figure 4) [101]. Genetic analyses showed that Lowe syndromeassociated mutations are mostly in exons 8-23 (5-phosphatase, ASH, and RhoGAP domains), whereas Dent-2-associated mutations are typically in exons 1-7 (PH domain) (Table 1) [101,104], suggesting that the distinct symptoms in these two disorders result from different mutations. Interestingly, one study using different Lowe and Den-2 fibroblasts displays similar reduced protein and accumulated PtdIns(4,5)P 2 levels but milder ciliogenesis defects in Den-2 [96], indicating the ciliation defect may be related to the disease severity. Moreover, the epigenetic differences among individual patients and/or secondary genetic mutations should be considered because the disease severity varies widely between patients who carry the same OCRL mutation [105,106]. However, INPP5B has not been associated with any human diseases. Thus, although INPP5B's function may be partially redundant with OCRL [107], OCRL is the dominant enzyme accountable for corresponding developmental needs in human.
Results from animal studies suggest that mice respond differently from humans to the loss of OCRL [107,108]. Ocrl knockout mice are fertile with normal kidneys, eyes, and brains, failing to recapitulate the phenotypes of Lowe syndrome [107]. However, Ocrl;Inpp5b double knockout mice are embryonic lethal [107] and the kidney-tubulespecific deletion of Inpp5b in Ocrl −/− mice phenocopy the tubulopathy disorder of Lowe Syndrome/Dent-2 [109]. These results indicate that Ocrl and Inpp5b carry redundant functions in mice, and normal mouse development can be conducted if the combined enzyme activity of Ocrl and Inpp5b is above certain threshold. Once levels of Ocrl and Inpp5b drop below this threshold, the severity of developmental defects negatively correlates with the remaining dosage of functional Ocrl and Inpp5b. Interestingly, knock-in of human INPP5B in Ocrl;Inpp5b mice corrected the lethality, but animals still showed phenotypes such as Lowe syndrome/Dent-2 disease including reduced postnatal growth, low molecular weight proteinuria, and aminoaciduria [108]. This is consistent with the previous discovery that although INPP5B and Inpp5b are highly conserved in most exons, the significant differences in exons 7 and 8 lead to different gene transcription, mRNA splicing, and primary protein sequence of between human and mouse [110]. Moreover, one study showed that Inpp5b expression level was dramatically higher in mouse trabecular meshwork cells than the same human cells [100]. This distinct expression may partly explain why Inpp5b and INPP5B compensate the loss of Ocrl in mice differently.
Although several studies have confirmed the ciliary localization of OCRL and INPP5B [94,95,100], the investigation of OCRL and INPP5B in cilia has just begun. OCRL can be recruited to cilia by RAB8, which is essential for ciliary assembly, and regulate ciliary protein trafficking in an Rab8-and endosome-dependent manner [94,95]. Knockdown of Ocrl or Inpp5b in zebrafish resulted in defective cilium formation in Kupffer's vesicle and ciliopathy-like phenotypes including microphthalmia, body-axis asymmetry, microlens, distorted retinas, and hydrocephalus [95,100]. Double knockdown of both Ocrl and Inpp5b in zebrafish showed synergistic effects, suggesting that these two ciliary PI phosphatases may play some non-redundant function in zebrafish development [100]. Consistent with the observation in zebrafish, the lack of OCRL in human retinal pigmented epithelial cells and patients' fibroblasts results in defective ciliogenesis and shortened cilia [94,95]. However, one study using MDCK cells demonstrated an increased ciliary length when knocking down OCRL [98], indicating a potential cell-specific function of OCRL. Further investigation on the molecular mechanism underlying these differences may help the understanding of the tissue-specific manifestations in Lowe syndrome/Dent-2 disease. As PI phosphatases, how OCRL and INPP5B regulate the disruption of ciliary PIs is also unclear. One study using Lowe syndrome patients' fibroblasts and Ocrl-null MEFs shows increased PtdIns(4,5)P 2 and decreased PtdIns4P in cilia, which is similar to the observation in Inpp5e-null MEFs [97]. Intriguingly, OCRL is also necessary for the activation of Hedgehog signaling, but through a different mechanism from INPP5E. Unlike INPP5E that suppresses the ciliary entry of GPR161 but not SMO [27,28], OCRL deficiency has no effect on GPR161, but disrupts the ciliary translocalization of SMO upon SAG treatment [97]. The underlying molecular mechanism and involvement of corresponding PI species is highly interesting and should be determined in future studies.
Homozygous global knockout of Pik3c2a (deletion of exon 1) in mice leads to embryonic lethality between E10.5 and E11.5 with abnormalities in multiple organs [119]. In addition to the defective angiogenesis and vascular maturation observed in Pik3c2a -/embryos [119], some developmental disorders typically detected in mice with deficient Hedgehog signaling were identified, such as disrupted cardiac tube looping and impaired left-right patterning [58,120,121]. Consistent with the observation, Pik3c2a -/-MEFs and embryos exhibit impaired cilium elongation and ciliary reduction in the Hedgehog signal transducer SMO [58]. Furthermore, the loss of one Pik3c2a allele worsened the renal cystic burden in two PKD mouse models (Pkd2 +/and Pkd1 +/-) [59]. In vitro studies using MEFs and IMCD3 cells also confirmed the impaired ciliary targeting of polycystin-2, encoded by Pkd2, in PI3K-C2α-depleted cells [59], supporting an essential role of PI3K-C2α in cilia in mouse development.
Recently, a novel human syndrome with short stature, cataracts, secondary glaucoma, and skeletal malformations has been identified in homozygous loss-of-function mutations in PIK3C2A (which encodes PI3K-C2α) ( Table 3) [122]. This syndrome shares many similar features to classical ciliopathy syndromes and is especially close to the Lowe syndrome caused by mutations in the PI 5-phosphatase OCRL, as discussed above [122]. Similar to the observation in mice models, patients' fibroblasts exhibited decreased PtdIns3P and RAB11 levels at the ciliary base, as well as reduced cilia length, suggesting that PIK3C2A is a candidate ciliopathy gene [122]. Further identification of additional patients will improve our understanding of the genotype-phenotype correlation associated with PIK3C2A mutations. Although PI3K-C2α can generate both PtdIns3P and PtdIns(3,4)P 2 , current evidence suggests that PI3K-C2α regulates cilia by producing PtdIns3P at the PC-REC [58,59]. It was proposed that PtdIns3P activates the regional RAB11/RAB8 cascade and regulates the trafficking of membrane proteins such as SMO and polycystin 2 into cilia [58,59]; however, the underlying molecular mechanisms remain to be illustrated. Nevertheless, the ciliopathy phenotypes observed in mouse models and human patients endorse the importance of PI3K-C2α in the context of primary cilia and cilia-dependent developmental events [58,59,122].

PIPKIγ
Type Iγ phosphatidylinositol-4-phosphate 5-kinase (PIPKIγ) is the main kinase generating PtdIns(4,5)P 2 by phosphorylating PtdIns4P (Figure 1) [1]. PIPKIγ is encoded by PIP5K1C. Currently, six alternative splicing isoforms of PIPKIγ have been identified that differ from each other solely by the C-terminal extension sequences [123]. Only isoform 3 with a unique motif at the C-terminus targets the proximal ends of centrioles or the basal body (Figure 3) [123]. While suppressing the centriole amplification in proliferating cells, PIPKIγ in non-duplicating cells regulates ciliogenesis in a kinase-dependent manner [26,124]. In proliferating cells, a centrosomal PtdIns4P pool generated by INPP5E inhibits the recruitment of tau tubule kinase 2 (TTBK2) and the subsequent removal of microtubule capping protein CP110 from the distal end of the mother centriole, thus blocks the initiation of axoneme assembly [26]. Signals promoting ciliogenesis triggers the removal of INPP5E from the mother centriole, leaving the centrosomal PtdIns4P pool to be exhausted by PIPKIγ, promoting the TTBK2 recruitment and the downstream axoneme elongation [26,124].
Although PIPKIγ appears indispensable for ciliogenesis, it also plays essential role in a wide range of cellular events at various subcellular locales such as focal adhesion assembly and endocytic trafficking, likely via different splicing variants [123,[125][126][127][128]. Up to date, the only reported human developmental disorder associated with PIPKIγ is the lethal congenital contractural syndrome type 3 (LCCS3), a severe form of arthrogryposis [129]. Caused by a single homozygous mutation in PIP5K1C (PIPKIγ p.D253N, a kinase-dead mutation), LCCS3 patients exhibit multiple joint contractures, severe muscle wasting, and atrophy (Table 3) [129], sharing similar features observed in LCCS10, which is in the same phenotypic series and defined as a ciliopathy [130]. However, the LCCS3-associated PIP5K1C mutation affects all splicing isoforms of PIPKIγ. Whether and to what extent the cilia-associated PIPKIγ isoform 3 contributes to the pathogenesis of LCCS3 remain to be determined. Future studies in the context of cilia using patient-derived fibroblast or epithelial cells may provide useful evidence to answer these questions. Similar to LCCS3 patients, Pip5k1c-interrupted mice are embryonic or postnatal lethal with severe developmental defects. Among three different Pip5k1c-KO mouse models [131][132][133], Pip5k1c knockout mice with a fusion of β-Gal to the first 32 amino acids of PIPKIγ were embryonic lethal at E10.5 and exhibited ciliopathy-like phenotypes including exencephaly and pericardial effusion [131]. However, the other two Pip5k1c KO mice models with deletion of exons 2-6 or deletion of exon 18 died within 24 h after birth, with no apparent ciliopathy-like phenotypes observed in this time frame [132,133]. This is not surprising because all three Pip5k1c editing approaches affect all PIPKIγ equally. To understand the cilium-dependent functionality of PIPKIγ in development, it is necessary to generate mouse models in which individual PIPKIγ isoform is inactivated in specific ciliated tissues/organs. In addition, we recently showed that hydrocephalus syndrome protein 1 (HYLS1), a ciliopathy protein, functions as a PIPKIγ activator at the ciliary base. In cultured renal epithelial cells, both PIPKIγ and HYLS1 are necessary for the assembly of NPHP module at the transition zone and promote the removal of ciliary GPR161 and activation of the Hedgehog pathway [124]. Surprisingly, the kinase activity of PIPKIγ is vital for HYLS1 to regulate the axoneme elongation but appears dispensable for the ciliary trafficking of Hedgehog signaling components, indicating that the PIPKIγ-HYLS1 axis plays multiple roles in the context of primary cilia [124].

Unanswered Questions and Perspectives
Comparing to what has been known for PIs in the plasma membrane and various cytosolic membrane compartments, the role of each PI species in the context of primary cilia just started to emerge. In addition to the PI species, PI kinases and phosphatases, and PI effectors we discussed above, recent studies revealed more PI metabolic enzymes that may function in the context of primary cilia [134][135][136]. For example, the PI 3-phosphatase and tensin homolog (PTEN) (Figure 1) [1], which is mainly identified as a potent tumor suppressor [137], also functions significantly in multicilia formation and cilia disassembly by controlling the phosphorylation of the WNT signaling component Dishevelled in Xenopus [134]. Loss of phosphatidylinositol 4-kinase β (PI4KB), the main generator of PtdIns4P in the Golgi membrane ( Figure 1) [1], in zebrafish led to the absence of primary cilia and ciliopathy phenotypes including neuromasts, pronephric ducts, and edema [135]. Moreover, the Src homology-2-domain-containing inositol-5-phosphatase SHIP2 was reported to increase the ciliary length and stability by confining AURKA at the basolateral membrane in polarized MDCK cells that formed spherical acini/cysts in 3D Matrigel, which is important for the lumen formation during the morphogenesis of epithelial tubules [136]. Future studies are needed to determine the mechanistic connection of these PI enzymes with the primary cilia and cilia-dependent cellular and biological processes that influence the development and tissue/organ homeostasis.
PIs are not only essential signaling components in various cell membranes, but also critical structural components to retain specific PI-binding proteins and determine the identity of these membrane domains/compartments [2,3,5]. Due to the interconvertibility, PIs together with PI kinases and phosphatases at the interface between different membrane compartments are essential integrators of membrane dynamics and facilitate the exchange/trafficking of proteins between membrane compartments [5]. In the context of primary cilia, PIs should act following similar mechanistic principles, which is proven true for the dynamics of PtdIns4P and PtdIns(4,5)P 2 in the context of cilia [27,28]. However, systemic studies are needed to understand whether, which, and how PIs contribute to the transport of proteins and lipids from the endosomal membrane compartments to pericentriolar vesicles surrounding the basal body, through the ciliary gate, in the ciliary membrane to the ciliary tip, and then back to the ciliary base for recycling or activating downstream signaling molecules. To this end, determination of the precise localization and dynamics of PIs, as well as PI enzymes and effectors, at specific ciliary subdomains is critical. Although PI biosensors and antibodies were successfully employed to visualize various PI species and PI-binding proteins in cytoplasmic membranes, their application in cilia is harder due to the low abundance of PIs, the highly selective ciliary targeting of exogenous proteins, or the unique physicochemical property of primary cilia which is unfriendly to common immunostaining and imaging approaches [138][139][140][141]. Utilizing super-resolution microscopy, a recent study reported that PtdIns(3,4,5)P 3 and PtdIns(4,5)P 2 were visualized using validated antibodies in a ring shape at distinct subdomains of the inner transition zone membrane [53], indicating that more advanced microscopy imaging techniques are necessary for studying the sub-ciliary localization of PIs and PI-binding proteins.
In addition to identify more cilium-associated PIs and PI enzymes, it is vital to define more PI effectors. Many cilium-associated proteins are identified with PI-binding domains such as PH, C2, Tubby, and B9 domains [53,61,62,[66][67][68][69][70]142,143]. Potential PI-binding motifs are also suggested in the cytoplasmic domains of ciliary transmembrane proteins, such as polycystins [144], via in vitro lipid binding assays. However, it has not been confirmed whether and to which PI species these proteins bind in vivo, and the consequent physiological significance of binding to these PIs remain unclear. In the future, it is vital to develop new tools, from novel biosensors and imaging methods to new mutations and animal models, to answer these questions. These studies will yield adequate knowledge to complete the route map of PI signaling in the context of primary cilia and development.