ABSTRACT
Cilia perform essential signalling functions during development and tissue homeostasis. A key event in ciliogenesis occurs when the distal appendages of the mother centriole form a platform that docks ciliary vesicles and removes CP110-Cep97 inhibitory complexes. Here, we analysed the role of LRRC45 in appendage formation and ciliogenesis. We show that the core appendage proteins Cep83 and SCLT1 recruit LRRC45 to the mother centriole. Once there, LRRC45 recruits the keratin-binding protein FBF1. The association of LRRC45 with the basal body of primary and motile cilia in both differentiated and stem cells reveals a broad function in ciliogenesis. In contrast to the appendage components Cep164 and Cep123, LRRC45 was not essential for either docking of early ciliary vesicles or for removal of CP110. Rather, LRRC45 promotes cilia biogenesis in CP110-uncapped centrioles by organising centriolar satellites, establishing the transition zone and promoting the docking of Rab8 GTPase-positive vesicles. We propose that, instead of acting solely as a platform to recruit early vesicles, centriole appendages form discrete scaffolds of cooperating proteins that execute specific functions that promote the initial steps of ciliogenesis.
INTRODUCTION
The primary cilium is a microtubule-based organelle that protrudes from the surface of several cell types in the human body. It has mechanosensory and signalling functions as a result of its association with receptors of important signalling pathways, such as hedgehog (Shh) and Wnt (Goetz and Anderson, 2010). Defects in primary cilia formation or function are the underlying cause of severe genetic diseases, collectively named ciliopathies, in which embryonic development and/or functioning of multiple organs in the human body can be affected (Reiter and Leroux, 2017).
The primary cilium emerges from the mother centriole that converges into the basal body of the cilium. Each centrosome is composed of two centrioles, cylindrical structures consisting of a nine-fold array of triplet microtubules and pericentriolar material (PCM). The two centrioles of a centrosome are different in function and composition. The mother, but not the daughter centriole, associates with an array of proteins, named appendage proteins, which give rise to the cilium upon cell cycle exit (Nigg and Stearns, 2011).
Ciliogenesis is a multi-step process that requires the concerted action of several components, including mother centriole-associated proteins and the vesicular transport machinery, which, together, are important for the establishment and maintenance of the ciliary compartment (Sánchez and Dynlacht, 2016). At early stages of ciliogenesis, the mother centriole associates with small ciliary vesicles, most likely derived from the Golgi (Das and Guo, 2011; Qin, 2012). After docking to the mother centriole, small vesicles are fused, thereby forming a large ciliary vesicle that caps this centriole (Sánchez and Dynlacht, 2016). The fusion process depends on the membrane-shaping proteins EHD1 and EHD3, as well as the SNARE component SNAP-29 (Lu et al., 2015). Ciliary membrane establishment also requires components of the Rab-GTPase cascade, which includes the GTPase Rab11, the small Rab GTPase Rab8 and the Rab8 activator Rabin8 (Knodler et al., 2010; Nachury et al., 2007; Yoshimura et al., 2007). Rabin8 and Rab8 accumulate at the centrosomes shortly after induction of ciliogenesis, implying a local activation of Rab8 to promote ciliary vesicle formation (Westlake et al., 2011). Alongside the process of ciliary vesicle establishment, the CP110-Cep97 complex is removed from the mother centriole by targeted protein degradation (Spektor et al., 2007). Removal of CP110-Cep97 allows the extension of microtubules to form the ciliary axoneme and establishment of the ciliary transition zone (Kobayashi et al., 2011; Yadav et al., 2016). The transition zone serves as a diffusion barrier at the base of the cilium (Reiter et al., 2012). The intraflagellar transport (IFT) machinery also controls cilia formation and extension (Follit et al., 2006; Jurczyk et al., 2004; Pazour et al., 2002, 2000; Pedersen and Rosenbaum, 2008). In addition, large cytoplasmic protein complexes, so-called centriolar satellites, are involved in the transport of building blocks to the basal body (Craige et al., 2010; Garcia-Gonzalo et al., 2011; Hori et al., 2014; Jin et al., 2010; Klinger et al., 2013; Kurtulmus et al., 2016).
Distal appendage proteins at the mother centriole are key for the initial steps of cilia biogenesis. Distal appendage components include the C2-calcium-dependent domain protein C2CD3, Cep83 (also known as CCDC41), the sodium channel and clathrin linker protein 1 (SCLT1), Cep123 (also known as Cep89), Fas binding factor 1 (FBF1) and Cep164 (Graser et al., 2007; Joo et al., 2013; Schmidt et al., 2012; Sillibourne et al., 2013; Tanos et al., 2013; Wei et al., 2013; Ye et al., 2014). The protein C2CD3 positively controls centriole length (Thauvin-Robinet et al., 2014). In the absence of C2CD3, centrioles lack Cep83 and, consequently, all additional distal components, given that Cep83 is necessary for binding of Cep123, SCLT1, Cep164 and FBF1 to the mother centriole (Tanos et al., 2013; Ye et al., 2014). Upon depletion of distal components, including Cep164, Cep123 and Cep83, the initial docking of small ciliary vesicles at the mother centriole does not occur. In addition, the inhibitory CP110-Cep97 complex is not displaced from the mother centriole and axoneme extension is blocked. Distal appendages interact with components of the vesicular transport machinery, including Rab8, as well as with protein kinases and phosphatases that influence CP110-Cep97 recruitment (Bielas et al., 2009; Cajanek and Nigg, 2014; Goetz et al., 2012; Humbert et al., 2012; Kuhns et al., 2013; Oda et al., 2014; Schmidt et al., 2012; Tanos et al., 2013; Xu et al., 2016; Ye et al., 2014). Therefore, the current model is that distal appendages are indispensable for vesicle docking at the mother centriole and all subsequent steps thereafter.
Despite the importance of appendage proteins for cilia biogenesis, it is not fully understood how appendages are formed and how they regulate the initial steps of ciliogenesis. Here, we investigated the function of the leucine-rich repeat protein, LRRC45, in cilia formation. LRRC45 was reported to be required for centrosome cohesion at the proximal end of centrioles, as part of the linker that holds duplicated centrosomes together (He et al., 2013). LRRC45 is recruited to the proximal end of the centrioles by the protein C-Nap1 (also known as Cep250; He et al., 2013). In addition, a pool of LRRC45 was shown to associate with mother centriole appendages (He et al., 2013), implying a direct function in ciliation. We now show that LRRC45 associates with the distal appendages of the mother centriole in a Cep83- and SCLT1-dependent manner. LRRC45 was required for FBF1 but not Cep164 or Cep123 centriolar localisation. Depletion of LRRC45 impaired ciliogenesis at the level of axoneme extension independently of C-Nap1. Our data indicate that LRRC45 is required for the establishment of the transition zone and recruitment of Rab8 and satellites to centrosomes but not for removal of the CP110-Cep97 complex or ciliary vesicle docking at the mother centriole. We propose that distal appendage proteins cooperate to control ciliary vesicle docking and axoneme extension at early stages of ciliogenesis.
RESULTS
LRRC45 contributes to cilia formation independently of C-Nap1
LRRC45 was shown to associate with the basal body of primary cilia in retina pigment epithelial (RPE1) cells (Fig. 1A) (He et al., 2013). We also observed that LRRC45 associates with the basal body of ciliated murine fibroblasts, neural stem cells and motile cilia of ependymal cells (Fig. 1A,B). This indicates that LRRC45 may have a function in promoting ciliogenesis in several cell types, including cells with motile cilia.
To understand the function of LRRC45 in ciliogenesis, we depleted LRRC45 using three independent small-interfering (si)RNAs and a pool of 4 siRNAs. Loss of LRRC45 led to a reduction of cilia formation in serum-starved RPE1 cells (Fig. 1C,D and Fig. S1A-D). This defective ciliogenesis was not due to inefficient cell cycle arrest, as the majority of control and LRRC45-depleted cells arrested in the G1/G0 phase (Fig. S1D). In addition, primary cilia in LRRC45 depleted cells were shorter in comparison to control depleted cells (Fig. 1E), indicating that LRRC45 is required for cilia formation and elongation. The cilia-loss phenotype was specific for LRRC45, as it could be rescued by the expression of murine Lrrc45 (Fig. S1E-G). However, we were unable to rescue the strong cilia-loss phenotype obtained with the published Lrrc45 siRNA sequence (siLRRC45-1, Fig. S1G) (He et al., 2013). This suggests that the LRRC45-1 siRNA has an off-target effect that exacerbates the cilia-loss phenotype.
LRRC45 associates with the proximal end of the mother and daughter centrioles independently of the protein C-Nap1 (encoded by CEP250) (He et al., 2013). In addition, a fraction of LRRC45 associated with the mother centriole independently of C-Nap1 (Fig. 1F). Because of this dual localisation, we asked which fraction of LRRC45 contributes to cilia formation. For this, we used CEP250 knockout (KO) cells (Panic et al., 2015). These cells can form cilia similarly to WT RPE1 cells (Fig. 1G). Depletion of LRRC45 in RPE1 CEP250 KO cells caused the same degree of cilia loss as observed in WT RPE1 cells (Fig. 1G). We concluded that LRRC45 regulates ciliogenesis independently of its function at the centrosomal linker.
Recently, a role for the daughter centriole has been implicated in ciliogenesis due to its proximity to the mother centriole (Loukil et al., 2017). We therefore asked whether lack of cilia upon LRRC45 depletion in WT and CEP250 KO RPE1 cells correlated with the distance between both centrioles (Fig. 2). Under serum starvation, mother and daughter centrioles were significantly more separated in CEP250 KO in comparison to WT RPE1 cells (Fig. 2A). Yet, CEP250 KO cells were able to form cilia even when mother and daughter centrioles were more than 2 µm apart (Fig. 2A-D). This indicated that the physical proximity of daughter and mother centriole is not required for ciliogenesis. There was no increase in the distance between mother and daughter centrioles irrespective of whether these cells had a cilium or not in either cycling (Fig. 2E) or serum-starved RPE1 cells lacking LRRC45 (Fig. 2F). Together, these data indicate that defective ciliation upon LRRC45 knockdown is most likely unrelated to the proximity of mother and daughter centrioles.
Cep83 and SCLT1 direct LRRC45 to distal appendages
Our data indicate that the mother centriole pool of LRRC45 contributes to ciliogenesis. Using specific anti-LRRC45 antibodies (Fig. S2), we determined the subcellular localisation of LRRC45 at the mother centriole by super-resolution microscopy (Fig. 3). Pairwise co-localisation analysis between LRRC45 and distal (Cep123, SCLT1, Cep83 and Cep164) or sub-distal (ODF2) components by stimulated emission depletion (STED) microscopy shows that LRRC45 appeared as dotted ring-like structures around centrioles (Fig. 3A-E, top views). Lateral images of the centriole revealed that LRRC45 did not co-localise with ODF2 (Fig. 3A). The LRRC45 signal mostly mapped with Cep83 (Fig. 3B) and partly with Cep164 or SCLT1 (Fig. 3C). Interestingly, Cep123 could be resolved in some cells as one or two rings (Fig. 3D, lower panel). In cells with two rings, LRRC45 localised between the two rings (Fig. 3D), whereas in cells with one Cep123 ring, LRRC45 located mostly proximal to it (i.e. between Cep123 and sub-distal appendages). Moreover, by using three dimensional-structured illumination microscopy (3D-SIM), we found out that LRRC45 at the mother centriole forms rings of 256-272 nm in diameter (Fig. 3F). The size of LRRC45 rings was significantly smaller than the one formed by the distal appendage protein Cep164 (350±36 nm) but similar to the rings of the distal component Cep123 (288±49 nm) or the sub-distal appendage protein ODF2 (260±32 nm) (Fig. 3F). These data indicate that LRRC45 associates with distal appendage components.
Previously, a hierarchical network of distal appendage proteins was unravelled by siRNA depletion experiments (Tanos et al., 2013). To position LRRC45 within this network, we first investigated its localisation upon depletion of appendage components. Cep83 targets the appendage proteins Cep123, Cep164, FBF1 and SCLT1 to the mother centriole (Fig. S3). Centriole association of LRRC45 was significantly decreased upon depletion of Cep83 (Fig. 4A,B) or SCLT1 (Fig. 4C,D), but remained unaffected upon depletion of Cep123, Cep164, FBF1 or ODF2 (Fig. S4A-D). This indicates that LRRC45 is mainly targeted to distal appendages via Cep83 and SCLT1. Upon LRRC45 depletion, the levels of FBF1 at mother centrioles were significantly decreased (Fig. 4E,F), whereas the levels of ODF2, Cep123, SCLT1 and Cep164 did not change in comparison to control-depleted cells (Fig. S5A-D). Together, our data are consistent with the model whereby LRRC45 is positioned between Cep83, SCLT1 and FBF1 in the organisational scheme of distal appendage components (Fig. 4G).
We next employed the yeast two-hybrid system to determine the interaction network of distal appendage proteins, including LRRC45. For this, we used a combination of full-length and truncated constructs to identify their minimal interaction domains (Fig. 5A). We reasoned that interactions obtained by the yeast two-hybrid system would reflect direct interactions, as genes encoding appendage proteins are not present in the yeast genome. Full-length LRRC45 interacted with Cep83 and SCLT1 constructs (Fig. 5B, red squares). LRRC45 failed to interact with Cep123 or weakly interacted with Cep164 (Fig. 5B). LRRC45 C-terminal truncations interacted with truncated forms of FBF1 carrying coiled-coil domains (Fig. 5B). Interestingly, the leucine-rich repeats of LRRC45 (LRRC45 1-222 aa) strongly associated with the coiled-coil domain of SCLT1 (SCLT1 C-terminus, 554-688 aa) but not with the coiled-coil domains of other proteins (Fig. S5), indicating interaction specificity to SCLT1. These results, together with the complete yeast two-hybrid data of appendage components (Fig. S6), led us to propose that LRRC45 interacts with distal appendage proteins Cep83 and SCLT1, and weakly with Cep164 (Fig. 5C and Fig. S6).
LRRC45 is required for Rab8 centrosomal localisation but not for CP110 removal
To determine which step of cilia biogenesis requires LRRC45, we first used transmission electron microscopy to visualise the mother centriole and basal body of non-ciliated cells. Ciliogenesis initiates at the mother centriole with the fusion of small pre-ciliary vesicles, which fuse to form a large ciliary vesicle that caps the distal end of the mother centriole. In control non-ciliated cells, all basal bodies inspected were capped by a ciliary vesicle (Fig. 6A,B). In a large number of non-ciliated LRRC45-depleted cells (63.6%), the basal bodies were associated with a ciliary vesicle, whereas in 36.4% of basal bodies, no vesicles were observed (Fig. 6A,B). This phenotype indicates that cilia membrane docking and fusion initiates in the majority of LRRC45-depleted cells upon serum starvation. This conclusion is consistent with the fact that the endosomal protein EHD1, which is required for pre-ciliary vesicle fusion at the mother centriole (Lu et al., 2015), was present in 60% of control cells and in more than 40% of non-ciliated LRRC45-depleted cells (Fig. 6C,D). The association of EHD1 with the mother centriole was drastically reduced in the absence of Cep164 or Cep123 (Fig. 6C,D), when vesicle docking is severely affected (Schmidt et al., 2012; Sillibourne et al., 2013). Together, these data imply that LRRC45 plays a role in axoneme extension after ciliary membrane establishment.
We further analysed the key steps in the initiation of primary cilium formation. We first looked into the CP110-Cep97 complex, which is a negative regulator of ciliogenesis and its removal from the mother centriole is a pre-requisite for axoneme extension (Spektor et al., 2007). To analyse whether LRRC45 is required for CP110-Cep97 removal at the mother centriole, we followed CP110 centrosomal localisation upon serum starvation (Fig. 6E,F). As expected, CP110 did not decorate the basal body of ciliated cells (Fig. 6E, top panel for representative image). In non-ciliated cells, CP110 was absent from the majority of the basal bodies (labelled by ODF2) in both control and LRRC45-depleted cells (Fig. 6F). The removal of the CP110-Cep97 complex was shown to require distal components including Cep164 and the kinase TTBK2 (Goetz et al., 2012; Schmidt et al., 2012). Consistently, Cep164 and TTBK2 centrosomal levels were not affected by LRRC45 depletion (Fig. S6D and S6F). Our data thus indicate that LRRC45 is dispensable for CP110-Cep97 complex regulation at the mother centriole at early stages of ciliogenesis.
Transition zone establishment occurs after EHD1 docking to the mother centriole and removal of CP110 from the mother centriole (Lu et al., 2015; Yadav et al., 2016). We thus analysed localisation of NPHP1 as a marker of transition zone to investigate whether LRRC45 influences transition zone formation upon serum starvation. As expected, NPHP1 was present at the basal body of control and LRRC45-depleted ciliated cells (Fig. 6G). NPHP1 could also be detected at the basal body of non-ciliated cells lacking LRRC45 (Fig. 6G,H) but at lower levels than in the control (Fig. 6I). This implies that LRRC45 may influence transition zone establishment. To further investigate this possibility, we analysed the localisation of Rab8. Upon serum starvation, Rab8 accumulates around centrosomes in a manner that requires a functional transition zone (Bachmann-Gagescu et al., 2015; Kim et al., 2008, 2014). The enrichment of Rab8 around centrosomes was significantly reduced in non-ciliated LRRC45-depleted cells in comparison to control cells (Fig. 6J,K). However, the loss of centrosomal Rab8 in LRRC45-depleted cells was not as pronounced as in cells lacking Cep164 where no ciliary vesicles, including Rab8a containing vesicles, dock to the mother centriole (Fig. 6K) (Schmidt et al., 2012). These data suggest that LRRC45 is required but not essential for Rab8 vesicle docking at the mother centriole upon induction of cilia biogenesis.
LRRC45 promotes centriolar satellite localisation
To better understand how LRRC45 contributes to ciliogenesis, we asked whether LRRC45 influenced the centrosomal enrichment of centriolar satellites, which are large protein assemblies required for transition zone establishment and cilia formation (Craige et al., 2010; Garcia-Gonzalo et al., 2011; Hori et al., 2014; Hori and Toda, 2017; Jin et al., 2010; Kim et al., 2008; Klinger et al., 2013; Kurtulmus et al., 2016). For this, we analysed the centriolar satellite proteins PCM1 and SSX2IP. In comparison to control cells, the levels of PCM1 and SSX2IP around centrosomes significantly decreased in the absence of LRRC45 (Fig. 7A,B and Fig. S7A). Moreover, a similar phenotype of satellite disorganisation was observed in cells depleted of Cep83, SCLT1, FBF1 and Cep123 (Fig. 7A,B) but not in Cep164-depleted cells (Fig. 7A,B). These results together suggest that distal appendages have a function in centriolar satellite organisation at centrosomes.
To test whether loss of satellite organisation could be a consequence of microtubule loss at centrosomes in response to LRRC45 depletion, we compared the kinetics of microtubule nucleation at centrosomes in control and LRRC45-depleted cells (Fig. S7B). To visualise microtubules, we followed the microtubule plus-end binding protein EB1 (Mimori-Kiyosue et al., 2000). Cold treatment depolymerised microtubules and decreased centrosomal EB1 (Fig. S7B). After raising the temperature to 30°C, microtubules re-formed from centrosomes with similar kinetics in WT and LRRC45-depleted cells (Fig. S7B). Similar results were obtained for cells lacking Cep83 (Fig. S7B). This indicates that distal appendage components, including LRRC45, are not required for microtubule nucleation at centrosomes. Furthermore, the centrosomal levels of the intraflagellar transport protein IFT88, which is transported to the centrosome in a microtubule-dependent manner (Kozminski et al., 1993; Pazour et al., 1998), were not decreased by depletion of LRRC45 (Fig. S6G). This indicates that LRRC45 does not impair microtubule-dependent transport to centrosomes in general.
DISCUSSION
Appendage proteins at the mother centriole play a critical role during the initial steps of cilia formation. Here, we have identified LRRC45 as a component that associates with distal appendages that contributes to the early phases of ciliogenesis (Fig. 7C). Given that LRRC45 localises to the basal body in diverse cell lines, we propose that the function of LRRC45 is conserved from cells that form a single primary cilium, such as stem and differentiated cells, to cells with multiple motile cilia.
LRRC45 is recruited to the proximal ends of mother and daughter centrioles by the linker protein C-Nap1. Super-resolution and quantitative fluorescence microscopy indicate that LRRC45 binds to the distal appendage of the mother centriole in a manner that requires Cep83 and SCLT1. Proximal and distal pools of LRRC45 co-exist at the mother centrioles and association of LRRC45 with distal appendages does not depend upon C-Nap1. This implies that C-Nap1 and Cep83-SCLT1 do not compete for LRRC45 binding. Distal appendage proteins were recently proposed to form two spatially distinct domains, named distal appendage blades and matrix. According to this model, Cep83, SCLT1, Cep164 and Cep123 form defined conical-shaped blades whereas FBF1 makes up the matrix by filling the space between the blades (Yang et al., 2018). The reduced level of FBF1 at the mother centriole observed in the absence of LRRC45 indicates that LRRC45 is a key determinant of FBF1 centriolar localisation. It is therefore tempting to speculate that LRRC45 might be part of the appendage blade complex that is recruited via Cep83 and SCLT1 to help maintain FBF1 at this location.
Recently, the proximity of the daughter to the mother centriole was proposed to be important for ciliogenesis (Loukil et al., 2017). Our data indicate that centriole separation cannot be the underlying mechanism explaining the lack of cilia formation in LRRC45-depleted cells. First, the depletion of LRRC45 did not cause a significant separation of mother and daughter centrioles in RPE1 serum-starved cells, even though these cells did not form cilia. Second, in CEP250 KO cells, which lose the LRRC45 proximal pool (Fig. 1) (He et al., 2013), LRRC45 remained at the distal ends of the mother centriole. In those cells, cilia were formed in a LRRC45-dependent manner even when centrioles were more than 8 µm apart. Therefore, our data do not show any correlation between cilia loss and distance between centrioles. We thus propose that LRRC45 plays a key function in ciliogenesis at distal appendages.
Ciliogenesis is a tightly regulated process that occurs at the G1/G0 phase of the cell cycle and involves the conversion of the mother centriole into the basal body. One critical step in this initial phase is the docking of small vesicles at the distal appendages of the mother centriole, which then fuses to generate a large vesicle that caps the distal end of this centriole (Graser et al., 2007; Joo et al., 2013; Kurtulmus et al., 2016; Schmidt et al., 2012; Sillibourne et al., 2013; Sorokin, 1962; Tanos et al., 2013; Ye et al., 2014). Vesicle fusion is promoted by components of recycling endosomes, including the transmembrane tethering protein EHD1 (Lu et al., 2015). Subsequent steps involve membrane elongation, removal of inhibitory components from the mother centriole and targeted delivery of proteins to the basal body (via IFT- and centriolar satellite-dependent transport) to allow axoneme extension (Fig. 7C) (Sánchez and Dynlacht, 2016). Interestingly, defects at early stages of membrane establishment block CP110-Cep97 removal and axoneme elongation, implying that these are highly coordinated processes. The common concept is that distal appendages are key components for this coordination, yet how this is achieved on a molecular level remains unclear. Our analyses show that, unlike Cep164 or Cep123, LRRC45 is not a major determinant for EHD1 centrosome accumulation or initial steps of ciliary vesicle docking. In the absence of LRRC45, the ciliary vesicle capped the majority of basal bodies; however, axoneme and ciliary membrane extension were blocked. The lack of axoneme extension in the absence of LRRC45 was not due to defects in CP110-Cep97 removal, as the majority of non-ciliated LRRC45-depleted cells had CP110-uncapped mother centrioles. This is in agreement with the fact that LRRC45 did not affect the ability of Cep164 to recruit the kinase TTBK2 to promote CP110 centrosomal removal (Goetz et al., 2012). LRRC45 also did not influence the centrosome levels of IFT88, which is required to promote axoneme extension (Sánchez and Dynlacht, 2016).
Our data indicate that the inability of LRRC45-depleted cells to ciliate is related to defective Rab8-dependent ciliary membrane extension, transition zone formation and centriolar satellite organisation at centrosomes. In the absence of LRRC45, the levels of the satellite proteins PCM1 and SSX2IP around centrosomes were significantly reduced. Similar results were observed after depletion of Cep83, SCLT1, Cep123 and FBF1 but not Cep164. Centriolar satellites have been implicated in transition zone establishment and targeting of Rab8-vesicles to the mother centriole upon induction of ciliogenesis (Hori et al., 2014; Kim et al., 2008; Klinger et al., 2013; Kurtulmus et al., 2016; Lopes et al., 2011; Nachury et al., 2007; Wang et al., 2016). It is thus tempting to speculate that a yet-to-be-identified component required for axoneme extension is targeted to the forming basal body by Rab8-positive vesicles. Alternatively, centriolar satellites might be required for centriolar microtubule remodelling to promote axoneme extension after removal of CP110-Cep97. Future research should thus concentrate on understanding how the function of centriolar satellites and Rab8-vesicles are coordinated by distal appendage components at early stages of ciliogenesis.
We propose that appendages can be sub-divided into three categories or modules based on their function in (I) promoting docking of small ciliary vesicles, (II) recruiting TTBK2 and removing inhibitory CP110-Cep97 complexes and (III) promoting Rab8-positive vesicle docking to initiate membrane extension and axoneme elongation in addition to satellite organisation (Fig. 7C). Both the Cep83-Cep123 and Cep83-SCLT1-Cep164 branch of distal appendages, are required for the early steps of small vesicle docking (Graser et al., 2007; Joo et al., 2013; Schmidt et al., 2012; Sillibourne et al., 2013; Tanos et al., 2013). Cep83-Cep123 has an additional function in centriolar satellite organisation (Sillibourne et al., 2013), which is not shared by the Cep83-SCLT1-Cep164 complex (Cajanek and Nigg, 2014; Graser et al., 2007; Oda et al., 2014; Schmidt et al., 2012). Without both modules, subsequent steps related to TTBK2 recruitment, CP110 removal, Rab8-vesicle docking and axoneme extension are blocked. In contrast, the Cep83-SCLT1-LRRC45-FBF1 module works downstream in this cascade of events to promote Rab8 vesicle docking, transition zone establishment as well as satellite organisation (Fig. 7C) (Graser et al., 2007; Joo et al., 2013; Schmidt et al., 2012; Sillibourne et al., 2013; Tanos et al., 2013; Wei et al., 2013; Ye et al., 2014). Cep164 might also assist in this process, given that Cep164 interacts with components of the Rab8 machinery (Schmidt et al., 2012; Burke et al., 2014). Importantly, the LRRC45 siRNA phenotype is distinct from those seen upon depletion of other distal appendage proteins, suggesting that centriole appendages are not simply a scaffold for vesicle recruitment but provide a multi-task platform of cooperating, rearranging proteins with defined functions during ciliogenesis.
MATERIALS AND METHODS
Plasmids and reagents
HsLRRC45 (BC014109), HsCep83 (NM_016122), HsSCLT1 (BC128051) and OFD1 (BC096344) were amplified from an RPE1 cDNA library; MmLRRC45 (BC023196) was amplified from an NIH 3T3 cDNA library. Briefly, total mRNA from RPE1 or NIH3T3 cells was isolated using RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol, and used for cDNA synthesis using RevetAid First Strand cDNA Synthesis Kit (Thermo Fisher Scientific). HsFBF1 (BC023549) and HsEHD1 (BC104825) cDNAs were purchased from Dharmacon (clones 4109753 and 8143828, respectively). HsCep123 cDNA was a gift from Michel Bornens, Curie Institute, Paris, France (Sillibourne et al., 2013). Coiled-coil domains of the proteins were determined using the COILS prediction programme (Lupas et al., 1991), and truncations were cloned into yeast two-hybrid vectors pMM5, pMM6 and frame-modified pMM5 and pMM6 (Schramm et al., 2001). For rescue experiments, MmLrrc45 (from mouse cDNA), IRES2 (from pQCXIP-GFP) and GFP (from pEGFP-C1) were cloned into pRetrox-TRE3G (Takara Bio) using NEB HiFi assembly (New England Biolabs). HsEHD1 was cloned into pMSCV-Zeo-C1-mNeonGreen [pMSCV-Zeocin, Addgene plasmid #75088 deposited by David Mu (Kendall et al., 2007)]. More detailed information can be found in Table S1.
Cell culture and transfection
h-TERT-immortalised retinal pigment epithelial (RPE1, ATCC, CRL-4000, USA) cells were grown in DMEM/F12 (Sigma Aldrich) supplemented with 10% fetal bovine serum (FBS, Biochrom), 2 mM L-glutamine (Thermo Fischer Scientific) and 0.348% sodium bicarbonate (Sigma Aldrich). Medium of RPE1 cells stably expressing GFP-Rab8a was supplemented with G418 (500 µg/ml, Thermo Fischer Scientific) (Schmidt et al., 2012). NIH 3T3 cells (ATCC, CRL-1658) were grown in DMEM high glucose (Sigma Aldrich) supplemented with 10% newborn calf serum (NCS, PAN-Biotech). HEK293T (ATCC CRL-3216) and GP2-293 cells (Takara Bio) were cultured in DMEM high glucose supplemented with 10% FBS. Adult neuronal stem cells were isolated from sub-ventricular zone of 8-week-old C57BL/6 mouse and cultured in DMEM/F12 supplemented with 0.236% sodium bicarbonate, 6% Glucose (Sigma Aldrich), 50 mM HEPES (Sigma Aldrich), 3.14 µg/ml progesterone (Sigma Aldrich), 4.8 µg/ml putrescine dihydrochloride (Sigma Aldrich), 0.37 U/ml heparin (Sigma Aldrich), 2% B-27 supplement (Thermo Fischer Scientific), 0.1% insulin, transferrin, sodium selenite supplement (ITSS, Sigma Aldrich), 10 ng/ml fibroblast growth factor-basic (FGF-2, Sigma Aldrich), 20 ng/ml epidermal growth factor (EGF, Sigma Aldrich). All cell lines were grown at 37°C with 5% CO2. Transient transfection in HEK293T and GP2-293 cells was performed using polyethyleneimine (PEI 25000, Polysciences). RPE1 cells were transiently transfected with plasmid DNA by either electroporation using the Neon® Transfection System or TurboFect (Thermo Fisher Scientific) according to the manufacturer's protocol. Stable cell lines expressing MmLrrc45 and HsEHD1 were generated using retroviral-mediated transduction. GP2-293 cells were transiently transfected with pRetrox-TRE3G-MmLrrc45-IRES2-GFP or pMSCV-Zeo-C1-mNeonGreen-HsEHD1. 48 h after transfection, supernatant was used to transduce RPE1 cells. Expression of TET-ON inducible constructs was induced by the addition of doxycycline (Sigma Aldrich) at a concentration of 1-10 ng/ml. RPE1 CEP250 knockout cell lines were described previously (Panic et al., 2015). To induce cilia formation, 1.2×105-1.5×105 RPE1 or NIH 3T3 cells were seeded on coverslips in 6-well plates and incubated in serum-free medium for 16-48 h. Cell number was determined using the Luna automated cell counter (Logos Biosystems). At least 100 cells were counted for the quantification of ciliated cells for each experimental condition. All cell lines used in this study were authenticated by Multiplex Cell line Authentication (MCA) and were tested for contamination by Multiplex cell Contamination Testing (McCT) service provided by Multiplexion (Heidelberg, Germany).
Antibodies
Antibodies against human LRRC45 (rabbit, 1:800 and guinea pig, 1:400) were produced against C-terminal recombinant protein (240-670 aa) in rabbits and guinea pigs. In brief, 6×His-tagged antigen was purified from BL21 and used to immunise animals. For purification of the antibodies, GST-LRRC45-C (240-670 aa) was used. Antibodies against human Cep123 (1:400) were produced against the N-terminus of Cep123 (1-230 aa) (Sillibourne et al., 2011) in guinea pigs. Recombinant protein GST-Cep123-N (1-230 aa) was used for purification of the antibodies. Specificity of the antibodies was tested by immunofluorescence depletion of LRRC45 or Cep123 using siRNAs in RPE1 cells.
Primary antibodies used for indirect immunofluorescence were rabbit and guinea pig polyclonal anti-Cep164 (1:500 and 1:2000, respectively; Schmidt et al., 2012); rabbit and guinea pig polyclonal anti-ODF2 (1:100 and 1:500, respectively; Kuhns et al., 2013); guinea pig polyclonal anti-IFT88 (1:250; Gazea et al., 2016). Mouse monoclonal anti-polyglutamylated tubulin clone GT335 (1:2000) was a gift from Carsten Janke (Institute Curie, Paris, France) (Wolff et al., 1992). Mouse monoclonal anti-Chibby (1:100) was a gift from Ryoko Kuriyama (Department of Genetics, Cell Biology and Development at the University of Minnesota, USA) (Burke et al., 2014). Rabbit polyclonal antibodies anti-PCM1 (1:2000) and anti-SSX2IP (1:250) were a gift from Oliver Gruss (Institute of Genetics, University of Bonn, Germany) (Bärenz et al., 2013).
Antibodies from commercial sources used in this study were as follows: rabbit polyclonal and mouse monoclonal anti-ARL13B (1:500, Proteintech, #17711-1-AP and 1:50, UC Davis/NIH NeuroMab Facility, clone N295B/66, #73-287, respectively); mouse monoclonal anti-γ-tubulin clone GTU-88 (1:500, Sigma Aldrich, #T6557); rabbit polyclonal anti-TTBK2 (1:1000, Sigma Aldrich, #HPA018113); rabbit polyclonal anti-Cep83 (CCDC41) (1:400, Sigma Aldrich, #HPA038161); rabbit polyclonal anti-SCLT1 (1:250, Sigma Aldrich, #HPA036561); rabbit polyclonal anti-FBF1 (1:500, Sigma Aldrich, #HPA023677); rabbit polyclonal anti-MAPRE1 (EB1) (1:300, Abcam, #ab53358); rabbit polyclonal anti-CP110 (1:300, Bethyl Laboratories, #A301-343A).
RNA interference
Transfections of siRNA were performed using Lipofectamine RNAiMAX transfection reagent (Thermo Fischer Scientific). siRNA transfections were performed briefly as following, 1.2×105 cells were reverse-transfected in one well of a 6-well plate. For double depletion, 48 h after the initial transfection, cell were dissociated, counted and 1.5×105 cells were seeded in 6-well plates and reverse transfected for the second time. For LRRC45 and Cep83 depletions, double depletion method was used. In experiments where LRRC45 depletion was compared to depletion of other proteins, double depletion method was used. Cycling cells were analysed 48 h hours after the second knockdown. For cilia analysis, 16 h after second transfection, cells were serum starved for 48 h, if not stated otherwise. Phenotypes observed by LRRC45 depletion using individual LRRC45 siRNAs were averaged and shown as single value. Detailed information about the siRNAs, along with final molar concentrations used in the study, can be found in Table S2.
Rescue experiments
RPE1 TRE3G-MmLrrc45-IRES2-GFP cells were incubated in medium containing 0, 1, 5 or 10 ng/ml doxycycline for 6 h prior to LRRC45 double depletion. Cells were then maintained in doxycycline-containing medium.
Indirect immunofluorescence and microscopy
Cells were grown on coverslips (No. 1.5, Thermo Fischer Scientific) and fixed in ice-cold methanol at −20°C for 10 min. Cells expressing fluorescent protein fusions were fixed with 3% PFA for 3 min prior to methanol fixation. Coverslips were coated with 0.1 mg/ml collagen A (Biochrom) for culturing NIH 3T3 cells. For adult NSCs, coverslips were coated with 0.1 mg/ml poly-L-lysine (Sigma Aldrich) and 0.05 mg/ml laminin (Sigma Aldrich) sequentially. Cells were blocked with blocking solution containing 3% IgG-free BSA (Jackson ImmunoResearch), 0.1% Triton X-100 (Sigma Aldrich) in PBS for 30 min and incubated with primary antibodies at 37°C for 1 h. The cells were then incubated with Alexa Fluor 488-, 546-, 594- or 647-conjugated secondary antibodies (Thermo Fisher Scientific) together with DAPI (4′,6-diamidino-2-phenylindole) for 45 min at room temperature. All antibodies were diluted in blocking solution. Coverslips were mounted with Mowiol (EMD Millipore).
Images were acquired as Z stacks using either Zeiss Axiophot equipped with 63× NA 1.4 Plan-Fluor oil immersion objective, and Cascade:1K EMCCD camera using Meta.Morph software, or using Zeiss Axio Observer Z1 equipped with 63× NA 1.4 Plan-Apochromat oil immersion objective, and AxioCam MRm CCD camera using ZEN software.
3D-SIM images were acquired as Z-stacks using a Nikon Ti inverted microscope equipped with 488 nm, 561 nm and 647 nm laser lines, a Nikon Apo TRIF 100× NA 1.49 oil immersion objective and an Andor iXon3 DU-897E single photon detection EMCCD camera. After acquisition, images were reconstructed using NIS Elements program.
STED images were acquired on with STEDYCON STED microscope platform mounted on Zeiss Axio Imager Z2 with a 100× NA 1.46 oil immersion objective. All images were taken at room temperature. Proteins fused to fluorescent proteins were visualised by the fluorescence signal. Figures were assembled in Adobe Photoshop and Illustrator CS3 (Adobe).
Preparation and imaging of brain tissue
Eight-week-old C57BL/6J mice (Jackson Laboratory) were euthanised and brains were extracted and fresh-frozen in Tissue-Tek OCT compound (Sakura FineTek). Coronal sections of 12 µm were cut on a Leica cryomicrotome. Sections were permeabilised with PBS containing 0.1% Triton X-100 prior to fixation with ice-cold methanol at −20°C for 30 min and immunostaining was performed with overnight primary and secondary antibody incubations at 4°C. Sections were imaged on a Nikon A1R confocal microscope equipped with Nikon N Apo 60× NA 1.4 λs oil immersion objective, using NIS Elements software. Mice were obtained from the Deutsches Krebsforschungszentrum (DKFZ) Centre for Preclinical Research facility, housed under standard conditions of 12 h:12 h light:dark and fed ad libitum. All procedures were in accordance with the DKFZ guidelines and approved by the Regierungspräsidium Karlsruhe.
Transmission electron microscopy
RPE1 cells were grown on coverslips, rinsed with PBS and fixed with a mixture of 2.5% glutaraldehyde, 1.6% paraformaldehyde for 30 min at room temperature. The fixative was washed out with 50 mM cacodylate buffer. After post-fixation in 2% OsO4 for 1 h at 4°C, cells were incubated in 0.5% aqueous uranyl acetate overnight, rinsed in water and then dehydrated at room temperature using ascending ethanol concentrations. The coverslips were subsequently set on spurr (Serva)-filled capsules and polymerised overnight at 60°C for 48 h. Serial ultrathin sections (70 nm) were post stained with uranyl acetate and lead citrate and examined at a JEM-1400 electron microscope (JEOL, Tokyo), operating at 80 kV and equipped with a 4K TemCam F416 (Tietz Video and Image Processing Systems GmBH, Gautig). Image brightness and contrast were adjusted in ImageJ.
Microtubule re-growth
Cells grown on coverslips were incubated on ice for 2 h in order to depolymerise microtubules. Cells were fixed in ice-cold methanol after incubation at 30°C for 0, 5 and 20 s, respectively.
Yeast two-hybrid assay
Full-length or partial ORF of the indicated genes were cloned into pMM5 (LexA) and pMM6 (Gal4), and the assay was performed as described (Geissler et al., 1996). Yeast growth conditions were as described (Sherman, 1991). Briefly, plasmids encoding LexA and Gal4 fusion proteins were transformed into yeast strains that have opposite mating types and selected according to their auxotrophy markers. A pool of colonies (more than 30 colonies) from the transformation plates were picked and streaked as individual lines on selection plates. Transformants of LexA fusion proteins were then mated with the Gal4 fusion proteins on non-selective plates for 2 days, and the diploid strains were selected on double selection plates. Two to three days after selection, the plates were covered with Overlay Solution containing 0.25 M sodium phosphate buffer pH 7.0, 0.1% SDS, 10 mM KCl, 1 mM MgCl2, 0.4% low melting agarose and 0.04% X-Gal. Development of blue colour was observed and recorded. The assay with the complete set of genes was repeated twice.
Image processing and analysis
Quantification of fluorescence intensity was performed using maximum projection of images using Fiji and CellProfiler (Carpenter et al., 2006; Schindelin et al., 2012; Schneider et al., 2012). Briefly, images were corrected for background signal by applying a top hat filter, and then centrosomes were segmented using γ-tubulin signal. Identified centrosome areas were expanded by 2 pixels and 10 pixels for intensity measurement of centrosomal proteins and centriolar satellites, respectively. Intensity measurement from replicate experiments were normalised and combined. Intercentriolar distances were manually measured using Fiji. Statistical analyses of fluorescence intensity measurements and ciliation assays were performed using Wilcoxon and two-tailed Student's t-tests; significance probability values are: *P<0.05, **P<0.01, ***P<0.001. Statistical tests were performed in Excel (Microsoft) and KaleidaGraph (Synergy Software).
FACS analysis
Cell cycle profile of control and LRRC45-depleted serum-starved RPE1 cells were determined by analysing total DNA content using propidium iodide staining. Briefly, cells were fixed with ice-cold ethanol (70%), and incubated with staining solution (50 µg/ml propidium iodide, 0.08% Triton X-100, 0.2 mg/ml RNase A, in PBS/1 mM EDTA). Cells were subjected to analysis on a BD FACS Canto.
Acknowledgements
We would like to thank to Marko Panic, Carsten Janke, Michel Bornens, Andrew Fry, Ryoko Kuriyama and Oliver Gruss for sharing reagents; Astrid Hofmann for excellent technical support; Rafael Dueñas-Sánchez for help with antibody production; Berati Cerikan for working on the initial stage of the project and cloning SCLT1 and Cep83; Monika Langlotz from the ZMBH FACS facility for help with the cell cycle analysis, Damir Krunic and Felix Bestvater from the microscopy facility of the DKFZ, and Ulrike Engel from the Heidelberg-University Nikon Imaging Center for providing access to microscopes, help with imaging and image processing. We are grateful to Abberior Instruments GmbH for allowing us access to the STEDYCON STED microscope platform.
Footnotes
Author contributions
Conceptualization: B.K., G.P.; Methodology: B.K., G.K., G.P.; Software: B.K.; Validation: B.K., G.P.; Formal analysis: B.K., G.P.; Investigation: B.K., C.Y., J.S., A.N.; Resources: B.K., S.H., G.K., A.M.-V., G.P.; Data curation: B.K.; Writing - original draft: B.K., G.P.; Writing - review & editing: B.K., C.Y., J.S., A.N., S.H., G.K., A.M.-V., G.P.; Visualization: B.K.; Supervision: G.P.; Project administration: G.P.; Funding acquisition: G.P.
Funding
This project was funded by the collaborative research grant from the Deutsche Forschungsgemeinschaft (DFG) (SFB873) granted to G.P. (Project A14). Core funding for microscopy was provided by the SFB873, Project Z03. B.K. is a member of the Heidelberg Biosciences International Graduate School (HBIGS) and funded by the SFB873 grant. C.Y. is a member of the Helmholtz International Graduate School (HIGS). G.P. holds a Heisenberg Professorship from the DFG (PE1883/3).
References
Competing interests
The authors declare no competing or financial interests.