Glucose deprivation induces primary cilium formation through mTORC1 inactivation

Primary cilia are non-motile cilia protruding from the surface of many cell types that act as antennae to perceive and transmit diverse chemical and mechanical signals from the cell exterior (Goetz and Anderson, 2010; Ishikawa and Marshall, 2011; Basten and Giles, 2013). Since primary cilia play important roles in tissue development and homeostasis, defects in their formation are associated with a wide range of human disorders, collectively called ciliopathies, including polycystic kidney disease, retinal degeneration, polydactyly, visceral inversion, obesity and mental retardation (Fliegauf et al., 2007; Gerdes et al., 2009).

Primary cilia are formed by the extension of a microtubule-based axoneme from the mother centriole-derived basal body. Extensive studies have identified numerous molecules that are involved in primary cilium formation (Kim et al., 2010; Boldt et al., 2016), but the signaling mechanisms controlling ciliogenesis remain unclear. In cultured cells, primary cilia are efficiently induced by serum starvation or highly confluent culture, implying that ciliogenesis is correlated with the arrest of cell growth and proliferation (Tucker et al., 1979; Izawa et al., 2015; Malicki and Johnson, 2017). Recent studies also showed that ciliogenesis is triggered by various cellular stresses, such as UV radiation, heat shock, actin destabilization and loss of mechanical stresses (Villumsen et al., 2013; Kim et al., 2015). Because nutrient or energy stresses suppress cell growth and proliferation, it is predicted that these stresses also affect the efficiency of ciliogenesis. However, the role of nutrient signals in primary cilium formation is largely unknown.

Mammalian target of rapamycin (mTOR) is a protein kinase that plays a central role in transducing nutrient signals to cell responses, such as growth, metabolism, and proliferation (Dibble and Manning, 2013; Yuan et al., 2013; Saxton and Sabatini, 2017). mTOR interacts with several proteins to form two functionally distinct complexes, termed mTORC1 and mTORC2. In response to nutrient signals, such as glucose and amino acids, mTORC1 stimulates anabolic processes including protein and lipid biosynthesis, via phosphorylation of diverse target proteins including p70 ribosomal S6 kinase 1 (S6K1; also known as RPS6KB1) (Fingar et al., 2004; Saxton and Sabatini, 2017). mTORC1 also functions to inhibit autophagy by phosphorylating ULK1 (Noda and Ohsumi, 1998; Jung et al., 2010; Kim et al., 2011). Under nutrient-deprived conditions, mTORC1 activity is downregulated through a signaling cascade involving activation of AMP-activated protein kinase (AMPK) and the tuberous sclerosis complex (TSC, composed of TSC1, TSC2 and TBC1D7, a GAP for Rheb), and inactivation of Rheb, a small GTPase that activates mTORC1 (Saxton and Sabatini, 2017). Thus, mTORC1 serves as a sensor of nutrient status.

Recent reports have shown that treatment with rapamycin, a potent inhibitor of mTORC1, shortens the length of primary cilia in zebrafish embryos and in cultured fibroblasts (Yuan et al., 2012; Jin et al., 2015) and that Tsc1 depletion (which enhances mTORC1 activity) elongates cilium length in zebrafish embryos (DiBella et al., 2009). Thus, mTORC1 appears to be involved in elongation and maintenance of cilium length. Inactivation of mTORC1 causes cilium shortening probably through reduction in protein synthesis. By contrast, as inactivation of the mTORC1 pathway causes the arrest of cell proliferation (Nourse et al., 1994; Luo et al., 1996; Dalvai et al., 2010), it is possible that mTORC1 inactivation promotes ciliogenesis by triggering cell quiescence. In this respect, previous studies have shown that mTORC1 is aberrantly activated in polycystic kidney diseases and that rapamycin treatment slows the disease progression (Tao et al., 2005; Ibraghimov-Beskrovnaya and Natoli, 2011). The increased mTORC1 activity may contribute to the defect in ciliogenesis in this disease. However, the role of the mTORC1 pathway in ciliogenesis and the mechanism of mTORC1-mediated control of ciliogenesis remain to be established.

Recent studies have also shown the functional interaction between ciliogenesis and autophagy. Primary cilia are required for fluid flow-induced mTORC1 inactivation and autophagy stimulation in the size control of kidney epithelial cells (Boehlke et al., 2010; Orhon et al., 2016). Conversely, autophagy regulates ciliogenesis in a complex way. Autophagic degradation of negative [such as the oral-facial-digital syndrome protein-1 (OFD1)] and positive [such as intraflagellar transport protein-20 (IFT20)] regulators of ciliogenesis promotes and suppresses primary cilium formation, respectively, depending on cell types, cell conditions and types of autophagy (Tang et al., 2013; Pampliega et al., 2013). Wang et al. (2015) have also reported the reciprocal regulation of ciliogenesis and autophagy; shortening of cilia activates mTORC1 and represses autophagy, whereas inhibition of autophagy leads to short cilia. Given that mTORC1 inhibits autophagy, nutrient-dependent changes in mTORC1 activity may affect ciliogenesis through the regulation of autophagy.

In this study, we examined the effects of glucose deprivation and mTORC1 inactivation on primary cilium formation. We show that these treatments increase the proportion of ciliated cells, but shorten the cilium length. We also provide evidence that mTORC1 inactivation induces ciliogenesis through p27^KIP1 (also known as CDKN1B) upregulation but not through autophagy stimulation. These findings suggest that low glucose and mTORC1 inactivation play dual roles in ciliogenesis; they promote the initiation of primary cilium formation and shorten the length of cilia.

To investigate the role of nutrient stresses in primary cilium formation, we examined the effect of glucose deprivation on ciliogenesis in human telomerase reverse transcriptase (hTERT)-immortalized retinal pigmented epithelium (RPE) 1 cell lines (hereafter referred to as RPE1 cells). RPE1 cells at subconfluent density were cultured in a medium containing either 17.5 mM glucose or no glucose for 24 h under serum-fed or serum-starved conditions, and cilium formation was then analyzed by immunostaining with antibodies against Arl13b and acetylated (Ac)-tubulin (Fig. 1A). Under serum-fed conditions, less than 3% of cells formed cilia when cultured in normal glucose-containing medium, whereas ∼15% of cells formed cilia when cultured in glucose-deprived medium (Fig. 1A,B). Under serum-starved conditions, ∼30% of cells formed cilia when cultured in glucose-containing medium, but the number of ciliated cells further increased to 52% upon glucose deprivation (Fig. 1A,B). These results suggest that glucose deprivation promotes primary cilium formation under both serum-fed and serum-starved conditions and that glucose deprivation acts to promote ciliogenesis additively or synergistically with serum starvation. Similar to the case of RPE1 cells, glucose deprivation also significantly promoted ciliogenesis in human embryonic kidney (HEK) 293T cells (Fig. S1A). Since glucose deprivation is expected to decrease the osmolality of the culture medium, we examined the effect of sorbitol addition on glucose-deprivation-induced ciliogenesis. Addition of 10–20 mM sorbitol to the medium had no apparent effect on glucose-deprivation-induced ciliogenesis (Fig. S1B), suggesting that glucose deprivation does not induce ciliogenesis through a decrease in osmolality.

We also analyzed the effect of glucose deprivation on the length of cilia. Cilia that were formed in cells cultured in glucose-depleted medium under serum-fed or serum-starved conditions were shorter than those in cells cultured in glucose-containing medium under serum-starved conditions (Fig. 1C). Thus, glucose deprivation increased the proportion of ciliated cells but shortened the length of the formed cilia.

Although glucose deprivation often leads to protein deacetylation (Zhao et al., 2010; Snider et al., 2013), this treatment enhanced tubulin acetylation in primary cilia (Fig. 1A). Immunoblot analysis revealed that glucose deprivation slightly decreases the level of tubulin acetylation in whole-cell lysates of RPE1 cells (Fig. S1C). These results suggest that tubulin acetylation in primary cilia is regulated differently from that in the cytoplasm.

Glucose deprivation is known to suppress mTORC1 kinase activity. To examine whether glucose deprivation actually reduces mTORC1 activity, we analyzed the levels of phosphorylated S6K1, a well-known substrate of mTORC1, in RPE1 cells cultured in glucose-fed or glucose-deprived medium under serum-starved conditions. Immunoblot analyses with an antibody specific to S6K1 phosphorylated on T389 [pS6K1(T389)] revealed that S6K1 phosphorylation decreased in cells cultured in glucose-deprived medium, whereas no apparent change in the amounts of total S6K1 and mTOR proteins was observed (Fig. 2A). These results indicate that glucose deprivation decreases mTORC1 activity in RPE1 cells.

To examine whether mTORC1 inactivation is involved in glucose deprivation-induced ciliogenesis, we analyzed the effect of overexpression of wild-type Rheb (Rheb-WT), an activator of mTORC1. As reported previously (Long et al., 2005), overexpression of Rheb-WT increased the level of pS6K1 in glucose-deprived cells, indicating that Rheb-WT expression actually induced mTORC1 activation (Fig. 2B). Whereas glucose deprivation promoted primary cilium formation in control YFP-expressing cells under serum-starved conditions, transfection of YFP–Rheb-WT suppressed glucose-deprivation-induced ciliogenesis for YFP-positive cells but not for YFP-negative cells (Fig. 2C,D). Furthermore, expression of a constitutively active Q64L mutant of Rheb suppressed glucose-deprivation-induced ciliogenesis, but expression of an inactive D60K mutant did not (Fig. S2). These results suggest that mTORC1 activation suppresses glucose-deprivation-induced ciliogenesis and that glucose deprivation promotes ciliogenesis through mTORC1 inactivation.

To further examine the role of mTORC1 inactivation in ciliogenesis, we analyzed the effect of rapamycin, a potent inhibitor of mTORC1, on ciliogenesis. RPE1 cells cultured at subconfluent density were treated with 1 µM rapamycin for 24 h under serum-fed or serum-starved conditions, and the ciliated cells were then analyzed by staining with anti-Arl13b and anti-Ac-tubulin antibodies. Similar to what was seen upon glucose deprivation, treatment with rapamycin notably increased the ratio of ciliated cells under both serum-fed and serum-starved conditions (Fig. 3A,B). Rapamycin treatment also promoted ciliogenesis in HEK293T cells (Fig. S3A) and mouse embryonic fibroblasts (MEFs) (see Fig. 7F). Rapamycin dose-dependently increased the percentage of ciliated cells, with significant effects observed at concentrations of >0.01 μM (Fig. S3B). The lower dose (0.01–0.1 μM) of rapamycin inhibits mTORC1 but not phosphoinositide 3-kinase (PI3K) (Chung et al., 1994; Tyagi et al., 2015), suggesting that rapamycin induces ciliogenesis by inhibiting mTORC1. We also examined the effect of overexpression of TSC2, a suppressor of mTORC1 pathway, on ciliogenesis. Overexpression of TSC2 increased the number of ciliated cells (Fig. S3C), which further suggests that mTORC1 inactivation induces ciliogenesis.

We also analyzed the effect of rapamycin treatment on the length of cilia. Cilia that were formed in rapamycin-treated cells cultured under serum-fed or serum-starved conditions were shorter than those in rapamycin-untreated cells cultured under serum-starved conditions (Fig. 3C). Immunoblot analyses with an antibody against pS6K1(T389) as well as immunoblot analyses with an anti-S6K1 antibody after Phos-tag-containing SDS-PAGE revealed that rapamycin treatment markedly decreased the level of S6K1 phosphorylation (Fig. 3D), indicating that rapamycin inactivates mTORC1 in RPE1 cells. Thus, similar to what is seen upon glucose deprivation, rapamycin treatment increased the proportion of ciliated cells but shortened the length of the formed cilia.

Fig. 3D also shows the effect of serum starvation on mTORC1 activity. Intriguingly, serum starvation had no apparent effect on the level of S6K1 phosphorylation in RPE1 cells (Fig. 3D, lanes 1 and 3). Similar results were obtained in HEK293T cells (Fig. S3D). These results suggest that serum starvation does not cause mTORC1 inactivation but induces cilium formation by a mechanism distinct from mTORC1 inactivation.

mTOR exists in two distinct complexes, mTORC1 and mTORC2, which contain the unique accessory proteins, Raptor and Rictor, respectively (Saxton and Sabatini, 2017). Rapamycin directly inhibits mTORC1, but not mTORC2, suggesting a preferential role for mTORC1 in ciliogenesis. To further define the role of mTORC1 and mTORC2 in ciliogenesis, we examined the effect of Raptor or Rictor knockdown on ciliogenesis. Transfection of siRNAs targeting Raptor or Rictor reduced the expression level of each target transcript in RPE1 cells (Fig. 3E). When RPE1 cells were transfected with Raptor- or Rictor-targeting siRNAs and cultured for 48 h under serum-fed conditions, the proportion of ciliated cells significantly increased upon treatment with Raptor siRNAs, but not with Rictor siRNAs, as compared with control siRNA (Fig. 3F). Immunoblot analyses revealed that Raptor knockdown substantially decreased the level of S6K1 phosphorylation, whereas knockdown of Rictor did not (Fig. 3G). These results indicate that inhibition of mTORC1, but not mTORC2, promotes primary cilium formation.

Previous studies have shown that overexpression of CPAP (also known as CENPJ) or C2CD3 or depletion of CP110 (also known as CCP110) or OFD1 leads to centriole elongation in some types of cells (Schmidt et al., 2009; Singla et al., 2010; Thauvin-Robinet et al., 2014). To determine whether the rapamycin-induced Ac-tubulin-positive structures are short cilia or elongated centrioles, we compared the localization of Ac-tubulin and several ciliary or centrosomal proteins in rapamycin-induced structures with that in primary cilia induced by serum starvation, using a super-resolution structured illumination microscopy (SR-SIM). Ciliary proteins, Arl13b and IFT88, were not detectable on the elongated centrioles (Schmidt et al., 2009), but they were localized on both rapamycin-induced structures and serum-starvation-induced primary cilia (Fig. 4A,B). Pericentrin was shown to localize along the elongated centrioles (Singla et al., 2010); however, it was localized only in the basal region of rapamycin-induced structures and primary cilia (Fig. 4C). Whereas two CP110 dots were detected at both mother and daughter centrioles in non-ciliated cells, only a single CP110 dot at the daughter centriole was observed near the rapamycin-induced structures and primary cilia (Fig. 4D). Localization of Cep164 (at distal appendages) and MKS1 (at the transition zone) in the rapamycin-induced structure was similar to that in primary cilia in serum-starved cells (Fig. 4E,F). Taken together, these observations indicate that rapamycin-induced structures are primary cilia but not elongated centrioles.

It is well known that primary cilia are formed in cells cultured at high cell density (Wheatley, 1971). In accordance with previous observations, the proportion of ciliated cells increased, depending upon a rise in the density of RPE1 cells, under both serum-fed and -starved conditions (Fig. 5A). To examine whether cell density affects mTORC1 activity, we analyzed mTORC1 activity in RPE1 cells cultured at distinct cell densities under serum-fed conditions by measuring the levels of S6K1 phosphorylation. Intriguingly, immunoblot analysis of pS6K1 and Phos-tag blot analysis of S6K1 revealed that the levels of S6K1 phosphorylation gradually decrease with an increase in cell density (Fig. 5B), indicating that mTORC1 activity declines in cells cultured at high density. Thus, the cell density-dependent cilium formation is correlated with decreased mTORC1 activity.

Inactivation of mTORC1 reduces the level of phosphorylation and the kinase activity of S6K1. To examine whether S6K1 inactivation is involved in cilium formation, we analyzed the effect of PF-4708671, an inhibitor of S6K1 (Pearce et al., 2010), on ciliogenesis in RPE1 cells cultured under serum-fed conditions. Treatment of cells with PF-4708671 for 24 h did not induce ciliogenesis, but treatment for 48 h significantly increased the number of ciliated cells (Fig. 5C). Under similar conditions, treatment with rapamycin induced ciliogenesis more effectively (Fig. 5C). These results suggest that inactivation of S6K1 partially contributes to mTORC1-inactivation-induced ciliogenesis.

OFD1 is a centrosomal protein involved in primary cilium formation (Singla et al., 2010). A recent study showed that OFD1 at centriolar satellites plays a role in suppressing ciliogenesis and that autophagic degradation of OFD1 at centriolar satellites promotes ciliogenesis (Tang et al., 2013). Since mTORC1 negatively regulates autophagy, it is possible that inactivation of mTORC1 promotes ciliogenesis by enhancing autophagy and thereby removing OFD1 from centriolar satellites. To examine this possibility, we analyzed the effects of rapamycin treatment on the localization and abundance of OFD1 in RPE1 cells. Immunostaining using an antibody specific to OFD1 revealed that rapamycin treatment has no apparent effect on OFD1 localization at the centrosome (Fig. 6A). Furthermore, immunoblot analysis also revealed that rapamycin treatment has no apparent effect on the OFD1 protein levels in cell lysates (Fig. 6B). These results suggest that mTORC1 inactivation does not cause OFD1 degradation from the centrosome. We also examined the effect of rapamycin treatment on autophagy by measuring the amount of LC3-II, a high-mobility band of LC3 protein (also known as MAP1LC3B protein) that specifically associate with the membranes of autophagosomes (Kabeya et al., 2000). Immunoblot analysis showed that rapamycin treatment had no apparent effect on the intensity of the LC3-II band (Fig. 6C). Bafilomycin A1 (BafA1), an inhibitor of lysosomal acidification and degradation functions, is known to cause accumulation of LC3-II. As expected, BafA1 treatment increased the amount of LC3-II; however, no further increase was observed upon rapamycin treatment (Fig. 6C). Thus, it seems likely that rapamycin treatment does not promote autophagy in RPE1 cells under the conditions tested. These results suggest that rapamycin induces ciliogenesis independently of autophagy-mediated OFD1 downregulation.

We also examined whether the cell density affects autophagy. The amount of LC3-II was slightly higher in cells cultured at high density than that in cells at low density (Fig. S4). Correspondingly, the amount of OFD1 was decreased in high-density cells. Thus, the autophagy-mediated OFD1 degradation appears to contribute at least in part to ciliogenesis induced by high cell density culture.

In this study, we showed that glucose deprivation or rapamycin treatment increases the number of ciliated cells but shortens the length of the formed cilia. Rapamycin treatment is also reported to reduce the length of cilia in epithelial cells in zebrafish embryos and in cultured NIH3T3 cells (Yuan et al., 2012; Jin et al., 2015). To examine the effect of rapamycin on pre-existing cilia in RPE1 cells, cells were cultured at high density for 24 h under serum-starved conditions. Under these conditions, more than 90% of cells formed primary cilia. Cells were then further cultured for 24 h in the presence or absence of rapamycin, and the proportion of ciliated cells and the length of cilia were analyzed. Rapamycin treatment had no apparent effect on the number of ciliated cells (Fig. S5A,B), but significantly reduced the length of cilia (Fig. S5A,C). Thus, similar to previously reported results, rapamycin treatment caused shortening of pre-existing cilia. These results suggest that mTORC1 inhibition has dual roles in ciliogenesis (i.e. induction of primary cilium formation and shortening of cilium length).

Previous studies demonstrated that inactivation of mTORC1 by glucose deprivation or rapamycin treatment induces cell cycle arrest via upregulation of p27^KIP1 cyclin-dependent kinase (CDK) inhibitor (Nourse et al., 1994; Luo et al., 1996; Dalvai et al., 2010). When RPE1 cells were treated with rapamycin, the level of p27^KIP1 protein increased, as reported previously (Fig. 7A). We next examined the effect of rapamycin treatment on the percentage of cells in the S-phase by analyzing 5-ethynyl-2′-deoxyuridine (EdU) incorporation into DNA. Rapamycin treatment markedly decreased the percentage of cells in the S phase (Fig. 7B,C), indicating that rapamycin treatment leads to cell cycle arrest. To examine whether mTORC1 inhibition promotes ciliogenesis through an increase in p27^KIP1, we analyzed the effect of p27^KIP1 deletion on rapamycin-induced ciliogenesis by using p27^KIP1-deficient (p27^KIP1−/−) MEFs (Fig. 7D). Primary cilium formation was promoted by rapamycin treatment in control p27^KIP1+/+ MEFs but was not induced in p27^KIP1−/− MEFs under serum-fed conditions (Fig. 7E,F). Moreover, expression of FLAG-tagged p27^KIP1 led to the recovery of the rapamycin-induced ciliogenesis in p27^KIP1−/− MEFs (Fig. 7G). These results suggest that rapamycin induces ciliogenesis through p27^KIP1 upregulation. We also examined the effect of p27^KIP1 knockdown on rapamycin- or glucose-deprivation-induced ciliogenesis in RPE1 cells. Cells were transfected with control siRNA or p27^KIP1-targeting siRNAs and then subjected to rapamycin treatment or glucose deprivation for 24 h. Rapamycin treatment or glucose deprivation significantly increased the number of ciliated cells in control cells, but these treatments had no significant effect on ciliogenesis in p27^KIP1-knockdown cells (Fig. S6). These results further suggest that p27^KIP1 is involved in rapamycin- or glucose-deprivation-induced ciliogenesis.

In this study, we showed that glucose deprivation promoted primary cilium formation under both serum-fed and serum-starved conditions. Glucose deprivation caused mTORC1 inactivation, and mTORC1 inactivation mediated by rapamycin treatment, Raptor knockdown or TSC2 overexpression promoted ciliogenesis. Furthermore, glucose-deprivation-induced ciliogenesis was suppressed by forced activation of mTORC1 by overexpression of an active form of Rheb. Taken together, these results strongly suggest that glucose deprivation promotes primary cilium formation through mTORC1 inactivation.

Glucose deprivation increased the number of ciliated cells even under serum-starved conditions, indicating that glucose deprivation promotes ciliogenesis additively or synergistically with serum starvation. Serum starvation did not cause mTORC1 inactivation (as measured by the level of S6K1 phosphorylation), indicating that serum starvation induces ciliogenesis independently of mTORC1 inactivation, and that glucose deprivation and serum starvation promote ciliogenesis through distinct pathways. We have shown that rapamycin treatment increases the protein level of the CDK2 and CDK4 inhibitor p27^KIP1, and decreases the proportion of S-phase cells in serum-fed RPE1 cells. Rapamycin- or glucose-deprivation-induced promotion of cilium formation was abrogated in p27^KIP1-depleted cells and the inhibitory effect of p27^KIP1 knockout was recovered upon exogenous expression of p27^KIP1. These results suggest that mTORC1 inactivation promotes ciliogenesis through p27^KIP1 upregulation.

Previous studies have shown that mTORC1 suppresses autophagy (Noda and Ohsumi, 1998; Jung et al., 2010; Kim et al., 2011) and that autophagy positively and negatively regulates ciliogenesis (Pampliega and Cuervo, 2016). In dividing cells, autophagic degradation of positive regulators of ciliogenesis (such as IFT20) suppresses ciliogenesis (Pampliega et al., 2013), whereas, in resting cells, autophagic degradation of OFD1, a negative regulator of ciliogenesis, at centriolar satellites promotes ciliogenesis (Tang et al., 2013). Since mTORC1 inactivation has been reported to promote autophagy, we hypothesized that rapamycin treatment might induce ciliogenesis though promoting autophagy and degradation of OFD1. However, rapamycin treatment had no apparent effect on the amount of LC3-II or the amount and centrosomal localization of OFD1 in RPE1 cells. These results suggest that mTORC1 inactivation does not always promote autophagy and that autophagic degradation of OFD1 is not a major factor that is involved in rapamycin-induced ciliogenesis in RPE1 cells.

Culture of cells at high density also promotes primary cilium formation additively or synergistically with serum starvation. We observed that mTORC1 activity is decreased in a manner dependent on increased cell density. Intriguingly, the amount of LC3-II was slightly increased and the amount of OFD1 was decreased in cells cultured at high density. Thus, it seems likely that culture at high cell density promotes ciliogenesis, at least in part, through mTORC1 inactivation, followed by stimulation of autophagy and OFD1 degradation. Further studies are required to understand the difference in mechanisms of ciliogenesis induced by glucose deprivation, serum starvation and culture at high density.

Previous studies have shown that rapamycin treatment shortened the length of primary cilia in zebrafish Kupffer's vesicles and in cultured NIH3T3 cells (Yuan et al., 2012; Jin et al., 2015), and that Tsc1 depletion (which enhances mTORC1 activity) elongates cilium length in zebrafish otic vesicles and pronephric ducts (DiBella et al., 2009). Consistent with these observations, we observed that rapamycin treatment shortened the length of pre-existing and newly formed cilia in RPE1 cells. It is likely that mTORC1 is involved in the elongation and maintenance of cilium size by promoting the biosynthesis of proteins and lipids required for cilium elongation and maintenance. Thus, mTORC1 inactivation has dual roles in ciliogenesis, namely, triggering cilium formation and controlling cilium length. Further studies on the precise mechanisms underlying mTORC1-mediated regulation of the frequency and length of primary cilia are required for better understanding the regulation mechanisms of nutrient state-dependent cilium formation and their functions in vivo.

Hartman et al. (2009) reported that depletion of TSC1 or TSC2 promoted cilium formation whereas rapamycin did not suppress TSC1- or TSC2-depletion-induced ciliogenesis, but rather promoted ciliogenesis. Based on these data, they proposed that TSC1 and TSC2 negatively regulate cilium formation, independently of mTORC1 inactivation. However, they did not further investigate the mechanism of the rapamycin-induced ciliogenesis. It also remains unknown how TSC1 and TSC2 negatively regulate ciliogenesis.

Recent studies have shown that in response to fluid flow, primary cilia transduce signals to activate the LKB1–AMPK pathway, which leads to downregulation of mTORC1 activity and stimulation of autophagy, which in turn reduces cell volume in kidney cells (Boehlke et al., 2010; Orhon et al., 2016; Zhong et al., 2016). Defects in cilium formation cause polycystic kidney diseases and renal cell carcinoma, which may be attributed to the increased mTORC1 activity caused by the loss of cilium formation. In fact, it has been reported that the mTORC1 pathway is activated in polycystic kidney diseases and that rapamycin treatment slows the disease progression (Tao et al., 2005; Ibraghimov-Beskrovnaya and Natoli, 2011). In this study, we showed that mTORC1 inactivation promotes cilium formation in various cell types. Thus, cilia-mediated mTORC1 inactivation might contribute to the maintenance of cilia by a feedback mechanism. Conversely, defects in cilium formation or function cause abnormal mTORC1 activation, which might result in further suppression of cilium formation and disease progression. Further studies on the precise role of the mTORC1 pathway in ciliogenesis are important for better understanding the pathogenesis of ciliopathies, including polycystic kidney diseases, and providing diagnostic and therapeutic perspectives for these diseases.

Rapamycin, bafilomycin A1 (BafA1) and sorbitol were purchased from Wako Pure Chemical (Osaka, Japan). PF-4708671 and 4′, 6-diamidino-2-phenylindol (DAPI) were purchased from Toronto Research Chemicals and Polysciences, respectively. Antibodies were purchased from the indicated suppliers as follows: mouse monoclonal antibodies against: α-tubulin (1:1000, B-5-1-2, Sigma-Aldrich), acetylated (Ac)-tubulin (1:1000, 6-11B-1, Sigma-Aldrich), γ-tubulin (1:1000, GTU-88, Sigma-Aldrich), and p27^KIP1 (1:1000, 610241, BD Biosciences), and rabbit monoclonal antibody against mTOR (1:400, 2983, Cell Signaling), and rabbit polyclonal antibodies against S6K1 (1:200, 9202, Cell Signaling), pS6K1(T389) (1:1000, 9205, Cell Signaling), Arl13b (1:1000, 17711-1-AP, Proteintech), GFP (1:1000, A6455, Thermo Fisher Scientific), OFD1 (1:1000, NBP1-89355, Novus Biologicals), LC3 (1:10,000, PM036, MBL, Nagoya, Japan), Cep164 (1:2000, 22227-1-AP, Proteintech), MKS1 (1:250, 16206-1-AP, Proteintech), IFT88 (1:1000, 13967-1-AP, Proteintech), CP110 (1:1000, A301-343A1, Bethyl), and pericentrin (1:2000, PRB-432C, Babco). Secondary antibodies for immunofluorescence were Alexa Fluor 488- or 568-labeled anti-mouse or anti-rabbit IgG (1:1000, Thermo Fisher Scientific). Secondary antibodies for western blotting were horseradish peroxidase (HRP)-labeled anti-mouse or anti-rabbit IgG (1:20,000, GE Life Sciences).

The cDNA for human Rheb was cloned by PCR amplification, using the primers 5′-AACTCGAGTTATGCCGCAGTCCAAGTCCC-3′ and 5′-AAGAATTCAATCACATCACCGAGCATG AAGAC-3′, and subcloned into the XhoI and EcoRI sites in the pEYFP-C1 vector (Clontech, Mountain View, CA). The cDNA plasmids for CFP-tagged active (Q64L) and inactive (D60K) mutants of Rheb were constructed as described previously (Zhou et al., 2009). The cDNA plasmid for p27^KIP1 was constructed as described previously (Nakayama et al., 1996). Dharmacon siRNAs targeting Raptor, Rictor and p27^KIP1 were purchased from GE Healthcare (Little Chalfont, UK). The siRNA target sequences were as follows: 5′-UGGCUAGUCUGUUUCGAAA-3′ (Raptor siRNA #1), 5′-CACGGAAGAUGUUCGACAA-3′ (Raptor siRNA #2), 5′-GACACAAGCACUUCGAUUA-3′ (Rictor siRNA #1), 5′-GAAGAUUUAUUGAGUCCUA-3′ (Rictor siRNA #2), 5′-CAAGUGGAAUUUCGAUUUU-3′ (p27^KIP1 siRNA #1), and 5′-CGACGAUUCUUCUACUCAA-3′ (p27^KIP1 siRNA #2).

hTERT-RPE1 (referred to as RPE1) cells were provided by Hiroyuki Nakanishi (Kumamoto University, Kumamoto, Japan) and Makoto Matsuyama (Okayama University, Okayama, Japan). RPE1 cells were cultured in Dulbecco's modified Eagle's medium (DMEM)/Ham's F-12 (Wako Pure Chemical Industries, Osaka, Japan) supplemented with 10% fetal calf serum (FCS) under 5% CO₂ at 37°C. HEK293T cells were obtained from the American Type Culture Collection and cultured in DMEM supplemented with 10% FCS. MEFs were obtained from p27^KIP1+/+ or p27^KIP1−/− mice, as reported previously (Nakayama et al., 1996). MEFs were cultured in DMEM supplemented with 2 mM L-glutamine, 10% FCS, 1 mM sodium pyruvate and 0.1 mM non-essential amino acids. Transfections with plasmids and siRNAs were performed using Lipofectamine LTX and Lipofectamine RNAiMAX (Thermo Fisher Scientific), respectively. To analyze the effect of glucose deprivation, RPE1 cells were cultured in glucose-depleted DMEM/Ham's F-12 (Nacalai Tesque, Kyoto, Japan) for 24 h under 10% FCS-containing or -starved conditions.

Immunofluorescence staining was carried out as described previously (Chiba et al., 2013; Oda et al., 2014). RPE1 cells grown on coverslips in six-well culture plates were washed once with phosphate-buffered saline (PBS) and fixed with 4% paraformaldehyde in PBS for 30 min at 20°C or with ice-cold methanol for 10 min at 4°C, followed by permeabilization with 0.1% Triton X-100 for 5 min. After blocking with 2% FCS in PBS for 30 min, cells were stained with the primary antibody diluted in Can-Get-Signal immunostaining solution (Toyobo, Osaka, Japan) for 1 h at 20°C and the secondary antibody diluted with 2% FCS in PBS. Fluorescence images were obtained with a DMI 6000B fluorescence microscope (Leica Microsystems, Wetzlar, Germany) equipped with a PL Apo 63× oil objective lens and CCD camera (Cool SNAP HQ, Roper Scientific, Martinsried, Germany). To detect DNA, cells were incubated in 5 mg/ml DAPI in PBS for 5 min and mounted in paraphenylenediamine. The SR-SIM imaging was performed on an ELYRA S.1 microscope (Carl Zeiss Microscopy) equipped with an Andor iXon 885 EMCCD camera, a 100×1.40 NA oil-immersion objective and four laser beams (405, 488, 561, 642 nm). Serial z-stack sectioning was carried out at 101 nm intervals. Z stacks were recorded with three phase changes and five grating rotations for each section. The microscope was calibrated with 100 nm fluorescent beads to calculate both the lateral and axial limits of image resolution. Images were reconstituted with Zeiss Zen software.

Primary cilium formation was analyzed as described previously (Chiba et al., 2013; Oda et al., 2014). RPE1 cells were plated on coverslips in six-well dishes, cultured with 10% serum-containing or serum-starved medium for 24 h, and then fixed with 4% paraformaldehyde and immunostained with anti-Arl13b and anti-Ac-tubulin antibodies. Alexa Fluor 488- or 568-labeled anti-rabbit IgG and anti-mouse IgG were used as secondary antibodies. The cells containing a cilium with a length of >0.5 μm were counted as the ciliated cells.

Immunoblotting was performed as described previously (Amagai et al., 2015). Cells were lysed with pre-heated SDS-containing lysis buffer (1% SDS, 50 mM Tris-HCl pH 7.4 and 1 mM dithiothreitol) and boiled for 20 min. After centrifugation, the supernatants were mixed with SDS sample buffer and boiled for 5 min. Lysates were loaded on 8–15% SDS-polyacrylamide gels and transferred to polyvinylidene difluoride membranes (GE Healthcare). To detect the S6K1 phosphorylation levels, cell lysates were loaded on SDS-polyacrylamide gels containing 25 µM Phos-tag (Wako Pure Chemical). The membranes were blocked with skimmed milk or Blocking One-P [Nacalai Tesque, for detecting pS6K1(T389)] and incubated with the appropriate antibodies. Horseradish peroxidase-conjugated anti-rabbit or anti-mouse IgG antibody (GE Healthcare) was used as the secondary antibody.

Total RNA was isolated from RPE1 cells using an Isogen II kit (Nippon Gene, Tokyo, Japan) and reverse transcribed to yield single-stranded cDNAs using a Transcriptor high fidelity cDNA synthesis kit (Roche Diagnostics, Indianapolis, IN). The cDNA fragments were amplified by PCR amplification (30 cycles of 95°C for 60 s, 58°C for 30 s, and 72°C for 60 s) using GoTaq DNA polymerase (Promega, Madison, WI) and analyzed by agarose gel electrophoresis. The primers used are as follows: 5′-CAGCCTGACCAACGATGT-3′ (Raptor sense), 5′-GGGTGGAATTTCACCACAG-3′ (Raptor antisense), 5′-CGAGTCTTTGGATGATTAAGAAGG-3′ (Rictor sense), 5′-CAAGACCTCCAGTTCCAGATG-3′ (Rictor antisense), 5′-AGAAAATCTGGCACCACACC-3′ (β-actin sense) and 5′-CCATCTCTTGCTCGAAGTCC-3′ (β-actin antisense).

Cell proliferation assay was carried out by using the Click-iT Plus EdU Alexa Fluor 647 imaging kit (Thermo Fisher Scientific), according to the manufacturer's protocols. Briefly, cells were cultured on the slide glass in six-well plates and labeled with 10 µM EdU for 2 h. The EdU-incorporated cells were analyzed by means of Alexa Fluor 647 fluorescence.

Data are expressed as the mean±s.e.m. of more than three independent experiments. P-values were calculated using a paired Student's t-test. Data were considered statistically significant at P<0.05.

We thank H. Nakanishi and M. Matsuyama for providing hTERT-RPE1 cells, Y. Amagai for providing Rheb cDNA constructs, D. Umetsu and E. Kuranaga for support and advice in EdU assays, and T. Nakagawa, S. Tsukita, K. Ohashi and K. Yasumoto for valuable comments. We thank the SR-SIM imaging facility in the Center for Medical Research and Education, Graduate School of Medicine, Osaka University.