Roles of the actin cytoskeleton in ciliogenesis

The primary cilium, present in most vertebrate cells, is an essential sensor and regulator of signaling in response to extracellular cues (Goetz and Anderson, 2010; Wheway et al., 2018; Anvarian et al., 2019). Defects in primary ciliogenesis cause genetic multisystemic diseases termed ciliopathies (Fliegauf et al., 2007; Hildebrandt et al., 2011; Mitchison and Valente, 2017; Reiter and Leroux, 2017). Primary cilia nucleate from the mother centriole, also called a basal body, and this requires extension of axoneme microtubules, membrane remodeling to sheath the axoneme and creation of a boundary, called the transition zone, which restricts access to and from the cilium, creating a unique cellular compartment (Pedersen et al., 2008; Reiter et al., 2012). In the past decade, numerous studies have started defining the molecular machinery mediating formation of primary cilia. However, most of these studies use the retinal pigment epithelial (RPE-1) cell line. Although RPE-1 cells are a powerful experimental model to study primary ciliogenesis, they do not always recapitulate ciliation in differentiated and polarized cells in vivo, especially in polarized epithelial cells (Molla-Herman et al., 2010; Bernabé-Rubio and Alonso, 2017). Although many pathologies in a variety of ciliopathies appear to be mediated by ciliation defects in polarized epithelia (Fliegauf et al., 2007; May-Simera et al., 2018; McConnachie et al., 2021), we are only beginning to understand the molecular machinery that regulates cilia formation and subsequent signaling in epithelial cells, where cilia are assembled exclusively at the apical pole. Indeed, there are several outstanding questions regarding ciliation in polarized epithelial cells. First, we still do not understand why cilia only form apically in epithelial cells and how ciliation cross-talks with the machinery mediating epithelial polarization. Second, we are only beginning to define how the basal body crosses the very dense apical cortical actin network to dock at the apical plasma membrane. Finally, specialized subsets of epithelial cells, such as bronchial epithelial cells, form hundreds of motile cilia that play a key role in clearing mucus from the airways (Brooks and Wallingford, 2014; Mitchison and Valente, 2017; Spassky and Meunier, 2017). The number of these motile cilia appears to correlate to the size of the apical surface, yet, how these cilia form remain to be defined. In this Review, we will overview the latest findings regarding cilia formation in epithelial cells, and the coordination of ciliation with epithelial polarization and apical cortical actin dynamics.

Two major pathways for ciliogenesis have been identified – an ‘extracellular’ route, which appears to be used by most polarized epithelial cells and produces cilia protruding directly from the cell surface, and an ‘intracellular’ route, which is used by most other cell types and produces cilia rooted in membrane invaginations termed ‘ciliary pockets’ (Molla-Herman et al., 2010; Sorokin, 1962, 1968) (see Box 1). The distinction between these cell-type-dependent ciliogenesis pathways, resulting in either the presence or absence of a ciliary pocket, has implications for ciliary signaling, which are not yet fully understood. The ciliary pocket is an active domain for clathrin-mediated endocytosis (Molla-Herman et al., 2010; Rattner et al., 2010), with a role in transforming growth factor (TGF)-β signaling (Clement et al., 2013), and is also a site of connections to the actin cytoskeleton (Molla-Herman et al., 2010; Rattner et al., 2010); it has been proposed that actin-mediated remodeling of the ciliary pocket could serve to regulate the exposure of the cilium to extracellular signals (Rattner et al., 2010). Cilia that are buried in a pocket cannot respond to fluid flow (Mazo et al., 2016), suggesting that the extracellular route of ciliogenesis is important for the mechanosensory function of epithelial primary cilia. Defects in the cell-type-specific machinery required for either intracellular or extracellular ciliogenesis might account for some of the diverse clinical manifestations of ciliopathies – that is, why different ciliopathy-associated mutations cause phenotypes in different sets of organ systems (Hildebrandt et al., 2011; Reitner and Leroux, 2017) – making it important to characterize which ciliogenesis factors are cell type specific.

Most of the core cellular machinery required for primary ciliogenesis is shared between the intracellular and extracellular routes, and this core ciliogenesis machinery has been well reviewed elsewhere (Bernabé-Rubio and Alonso, 2017; Sánchez and Dynlacht, 2016; Shakya and Westlake, 2021; Labat-de-Hoz et al., 2021); thus, in this Review, we will focus primarily on cilia formation in polarized epithelial cells. It remains unclear whether the intracellular and extracellular pathways represent fundamentally different mechanisms for ciliogenesis, or whether they can be better understood as variations of the same mechanism but occurring with different kinetics and at different subcellular locations. A few specific requirements for intracellular ciliogenesis have been reported. For instance, PACSIN proteins appear to be required for ciliary vesicle formation and for connecting the ciliary sheath to the plasma membrane in the intracellular pathway, but they do not seem to affect ciliogenesis in a kidney epithelial cell line that is thought to use the extracellular pathway (Insinna et al., 2019). Similarly, two recent studies have found that Rab34 is required for ciliary vesicle formation and ciliogenesis in several cell lines that use the intracellular route, although it is dispensable for ciliogenesis in kidney epithelial cells (Ganga et al., 2021; Stuck et al., 2021). Rab34 has also been observed to be important for ciliogenesis in epithelial cells plated at low density, in which case some cilia are assembled intracellularly, but not when the epithelial cells are plated at high density, in which case cilia are assembled extracellularly (Stuck et al., 2021). This is consistent with a previous report (Mazo et al., 2016) showing that bronchial epithelial cells, prior to forming a polarized monolayer, often produce cilia with deep ciliary pockets, while a polarized monolayer of the same cell line produce cilia without ciliary pockets. Thus, the choice of ciliogenesis pathway likely depends not only on cell type but also on aspects of the cellular context. Rab34-knockout mice (Dickinson et al., 2016; Xu et al., 2018) show tissue-specific and highly variable ciliopathy phenotypes, consistent with Rab34 mediating a cell-type- and cell-context-specific step of ciliogenesis in vivo.

There are also some factors that have been shown to be important for ciliogenesis in polarized epithelial cells and which might represent a specialized machinery for the extracellular ciliogenesis route. For example, the exocyst complex, which functions in polarized membrane trafficking, is important for ciliogenesis in kidney epithelial cells (Zuo et al., 2009), whereas in RPE-1 cells, which use the intracellular pathway, exocyst proteins are dispensable for cilia formation but regulate recycling of internalized cilia to the cell surface (Rivera-Molina et al., 2021). The Par and Crumbs protein complexes, which regulate apico-basal polarity of epithelial cells, are also important for ciliogenesis in polarized epithelial cells (Fan et al., 2004, 2007; Sfakianos et al., 2007), as is the post-mitotic midbody remnant (Bernabé-Rubio et al., 2016). To our knowledge, it is not known whether Par and Crumbs proteins or midbody remnants participate in intracellular ciliogenesis, but since these factors function in establishing cell polarity (for a review of the role of midbody remnants in cell polarity, see Peterman and Prekeris, 2019), it is plausible that they might be specifically required for the polarized-epithelia-specific extracellular ciliogenesis pathway.

In short, the primary cilia of polarized epithelia appear to be assembled at the apical plasma membrane and protrude directly from the cell surface, whereas primary cilia of other tissues are assembled intracellularly and remain partially submerged in a ciliary pocket, and this difference is likely important for the tissue-specific signaling functions of the cilia. There is recent evidence for a few factors being specifically required for one or the other of these pathways; however, many studies characterizing requirements for ciliogenesis rely on one model cell line and have not tested whether the mechanisms identified are conserved in cells that use the other ciliogenesis route. Moreover, considering the diverse phenotypes of ciliopathies, it is likely that there are additional cell-type-specific aspects of ciliogenesis beyond this division between extracellular and intracellular pathways.

Recent studies have identified the actin cytoskeleton as an important regulator of ciliogenesis. This was a somewhat unexpected finding since actin had not been observed inside cilia and traditionally microtubules were thought of as the key regulators of cilia formation and function. Nevertheless, although the function of actin during ciliogenesis remains to be fully understood, there are several lines of evidence indicating that primary ciliogenesis, especially in polarized epithelial cells, is closely tied to actin dynamics. First, many actin-binding proteins have been identified as regulators of primary ciliogenesis; in particular, factors that promote branched actin assembly tend to inhibit ciliogenesis (Bershteyn et al., 2010; Cao et al., 2012; Failler et al., 2021; Kim et al., 2010a). Second, mutations in known ciliopathy genes cause dysregulation of the actin cytoskeleton (Bozal-Basterra et al., 2020; Dawe et al., 2009; Hernandez-Hernandez et al., 2013; Valente et al., 2010). Third, pharmacological perturbation of actin polymerization promotes ciliogenesis (Bershteyn et al., 2010; Kim et al., 2010a; Nagai and Mizuno, 2017; Sharma et al., 2011). In this Review, we summarize the latest data connecting actin dynamics and ciliogenesis, with special focus on epithelial ciliation.

Among the first indications that actin dynamics regulate primary cilia formation came from a functional genomic screen to identify modulators of ciliogenesis (Kim et al., 2010a), which used RNA interference (RNAi) against a library of 7784 druggable genes in RPE-1 cells. Multiple actin-binding proteins were identified as positive or negative regulators of ciliogenesis. Among those confirmed as positive regulators of ciliogenesis were several factors that disrupt actin filaments and increase actin dynamics, such as two members of the gelsolin family of actin-severing proteins, whereas negative regulators included factors that stabilize or nucleate branched actin, such as Arp3, a component of the branched-actin-nucleating Arp2/3 complex (Kim et al., 2010a). Since then, a variety of other proteins involved in branched actin assembly have also been shown to negatively regulate ciliogenesis. For example, a study investigating mechanisms by which microRNA miR-129-3p promotes ciliogenesis found that it does so in part by downregulating Arp2 and several other mediators of branched actin formation (Cao et al., 2012). Another study found that ciliogenesis is inhibited by cortactin, which promotes actin polymerization and branching (Bershteyn et al., 2010). Several proteins involved in actin assembly were also identified in a recent whole-genome RNAi-based screen for negative regulators of ciliogenesis in breast cancer cells (Failler et al., 2021). These findings overall suggest that branched actin is inhibitory for ciliogenesis.

Various studies investigating the disease mechanisms of ciliopathy mutations have found that these mutations cause dysregulation of the actin cytoskeleton. For example, mutations in several genes that cause the ciliopathies Meckel–Gruber syndrome (MKS) and Joubert syndrome (JS) have been shown to cause changes in actin stress fibers along with a loss of ciliation in patient fibroblasts and cell culture models (Adams et al., 2012; Dawe et al., 2009; Valente et al., 2010; Wang et al., 2018; Yin et al., 2009). Mutations associated with Bardet–Biedl syndrome (BBS) produce a severely disrupted actin cytoskeleton with abnormal stress fibers, along with fewer and shorter cilia, in kidney cells in mice (Hernandez-Hernandez et al., 2013). Mutations that cause polycystic kidney disease (PKD) result in increased cortical actin and stress fibers, along with decreased ciliation in patient-derived cell lines, cell culture models and mouse kidneys (Streets et al., 2020). Some of these ciliopathy mutations lead to hyperactivation of RhoA (Dawe et al., 2009; Hernandez-Hernandez et al., 2013; Streets et al., 2020; Valente et al., 2010), which increases actin polymerization and contractility via Rho kinase proteins (ROCKs). In contrast, the mutation responsible for Townes–Brock syndrome, another disorder involving cilia defects, was shown to cause a reduction in actin and an increase in ciliation with aberrantly elongated cilia (Bozal-Basterra et al., 2020), highlighting the fact that not only positive but also negative regulation of ciliogenesis is crucial for ciliary function. Collectively, these observations of actin dysregulation in ciliopathy models bolster the conclusion that actin negatively regulates primary ciliogenesis.

Supporting a model in which actin polymerization inhibits primary ciliogenesis, several studies have shown that treatment with the actin polymerization inhibitors cytochalasin D (CytoD) or latrunculin B (LatB) increases the frequency of ciliation and the length of primary cilia in various cultured cell lines (Bershteyn et al., 2010; Kim et al., 2010a, 2015; Sharma et al., 2011; Nagai and Mizuno, 2017). CytoD treatment can also partially rescue the ciliation defects produced by knockdown of gelsolin or by loss-of-function mutation in IFT88, a core component of the intraflagellar transport (IFT) machinery that delivers materials for cilium growth (Kim et al., 2010a). Similarly, disrupting RhoA-ROCK-mediated actin contractility through ROCK inhibitor treatment enhances ciliogenesis in some cell lines and can rescue ciliary phenotypes of BBS and PKD ciliopathy models and those caused by various other mutations in which ciliogenesis defects are accompanied by actin dysregulation (Hernandez-Hernandez et al., 2013; Jewett et al., 2021; Kim et al., 2015; Rangel et al., 2019; Stewart et al., 2016; Streets et al., 2020).

From the evidence above, it might be expected that pharmacological stabilization of actin would block ciliogenesis, but surprisingly, treatment with the actin-stabilizing drug jasplakinolide also promotes ciliogenesis and cilium elongation, like the effect of CytoD (Nagai and Mizuno, 2017; Sharma et al., 2011). The discrepancy between the pro-ciliogenic effect of jasplakinolide and the anti-ciliogenic effect of genetic perturbations that stabilize actin may be because jasplakinolide produces disorganized actin aggregates, unlike the knockdowns of actin-remodeling proteins or overexpression of actin-stabilizing proteins, which give rise to organized actin structures such as stress fibers (Nagai and Mizuno, 2017); thus, jasplakinolide may be better considered as dysregulating rather than strengthening the actin cytoskeleton.

In order to understand the mechanisms by which actin remodeling regulates ciliogenesis, it is important to consider how the actin cytoskeleton is organized around the site of ciliogenesis and around the mature cilium (Fig. 1A). First, there is actin connected to the basal body. Although the centrosome (consisting of the mother centriole or basal body, the daughter centriole and pericentriolar material) is a microtubule-based structure, and is mainly considered as the microtubule-organizing center of the cell, it is linked to the actin cytoskeleton by focal adhesion proteins (Antoniades et al., 2014) and can also function as an actin-organizing center, which nucleates branched actin filaments via Arp2/3 (Farina et al., 2016). Second, there is the actin cortex, a dense mesh of actin filaments that forms a thin layer on the intracellular side of the plasma membrane. The actin cortex is present in most animal cells but is especially prominent at the apical membrane of epithelial cells (Doctor, 2006; Klingner et al., 2014). This cortical actin needs to be cleared away from the ciliation site to allow the basal body to dock at the apical plasma membrane in the extracellular ciliogenesis pathway, and cortical actin remains excluded from a region around the base of the cilium (Francis et al., 2011; Jewett et al., 2021). The ciliary pocket, specific to cilia generated by the intracellular pathway, is also connected to actin (Molla-Herman et al., 2010; Rattner et al., 2010). The presence of actin within the cilium itself has long been unclear, as actin-binding proteins have been detected in cilia (Kohli et al., 2017) and actin polymerization has been detected in cilia undergoing disassembly (Phau et al., 2017), but actin had not been directly identified within stable cilia. However, recent live-cell confocal imaging with genetically encoded actin-binding probes (Lee et al., 2018) and cryo-electron tomography (Kiesel et al., 2020) have shown that there is indeed actin in cilia.

Actin around the basal body plays important roles in trafficking of components for ciliogenesis (Fig. 1B). Ciliogenesis begins with trafficking of myosin-Va-associated preciliary vesicles to the mother centriole (Wu et al., 2018). Dynein transports these vesicles along microtubules to the pericentrosomal region, after which the actin motor myosin Va transports them along the Arp2/3-associated branched actin network to the mother centriole (Wu et al., 2018). Low-dose CytoD treatment enhances the centrosomal branched actin network, increases the transport of preciliary vesicles to the mother centriole and promotes ciliation, whereas high-dose treatment with CytoD, an Arp2 inhibitor, or Arp2 depletion disrupts the branched actin network, impairs preciliary vesicle transport to the mother centriole and reduces ciliation (Wu et al., 2018). As well as demonstrating that preciliary vesicle transport is actin dependent, these findings highlight an important caveat to the earlier studies that used CytoD to show that actin depolymerization promotes ciliogenesis (Bershteyn et al., 2010; Kim et al., 2010a, 2015; Sharma et al., 2011) but did not directly examine the effects of the CytoD treatments on actin at the centrosome. Indeed, CytoD exhibits multiple and varied effects on actin dynamics, depending not only on the dosage of CytoD but also on the particular actin population in question and the local concentrations of actin filament ends, actin monomers and various actin-interacting proteins (Cooper, 1987; Shoji et al., 2012; Stevenson and Begg, 1994). Thus, the commonly held idea that CytoD depolymerizes actin is not always accurate; the most that can be reliably assumed is that CytoD disrupts actin organization, but the nature of the disruption can vary, which may explain some of the differing results regarding effects of CytoD on ciliogenesis.

In contrast to the report that branched actin is required for myosin-Va-associated preciliary vesicle transport (Wu et al., 2018), several previous studies have indicated that branched actin negatively regulates trafficking to cilia. The formation of a ‘pericentrosomal preciliary compartment’ (PPC) that supplies materials for cilium growth, and the trafficking of ciliary proteins, such as Smoothened (Smo), Arl13b and IFT machinery, to the cilium, were shown to be enhanced not only by CytoD treatment (Fu et al., 2016; Kim et al., 2010a, 2015; Yeyati et al., 2017), but also by depletion of Arp3 (Fu et al., 2016; Kim et al., 2010a), microRNA-mediated suppression of Arp2 and several other branched-actin regulators (Cao et al., 2012), and treatment with an Arp2/3 inhibitor (Yeyati et al., 2017). Thus, the phenotype for loss-of-function of the branched-actin-nucleating Arp2/3 complex differs between studies, appearing in many cases to promote (Cao et al., 2012; Fu et al., 2016; Kim et al., 2010a; Yeyati et al., 2017) but in another case to disrupt (Wu et al., 2018) trafficking of ciliary materials to the centrosome and cilium. The reason for this discrepancy remains unclear. PPC formation has also been shown to be positively regulated by myosin IIB (with the heavy chain MYH10), which increases actin dynamics, and negatively regulated by myosin IIA (with the heavy chain MYH9), which stabilizes actin and suppresses its dynamics (Rao et al., 2014), supporting the idea that PPC formation requires the disassembly and remodeling of actin. It could be that branched actin serves as the tracks for the initial myosin-Va-dependent preciliary vesicle transport step, but also constitutes a fence that restricts transport of other components to cilia. One or the other of these roles of branched actin in regulating trafficking events for ciliogenesis might be more apparent depending on the specific experimental conditions.

Some of the formins, a class of non-branched actin nucleators, appear to positively regulate trafficking for ciliogenesis. The formin Daam1 localizes to vesicles carrying ciliary components and is required for ciliogenesis in renal epithelial cells (Corkins et al., 2019), and the formin Diaph1 stimulates ciliogenesis in part by promoting vesicular trafficking to the base of cilia (Palander and Trimble, 2020). A pan-formin inhibitor also reduces ciliation (Copeland et al., 2018), supporting an overall positive role for formins in ciliogenesis. Since formins nucleate linear actin filaments, whereas Arp2/3 nucleates branched actin filaments (Pollard, 2007; Suarez and Kovar, 2016), these results might indicate that certain trafficking events for ciliogenesis are negatively regulated by branched actin and positively regulated by linear actin. Inhibition of formin activity can increase centrosomal actin density (Inoue et al., 2019), perhaps due to competition for actin monomers between linear and branched actin networks (Suarez and Kovar, 2016). Therefore, a positive role for formins in this context might be caused by depletion of inhibitory centrosomal branched actin, as opposed to linear actin itself being required for trafficking to cilia. Interestingly, in contrast to the evidence described above that branched actin inhibits IFT trafficking and ciliogenesis in mammalian cells, both branched and linear actin are important for IFT trafficking and ciliogenesis in the unicellular flagellate Chlamydomonas (Avasthi et al., 2014; Jack et al., 2019; Bigge et al., 2020 preprint) (see Box 2).

Another actin-dependent step in ciliogenesis is the migration of the basal body to the apical region of the cell (Fig. 1C). Apical migration of the basal body is particularly crucial for the extracellular ciliogenesis pathway, which requires the basal body to dock at the apical plasma membrane. However, positioning the basal body on the dorsal side of the nucleus is important for ciliogenesis even in RPE-1 cells, which use the intracellular pathway, as contractile actin stress fibers impede ciliogenesis on the ventral side (Pitaval et al., 2010). Control of basal body migration by actin is evidenced by the fact that certain MKS and JS mutations, which dysregulate actin and impair primary ciliogenesis, disrupt the apical localization of the basal body (Adams et al., 2012; Dawe et al., 2007, 2009; Valente et al., 2010). Some of these mutations cause RhoA hyperactivation and its mislocalization away from the basal body, accompanied by an excess of polymerized actin (Valente et al., 2010), whereas one mutation instead causes a loss of RhoA activity and a decrease in polymerized actin (Adams et al., 2012). This suggests that basal body migration requires a balanced regulation of RhoA-mediated actin polymerization and contractility. A study investigating the effect of cell spatial confinement on ciliogenesis in RPE-1 cells found that confined cells develop a polarized actin cytoskeleton with an ezrin-rich network on the apical side, and contractile stress fibers on the basal side, which was associated with apical positioning of the centrosome followed by ciliogenesis (Pitaval et al., 2010). In contrast, widely spread cells have more contractile basal stress fibers and lack the apical actin pole, and, in these cells, the centrosome is positioned below the nucleus and ciliogenesis is blocked. Inhibiting actin polymerization with CytoD rescued basal body migration and ciliogenesis in the spread cells, while reducing actomyosin contractility with a ROCK inhibitor or a myosin II inhibitor disrupted basal body apical migration and ciliogenesis in confined cells (Pitaval et al., 2010). Thus, control of basal body migration by actomyosin cytoskeleton remodeling might serve to couple primary ciliogenesis to changes in cellular shape and tension during tissue development. Depletion of myosin IIB (MYH10) reduces the apical enrichment of actin and ezrin and impairs the apical positioning of the basal body (Hong et al., 2015), further supporting a role for myosin-II-dependent actin dynamics in mediating basal body migration.

The mechanism by which actin remodeling drives this movement of the basal body remains unclear. One model is that contractility of actomyosin filaments linking the basal body to the apical cortex pulls the basal body toward the apical pole. Actin- and myosin-IIB-based pulling forces have been implicated in centrosome movement in other cellular contexts, such as neuronal migration (Solecki et al., 2009) and mitotic spindle assembly (Kwon et al., 2015). Alternatively, it has been suggested that the role of actin in driving centrosome migration instead involves microtubule stabilization and symmetry breaking, resulting in microtubule-based pushing forces from the basal side, rather than direct actomyosin-based pulling forces from the apical cortex (Pitaval et al., 2017); other studies have also supported the idea that the actin cytoskeleton controls centrosome positioning mainly via its effects on the microtubule cytoskeleton (Jimenez et al., 2021). Another possibility is that branched actin connecting the centrosome to the nucleus needs to be disassembled to allow the centrosome to migrate away from the nucleus, as has been shown for immune synapse formation (Obino et al., 2016). In short, actin remodeling and contractility is involved in the apical migration of the basal body for ciliogenesis, but the mechanisms for this role of actin may be indirect and have yet to be elucidated.

For the extracellular ciliogenesis pathway, actin remodeling is required not only to move the basal body to the apical region, but also to clear a space in the actin cortex allowing the basal body to dock at the apical plasma membrane (Fig. 1C). The decrease in cortical actin intensity directly above the centrosome closely correlates with an increasing proximity between the centrosome and the cortex, and precedes the emergence of the cilium (Jewett et al., 2021). This cortical-actin-clearing step in epithelial ciliogenesis requires the GTPase Rab19, which localizes to vesicles around the centrosome within the cortical clearing, and the HOPS complex (a membrane tethering complex involved in lysosomal trafficking). Disrupting actomyosin contractility with a ROCK inhibitor rescues the defects in cortical actin clearing caused by Rab19 or HOPS knockout (Jewett et al., 2021), suggesting that cortical actin contractility blocks basal body docking and ciliation, and that a Rab19- and HOPS-dependent mechanism locally antagonizes actomyosin contractility in order for cortical actin to be cleared away from the ciliation site so that ciliation can proceed. Interestingly, this mechanism is not limited to the extracellular ciliogenesis pathway – Rab19 is required for ciliogenesis not only in polarized epithelial cells but also in RPE-1 cells, which use the intracellular pathway, and inhibition of actomyosin contractility also rescues ciliation in Rab19-knockout RPE-1 cells (Jewett et al., 2021). Although the intracellular ciliogenesis pathway does not involve basal body docking at the plasma membrane, a similar cortical-actin-clearing step may nevertheless be required to allow the ciliary sheath to fuse with the plasma membrane and expose the cilium to the extracellular space. The cortical actin clearing required for ciliogenesis might thus be comparable to the localized myosin-II-dependent relaxation of the actin cortex that is required for secretory granule docking at the plasma membrane during regulated exocytosis (Li et al., 2018; Miklavc and Frick, 2020).

A set of proteins that were originally identified as regulating planar cell polarity (PCP) in Drosophila also regulate ciliogenesis in vertebrates – leading to their designation as ‘ciliogenesis and planar polarity effector’ (CPLANE) proteins (Toriyama et al., 2016) – and the mechanism underlying both of these roles of the CPLANE machinery appears to involve control of the actin cytoskeleton (Adler and Wallingford, 2017; Brücker et al., 2020). For example, loss of the CPLANE proteins Inturned and Fuzzy results in a reduction of apical actin and a lack of both motile and primary cilia as well as PCP defects in Xenopus (Park et al., 2006), and produces stunted primary cilia in mice (Gray et al., 2009; Zeng et al., 2010). Mutations in Fuzzy also cause ciliopathies in humans (Barrell et al., 2022; Seo et al., 2011), highlighting the clinical importance of CPLANE proteins. Inturned recruits the formin Daam1 to basal bodies, where it functions in actin assembly (Yasunaga et al., 2015). Daam1, which mediates RhoA activation in the PCP pathway (Habas et al., 2001), is also important for ciliogenesis in mammalian renal epithelial cells, and this function requires the formin activity of Daam1 (Corkins et al., 2019). Wdpcp (also known as Fritz) is another CPLANE protein that regulates both PCP and ciliogenesis in Xenopus and in mice by modulating the actin cytoskeleton through controlling septin localization (Cui et al., 2013; Kim et al., 2010b; Park et al., 2015), and mutations in Wdpcp are also associated with human ciliopathies (Kim et al., 2010b; Saari et al., 2015). Ciliogenesis effects in CPLANE mutants include defects in vesicular trafficking to the basal body and cilium (Gerondopoulos et al., 2019; Gray et al., 2009; Toriyama et al., 2016; Zilber et al., 2013), consistent with the roles of actin in trafficking as described above. Some actin-interacting proteins originally characterized as ciliary proteins, such as the BBS proteins, are also important for PCP (May-Simera et al., 2015; Ross et al., 2005). Indeed, it has been hypothesized that cilia proteins in general might be better considered as polarity proteins that function in multiple polarized cellular processes including, but not limited to, ciliogenesis (Hua and Ferland, 2018). Thus, it seems likely that CPLANE proteins and various other ciliogenesis-linked factors regulate both PCP and ciliogenesis by controlling the polarized organization of the actin cytoskeleton.

Actin remodeling also regulates ciliogenesis indirectly through its effects on the transcriptional coactivators YAP and TAZ (collectively YAP/TAZ; also known as YAP1 and WWTR1, respectively), which mediate cellular responses to mechanical cues (Fig. 1D). Actin destabilization or disorganization, which is induced by high cell density or treatment with CytoD or jasplakinolide, results in YAP/TAZ inactivation; this suppresses the expression of certain negative regulators of ciliogenesis, such as AURKA and PLK1, which are YAP/TAZ target genes (Kim et al., 2015; Nagai and Mizuno, 2017). Lysophosphatidic acid (LPA), a component of serum, activates YAP/TAZ by promoting actin assembly, contributing to serum-induced suppression of ciliogenesis (Kim et al., 2015). This effect of LPA may be specific to the intracellular ciliogenesis pathway, given that some renal epithelial cells that use the extracellular ciliogenesis pathway can ciliate in the presence of LPA (Walia et al., 2019). Consistent with this, loss of YAP or TAZ induces ciliation in cell lines that use the intracellular pathway (Kim et al., 2015) but decreases ciliation in renal epithelial cells in zebrafish and mice (He et al., 2015; Hossain et al., 2007). Regulation of YAP/TAZ activity by the actin cytoskeleton thus has a cell-type-dependent effect on ciliogenesis.

In conclusion, although cilia have been primarily viewed as microtubule-based organelles, there is increasing evidence that ciliation is also controlled by the actin cytoskeleton and that actin dysregulation contributes to the pathology of many ciliopathies. Overall, primary ciliogenesis is inhibited by branched actin (nucleated by factors such as Arp2/3) and requires dynamic remodeling of actin (mediated by factors such as RhoA and myosin IIB). Key actin-dependent steps in ciliogenesis include the trafficking of ciliary components and the apical migration and docking of the basal body in polarized epithelial cells. Actin regulation of ciliogenesis also intersects with the PCP pathway and with YAP/TAZ signaling. Furthermore, although we have focused here on roles of actin in cilia assembly, recent studies have also demonstrated roles for actin in cilia maintenance, function and disassembly (see Box 3).

Besides the single primary cilium found on most cells, certain specialized multiciliated epithelial cells develop hundreds of motile cilia, which beat to move fluid across the extracellular surface (Spassky and Meunier, 2017). The process of multiciliogenesis, and the roles of actin in this process (Fig. 2), has much in common with epithelial primary ciliogenesis, but there are some important differences. Multiciliogenesis begins with de novo generation of many centrioles in addition to those produced by centriole duplication (Brooks and Wallingford, 2014; Meunier and Azimzadeh, 2016). These centrioles migrate and dock at the apical plasma membrane and then extend ciliary axonemes. The apical migration and docking of these basal bodies requires the formation of an apical actin web (Boisvieux-Ulrich et al., 1990; Lemullois et al., 1988; Pan et al., 2007). The axonemes of the motile cilia protrude through holes in the apical actin network (Werner et al., 2011), comparable to the clearing of apical cortical actin around the epithelial primary cilium. Treatment with the actin polymerization inhibitors CytoD or LatB inhibits formation of the apical actin web, disrupts basal body docking to the apical membrane and impairs multiciliogenesis (Boisvieux-Ulrich et al., 1990; Pan et al., 2007) – a notable contrast to primary ciliogenesis, which, as discussed above, is often enhanced by CytoD or LatB treatment.

The apical actin assembly required for basal body docking in multiciliated cells is mediated by RhoA; inhibition of RhoA impairs apical localization of basal bodies and blocks multiciliogenesis (Pan et al., 2007). Other factors involved in this apical actin assembly include PCP proteins, which mediate RhoA activation at basal bodies (Park et al., 2006, 2008), actin nucleators such as Cobl (Ravanelli and Klingensmith, 2011) and several formins (Mahuzier et al., 2018; Sedzinski et al., 2016, 2017), the actin crosslinker filamin A, the small GTPase R-Ras, which negatively regulates this process by sequestering filamin A away from basal bodies (Chevalier et al., 2015), and WDR5, which stabilizes actin around basal bodies (Kulkarni et al., 2018). Focal adhesion proteins link basal bodies to the actin network in multiciliated cells, as in monociliated cells (Antoniades et al., 2014).

Some studies have distinguished a pool of subapical actin in multiciliated cells that is differently regulated and has distinct functions from the apical actin (Antoniades et al., 2014; Mahuzier et al., 2018; Werner et al., 2011; Yasunaga et al., 2015). The subapical actin develops later, after basal body docking, and consists of filaments linking each basal body to the rootlet of the neighboring basal body, thereby regulating basal body spacing and synchronizing ciliary beating (Antoniades et al., 2014; Werner et al., 2011). The apical actin, meanwhile, is not only required for basal body docking at the beginning of motile ciliogenesis but also for keeping the basal bodies of mature motile cilia anchored at the apical surface against the forces generated by ciliary beating (Mahuzier et al., 2018) (Fig. 2).

Although most motile cilia are formed by the multiciliogenesis pathway described above, a recent ultrastructural study provided evidence for an alternative cytosolic route for multiciliogenesis (Narita and Takeda, 2021); see Box 4 for discussion of this cytosolic pathway.

Although numerous studies have established the importance of the actin cytoskeleton during cilia formation, the molecular machinery governing this process remains largely unclear. That is partially because the actin cytoskeleton clearly can both stimulate and inhibit cilia formation, likely because it plays different roles during different ciliation stages. Thus, analyzing cytoskeleton dynamics during cilia formation and signaling in live cells will be one of the key issues to fully understand ciliation, especially in the context of polarized epithelial cells. To make things more complicated, it has become clear that the process of cilia formation and signaling is also context dependent, and in various tissues and cell types, this process appears to rely on different regulators and is differentially impacted by the actin cytoskeleton. That is, of course, fully consistent with the fact that some ciliopathies appear to affect only certain tissues. Thus, future studies using various cell types and different animal models will be needed to fully understand the complexities of the crosstalk between localized cytoskeleton dynamics and cilia formation.