The ciliary tip has been implicated in ciliary assembly and disassembly, and signaling, yet information on its protein composition is limited. Using comparative, quantitative proteomics based on the fact that tip proteins will be approximately twice as concentrated in half-length compared with full-length flagella, we have identified FAP256 as a tip protein in Chlamydomonas. FAP256 localizes to the tips of both central pair and outer doublet microtubules (MTs) and it remains at the tip during flagellar assembly and disassembly. Similarly, its vertebrate counterpart, CEP104, localizes on the distal ends of both centrioles of nondividing cells until the mother centriole forms a cilium and then localizes at the tip of the elongating cilium. A null mutant of FAP256 in Chlamydomonas and RNAi in vertebrate cells showed that FAP256/CEP104 is required for ciliogenesis in a high percentage of cells. In those cells that could form cilia, there were structural deformities at the ciliary tips.
The cilia tip is a specialized region in the ciliary apparatus that is involved in cilia assembly/disassembly, signaling and sensory function (Johnson and Rosenbaum, 1992; Marshall and Rosenbaum, 2001; Pedersen et al., 2006; Goetz and Anderson, 2010; Gluenz et al., 2010). In comparison to the basal body and axonemal shaft of the cilium, little is known about the ultrastructure or composition of the tip. However, it is an important hub where tubulin subunits are added or removed at the plus ends of the axonemal MTs during ciliary assembly and disassembly (Rosenbaum and Child, 1967; Johnson and Rosenbaum, 1992; Marshall and Rosenbaum, 2001). At the tip, tubulin and other axonemal precursors from the cytoplasm are delivered for assembly by intraflagellar transport (IFT), axonemal turnover products are removed for return to the cytoplasm, IFT is remodeled from long anterograde to short retrograde trains and the anterograde kinesin motor is inactivated with the simultaneous activation of the dynein retrograde motor (Kozminski et al., 1993; Cole et al., 1998; Pedersen et al., 2006; Pigino et al., 2009). Moreover, recent findings showing recruitment of the Gli family transcription activators Gli2 and Gli3, and a regulator of Sonic hedgehog pathway, Sufu, to the tip of the primary cilium upon hedgehog pathway activation, suggest that the ciliary tip acts as an important signaling center during mammalian embryonic development (Haycraft et al., 2005; Liem et al., 2009; Tukachinsky et al., 2010; Goetz and Anderson, 2010).
In addition to the activities primarily involving IFT, specific proteins probably involved in the assembly and disassembly of the ciliary MTs, are localized at the tips. Among these are the MT plus-end-binding proteins EB1 and EB3 and the nonprocessive kinesin motor kinesin-13. The latter has been shown to be involved in MT depolymerization at the plus and minus ends of MTs (Hunter et al., 2003; Wordeman, 2005) and homologs of Kinesin-13 have been linked to axonemal assembly and disassembly through overexpression and knockout studies in a variety of protistans (Blaineau et al., 2007; Chan and Ersfeld, 2010; Dawson et al., 2007). Another member of the kinesin superfamily, Kif19A, which depolymerizes MTs from the plus ends, also localizes at the tips of motile cilia. Mice with mutations of the Kif19a gene showed elongated cilia in brain, oviduct and tracheal epithelial cells, suggesting an important role for this protein in ciliary length regulation (Niwa et al., 2012). Although the Chlamydomonas homolog of kinesin-13 family proteins, CrKinesin-13 was not seen at the flagellar tip, RNAi-mediated depletion of this protein resulted in cells with shorter flagella that failed to regenerate after deflagellation (Piao et al., 2009), suggesting a role for CrKinesin-13 in flagellar assembly. Similarly, depletion of Kif24, a kinesin-13 family protein that specifically depolymerizes centriolar MTs, resulted in aberrant cilia (Kobayashi et al., 2011). Another class of MT binding proteins, the EB family of MT plus-end-binding proteins, which help in recruiting other proteins to the plus ends of MTs, were reported to be localized to the flagellar tips in Chlamydomonas, Giardia and mammalian motile cilia (Pedersen et al., 2003; Dawson et al., 2007; Schrøder et al., 2011 and Brooks and Wallingford, 2012). The Chlamydomonas homolog of an EB family protein, CrEB1, localizes to the tip of the flagellar axoneme and interacts with IFT protein IFT172, indicating its possible role in regulating IFT events at the flagellar tip (Pedersen et al., 2003; Pedersen et al., 2005). EB3 is localized to the tips of motile cilia in bronchial epithelial cells, and depletion studies showed that this protein is required for the formation of primary cilia in cultured human retinal pigment epithelial (RPE1) cells (Schrøder et al., 2011). Moreover, GFP-EB3 was recently reported to be localized to the tip in the motile cilia of Xenopus and the primary cilia of RPE1 cells (Brooks and Wallingford, 2012 and Larsen et al., 2013). Finally, a 97 kDa Tetrahymena protein antigenically related to mammalian kinetochore proteins was also reported to be present at the flagellar tip, but the biochemical composition and function of this protein were not known (Miller et al., 1990).
The fine structure of the ciliary tip was described decades ago through ultrastructural studies in various organisms including Chlamydomonas, Tetrahymena and vertebrates (Dentler and Rosenbaum, 1977; Dentler, 1980; Foliguet and Puchelle, 1986). In Chlamydomonas and Tetrahymena at the tip, plug-like structures with thin filaments were shown linking the A-tubule of each outer doublet to the ciliary membrane and the central pair MTs terminate in a cap-like structure connected to the ciliary membrane. These structures were reported to remain intact during flagellar regeneration and resorption (Dentler and Rosenbaum, 1977; Dentler, 1980). In mammalian tracheal and oviduct cilia, several filamentous structures called lateral spokes connect the A-tubule of the outer doublet to the ciliary membrane: the central pair and A-tubules terminate in an electron-dense capping structure, which, in turn, is connected to the ciliary membrane and claw-like structures called ciliary crowns project from the surface of the ciliary tip (Foliguet and Puchelle, 1986). Similarly, electron-dense structures were also observed at the tip of the flagellar axoneme in other protistans, such as Trypanosoma brucei, Leishmania major and Crithidia daenei (Woolley et al., 2006). Among the few known proteins that localize at the tip, sentan is perhaps the only protein related to a known structure at the ciliary tip. It localizes to the bridging structures between the ciliary membrane and peripheral singlet MTs at the distal tip in tracheal and oviduct motile cilia (Kubo et al., 2008). Although various large-scale comparative genomics and proteomics studies have identified many cilia- and ciliogenesis-related genes (Ostrowski et al., 2002; Li et al., 2004; Pazour et al., 2005; Ishikawa et al., 2012), the proteins that make up these apical tip structures remain unknown.
In this report, we describe a method to identify ciliary tip proteins based on the hypothesis that proteins specifically localized at the tip will be approximately twice as concentrated in half-length in comparison to full-length cilia. The relative abundance of individual proteins in half-length compared with the full-length flagella was quantified by a mass-spectrometry-based proteomic analysis, iTRAQ (Ross et al., 2004). Using this method, we identified a new ciliary tip protein, CEP104/FAP256, which is a conserved centrosomal protein (Jakobsen et al., 2011) that is required for cilia assembly in both RPE1 cells and Chlamydomonas.
Identification of flagellar tip proteins using iTRAQ
Identification of C. reinhardtii flagellar tip proteins was based on the hypothesis that proteins present exclusively at the flagellar tip will be enriched around twofold to threefold when equal quantities of short flagella are compared with the full-length flagella by quantitative mass spectrometry analysis, iTRAQ. Chlamydomonas cells grown on agar plates lose their flagella in a few days (Lewin, 1953) and will synchronously and rapidly regenerate them when transferred to liquid medium. Flagella were isolated from cells that were allowed to regenerate for 10 or 120 minutes. The isolated short (3.45±1.56 µm, n = 103) and full-length (9.8±1.1 µm n = 108) flagella were labeled separately with different iTRAQ tags (see the Materials and Methods) and the relative abundance of individual proteins in the two samples was obtained from the ratios of the tags as determined by quantitative mass spectrometry analysis. The ratios for a particular protein were averaged and proteins were categorized into groups based on their relative abundance in either short or full-length flagella. Proteins with an average fold increase of ≥1.4 were considered to be enriched in the short flagella. Out of the 348 proteins identified at ≥95% confidence cut-off in the proteomic analysis, 74 proteins were found to be enriched ≥1.4-fold in short flagella and 59 proteins were found to be enriched ≥1.4-fold in full-length flagella (supplementary material Table S1). Chlamydomonas end binding protein (CrEB1), a Chlamydomonas MT-plus-end-binding protein that is known to localize at the tip of the flagellum (Pedersen et al., 2003), was found to be enriched twofold in the short flagella and served as a proof of principle that tip proteins should be increased ∼twofold in the half-length flagella. In short flagella, several known Chlamydomonas proteins involved in flagellar assembly, such as IFT polypeptides (Marshall et al., 2005) and motor proteins were enriched. Also enriched in the short flagella were 18 uncharacterized flagellar-associated proteins (FAPs), which would be of particular interest as candidates for new tip proteins.
Eleven Chlamydomonas flagellar proteins that were increased twofold to fourfold and have close human homologs were selected for localization studies in the primary cilia of RPE1 cells. Antibodies against many mammalian proteins are commercially available, and screening could be done quickly by immunofluorescence microscopy (IFM) of the RPE1 cells, rather than making specific antibodies against Chlamydomonas proteins (see supplementary material Fig. S2). One of the proteins that was increased ∼twofold in half-length flagella, FAP256 (supplementary material Table S2), was a conserved uncharacterized protein identified in the flagellar proteome (Pazour et al., 2005). Using FAP256 as query, an NCBI-Blastp search identified only one human protein of low E-value (2E-76) with 85% coverage. Both FAP256 and mammalian CEP104 contain a GlyBP domain (glycine-glutamate-trienylcyclohexylpiperidine binding protein). Using antibodies against the vertebrate homolog of FAP256 (CEP104) on RPE1 cells, CEP104/FAP256 was localized to the tip of the primary cilium and at the centrioles (see results below).
FAP256 localizes to the flagellar tip in full-length, growing and resorbing flagella
To determine the localization of FAP256 in Chlamydomonas flagella, rabbit polyclonal antibodies were raised against a recombinant 150 amino acid internal fragment of this protein. Immunofluorescence analysis using affinity-purified anti- FAP256 antibodies revealed that FAP256 localizes to the tips of the flagella in Chlamydomonas cells (Fig. 1A). FAP256 was also seen at the basal bodies (Fig. 1A). To determine whether FAP256 localizes at the axonemal tip during flagellar growth and resorption, IFM was carried out using antibodies against FAP256 and acetylated α-tubulin (axonemal marker). Flagellar tip and basal body staining of FAP256 was clearly seen in the short flagellar stubs that appear at the early stages of flagellar assembly and during later stages of flagellar elongation (Fig. 1B,C). FAP256 was also localized at the tips in the flagella undergoing resorption (Fig. 1D,E). These results indicate that FAP256 localizes to the flagellar tips in full-grown flagella and remains at the tip of the flagellum during flagellar assembly and disassembly.
FAP256 has a role in flagellar assembly
Ciliary growth occurs by the addition of tubulin subunits at the plus ends of the axonemal MTs at the ciliary tip (Johnson and Rosenbaum, 1992; Marshall and Rosenbaum, 2001). The localization of a conserved centrosomal protein, FAP256, in this position suggests that it has a role in flagellar assembly. To test this possibility, a Chlamydomonas insertional mutant for the FAP256 gene, Roc22, was obtained (Matsuo et al., 2008). Immunofluorescence showed the absence of FAP256 from the flagellar tips and the basal bodies of the mutant cells (Fig. 2A). Western blot analysis also revealed its absence in the null mutants (Fig. 2B). The ability of cells to form flagella after pH-induced deflagellation was analyzed. About 70% of the mutant cells (69.4±8.2%) failed to regenerate their flagella after deflagellation (Fig. 2C). Among the Roc22 cells that regenerated their flagella, the flagella were slightly shorter than those of wild-type cells (CBR34 mt+) (Fig. 2D). The flagella of Roc22 cells were 10.1±1.40 µm (n = 126) compared with 12.8±1.51 µm (n = 121) in the wild-type cells. Together, these results indicate that FAP256 plays a role in flagellar assembly but that its absence is not completely inhibitory to flagellar assembly.
FAP256 localizes to the tips of outer-doublet and central-pair MTs
To determine the precise localization of FAP256 at the flagellar tip, immunofluorescence and immunogold antibody labeling was performed on whole mounts of isolated Chlamydomonas flagellar axonemes splayed apart at their distal tips (see the Materials and Methods). IFM revealed that FAP256 localizes at the distal ends of the splayed out axonemal MTs (Fig. 3A), indicating that FAP256 is present at the tips of both outer-doublet and central-pair MTs (Fig. 3A). To determine this more precisely, immunogold labeling with EM was performed on negatively stained whole mounts of flagellar axonemes. Clear labeling by the gold particles was seen at the distal ends of individual outer-doublet and central-pair MTs (Fig. 3B).
Lack of FAP256 results in abnormalities of the flagellar tip region
Immunofluorescence and immunogold EM results indicated that FAP256 is present at the distal ends of the axonemal MTs at the flagellar tip. Mid-sagittal sections of flagella from control and Roc22 cells were examined with TEM to determine whether the loss of FAP256 altered the ultrastructure of the flagellar tip. As previously described (Ringo, 1967; Dentler and Rosenbaum, 1977; Dentler, 1980; Rogowski et al., 2013), the flagellar tip of wild-type cells appears different from the rest of the axoneme (Fig. 4A,B). At the tip, the matrix of the axoneme becomes much less dense, which is due in part to the lack of radial spokes. The central-pair MTs end together, proximal to the flagellar membrane and are covered with a bipartite endplate. The outer-doublet MTs are reduced to the single A-tubules, which terminate short of the central pair. Because of this early termination of the doublets, the diameter of the flagellum decreases near the distal end (Ringo, 1967). Often a small cone of membrane is seen at the very tip of the flagellum (Fig. 4A,B). These tip specializations were missing or less apparent in Roc22 flagella (Fig. 4C–E). The central-pair MTs often appeared to extend all the way to the membrane and were sometimes of slightly different lengths. The central pair generally did not appear to be capped with an end plate. The outer doublets often extended nearly to the overlying membrane of the tip and radial spokes extending from the A-tubule of the doublet could be seen almost to the very tip. The diameter of the distal end did not appear to be normally reduced, and the entire tip appeared blunt and more rounded than in the wild type, lacking the membranous cone above the central pair. These diverse alterations in the structure of the flagellar tip of Roc22 cells suggest that FAP256 is important in the formation of the overall configuration of the flagellar tip.
CEP104, the mammalian homolog of FAP256, moves to the tip during ciliary growth
The mammalian homolog of FAP256, CEP104, is a conserved protein recently reported in a proteomic analysis of isolated human centrioles and hence named centrosomal protein of 104 kDa or CEP104 (Jakobsen et al., 2011). Localization studies using mouse polyclonal antibodies against this protein revealed that in ciliated RPE1 cells, CEP104 localizes to the tip of the primary cilium (Fig. 5A). The tip localization of CEP104 was also seen in the motile cilia of frozen human tracheal sections (Fig. 5B). In addition to its localization at the ciliary tip in RPE1 cells, CEP104 is present at the distal end of the daughter centriole, but not on the mother centriole (Fig. 5A). This is different from the localization seen in interphase non-ciliated cells, where CEP104 localized to the distal ends of both the mother and daughter centrioles (Fig. 6A) (see also Jiang et al., 2012). In non-ciliated cells, the intensity of the CEP104 signal appeared to be similar at both the centrioles. To determine the fate of CEP104 on the mother centriole during the formation of the primary cilium, CEP104 localization was studied in cells fixed at different time intervals during cilia assembly induced by serum removal. During cilia assembly, even in the very early stages of ciliary growth (i.e. ∼1 µm long), CEP104 became relocalized to the tip of the elongating cilium and remained at the ciliary tip throughout the growth of the cilium (Fig. 6B–D); the daughter centriole retained the distal localization of CEP104 throughout ciliary growth.
CEP104 remains at the ciliary tip during ciliary resorption
To determine whether CEP104 is retained at the tip during ciliary disassembly, cells with fully-grown cilia were induced to disassemble their cilia by serum addition. Ciliary resorption was clearly seen after 3 hours of serum addition with most of the cells having short cilia of ∼3 µm. After 24 hours of serum addition, most of the RPE1 cells in the culture had lost their cilia. CEP104 was localized to the tip of the full-length cilium at 0 hours of serum addition (Fig. 7A) and at various time points during cilia shortening (Fig. 7B). Finally by 24 hours of serum addition, most of the non-ciliated interphase cells again contained CEP104 on both the mother and daughter centrioles (Fig. 7C), as seen at the beginning of ciliogenesis. These results show that throughout the ciliary disassembly process, CEP104 remained at the tip of the cilium.
CEP104 localizes to the centrioles during cell division
To determine whether the centriolar localization of CEP104 is maintained during cell division, actively growing RPE1 cells were observed during growth by staining with CEP104 and γ-tubulin antibodies (marker for centrosomes). As described earlier, CEP104 is present at the distal end of both mother and daughter centrioles in the interphase non-ciliated cells (supplementary material Fig. S1A). In S phase, the centrosome splits and separates and assembly of new daughter centrioles is initiated. Throughout S phase, CEP104 appears as two distinct spots in each centrosome, one spot for each mother and daughter centriole (supplementary material Fig. S1B,C). CEP104 was present at both the spindle poles during metaphase and telophase stages of cell division (supplementary material Fig. S1D,E). These results indicate that in actively dividing RPE1 cells, CEP104 is recruited to the daughter centrioles at very early stages of their assembly and remains associated with the centrosome at all stages of mitosis prior to interphase and ciliogenesis.
CEP104 is indispensible for ciliogenesis in RPE1 cells
The presence of CEP104 at the distal tip on the growing RPE1 cilia raises the possibility that it plays an important role in regulating ciliary assembly similar to that observed with its Chlamydomonas homolog, FAP256. To assess the role of CEP104 in the formation of primary cilia in mammalian cells, RNAi using siRNA oligos specific to CEP104 was used to deplete this protein in RPE1 cells (see the Materials and Methods). Nontargeting siRNAs were used for control ‘mock’ transfections. Western blots of the whole-cell lysates showed that the level of CEP104 protein in the siRNA-treated samples was significantly reduced when compared with the mock-transfected cells, confirming the specificity of anti-CEP104 antibodies (Fig. 8D). Immunofluorescence analysis showed that the amount of CEP104 was highly reduced at both mother and daughter centrioles in siRNA-treated non-ciliated RPE1 cells compared with centriolar localization seen in mock-transfected cells (Fig. 8A). Staining of CEP104 at the tip of the primary cilium and the daughter centriole was completely lost in the depleted cells compared with the staining observed at the tip and daughter centriole in mock-transfected cells (Fig. 8B). The ability of cells to grow cilia was significantly reduced in CEP104-depleted cultures: only 23.7±0.4% of the depleted cells formed cilia, compared with 85.6±3.2% in mock-transfected cells (Fig. 8C). Finally, the length of the cilia that did form in the CEP104-knockdown culture was significantly reduced when compared with mock-transfected cells. The average length of the cilia in CEP104-depleted cells was ∼3.0 µm (3.04±0.69 µm; n = 40) compared with ∼5.4 µm (5.39±0.8 µm; n = 53) in mock-transfected cells. Together, these results indicate that ciliogenesis in mammalian cells, is severely compromised by the depletion of CEP104, similar to Chlamydomonas flagellogenesis seen with the FAP256-null mutant.
Comparative proteomics of short and full-length flagella reveals tip proteins
In this report, we show that proteins stationed at the ciliary or flagellar tip can be identified by quantitative comparison of the proteomes of isolated short and full-length flagella. The reasoning behind the strategy is that proteins present at the tip of short flagella should be about twice as concentrated on a per-protein basis as in full-length flagella. The results presented here show that one of the proteins enriched in short flagella, FAP256, localizes at the tip during flagellar assembly and disassembly and is involved in flagellar biogenesis in Chlamydomonas. The mammalian homolog of FAP256, CEP104, also localizes at the ciliary tip in primary, as well as motile, cilia and is indispensible for ciliogenesis in RPE1 cells.
The present study was carried out using Chlamydomonas because it is a well-studied model genetic system where intact flagella of defined lengths can be isolated. The current approach for identification of proteins enriched in short flagella is a modification of a previous attempt for the identification of flagellar tip proteins by Schneider and colleagues (Schneider et al., 2008) wherein regenerating flagella were isolated by two deflagellation steps and protein identification was done by difference gel electrophoresis. In our case, short and full-length flagella from Chlamydomonas cells grown on plates were isolated by a single deflagellation step; this allowed the isolation of highly purified flagella uncontaminated by other cell organelles (Lewin, 1953; Marshall et al., 2005). The identification of proteins that were twice as concentrated in short versus full-length flagella relied on iTRAQ analysis, which is a sensitive, mass-spectrometry-based procedure used to compare quantitatively, the amount of individual proteins in two different samples, such as full-length and short flagella. CrEB1, a known flagellar tip protein in Chlamydomonas (Pedersen et al., 2003) was enriched twofold in short flagella, indicating that the strategy was successful.
Through iTRAQ analysis, a list of putative new tip proteins was generated for Chlamydomonas flagella, some of which have homologs in vertebrates. Mammalian homologs of six of the Chlamydomonas putative tip proteins identified in the iTRAQ procedure (RABGDI, SBP-1, CDPK-1, MSD-3, Serpin and FBB-11), localized to the basal bodies or centrioles, but not to the primary cilium in RPE1 cells (supplementary material Fig. S2). All of these proteins, however, are present in the Chlamydomonas flagellar proteome (Pazour et al., 2005). The absence of these flagellar proteins in primary cilia of RPE1 cells might be due to the differences in the tip structures between nonmotile primary cilia and the motile Chlamydomonas flagellum. Indeed, the central-pair MTs, which form a part of the distal cap in Chlamydomonas, are not present in primary cilia. Reflecting the difference in tip structure, the composition of the ciliary tips also differs. For example, Sentan is present at the tip of the tracheal and oviduct motile cilia but not in the primary cilia of RPE1 cells (Kubo et al., 2008). Alternatively the inability to detect these proteins at the tip could be due to (1) the very low abundance of these proteins in the primary cilia of RPE1 cells; (2) the inaccessibility of certain antigens in the ciliary compartment as previously documented for IFT20 (Follit et al., 2006); or (3) the methodology used, which enriched proteins that are concentrated during early stages of ciliary growth.
FAP256 is involved in flagellar assembly and disassembly
One of the putative tip proteins identified in the comparative iTRAQ analysis of Chlamydomonas flagella and followed in detail in this study, FAP256, is a centriolar protein that can also be found at the tip of the flagellum. The inability of a majority of FAP256-null mutants to grow flagella after deflagellation suggests that this protein is involved in flagellar assembly, but because some cells do grow flagella, it seems that FAP256 is not essential for the assembly process, or that some other protein can partially assume its function. The occurrence of structural abnormalities at the tips of null mutants that do form flagella suggest that FAP256 is involved in flagellar assembly by regulating the stability of the apical structures at the flagellar tip (described later). In addition to this function at the tip, FAP256 could also have a role at the basal bodies during initial stages of Chlamydomonas flagella formation as seen with its vertebrate homolog, CEP104 (described later).
Flagellar assembly occurs by the addition of axonemal precursors to the distal tips (Witman, 1975; Johnson and Rosenbaum, 1992; Marshall and Rosenbaum, 2001; Rosenbaum and Child, 1967), and resorption likewise appears to occur by MT disassembly at the tips (Johnson and Porter, 1968; Rosenbaum et al., 1969; Marshall and Rosenbaum, 2001). Generally, cells resorb their flagella before cell division (Bloodgood, 1974; Cavalier-Smith, 1974) and flagellar resorption can also be triggered chemically with isobutyl methylxanthine (IBMX) or sodium pyrophosphate (Lefebvre et al., 1980; Lefebvre et al., 1978), yet very little is known about the molecular events that occur in the flagellum upon induction of flagellar disassembly. Recent studies in Chlamydomonas showing increased methylation and ubiquitylation of flagellar proteins during flagellar disassembly suggest an active role of these protein modifications in the disassembly process (Schneider et al., 2008; Huang et al., 2009). Furthermore, various genetic and biochemical data suggest that protein phosphorylation has a key role in flagellar disassembly (Cao et al., 2009). Comparative phosphoproteome analysis of shortening versus steady-state flagella revealed that ∼89 flagellar proteins were specifically phosphorylated in the resorbing flagella compared with the control (Pan et al., 2011). Interestingly, FAP256 was detected in the phosphoproteome of disassembling flagella but not in the control (Pan et al., 2011), indicating that this protein is specifically phosphorylated during flagellar shortening. The presence of FAP256 at the flagellar tip at steady state and during resorption, shown in this study, together with its phosphorylation during flagellar resorption, suggests that this protein plays an important role in maintaining flagellar length and its role might be altered or regulated by phosphorylation during flagellar resorption.
Loss of FAP256 results in structural abnormalities at the flagellar tip
At the flagellar tip, the central pair normally end just short of the flagellar membrane, and are attached to it by a capping structure, whereas the outer doublets become singlets and end ∼100–200 nm from the flagellar tip (Ringo, 1967; Dentler and Rosenbaum, 1977). The A-tubule of each outer doublet has a carrot-shaped plug in it, and each plug has a thin filament connecting it to the flagellar membrane. These structures are present in the A-tubule throughout both assembly and disassembly of the flagellum (Dentler and Rosenbaum, 1977; Dentler, 1980). Although these MT-end-binding structures have been known for over 35 years, nothing is known about their composition or function. Our microscopic analysis with anti-FAP256 antibodies and EM localizations with gold-labeled antibodies showed that in addition to its presence at the central pair, FAP256 also localizes to the tips of individual outer doublets. Because the outer doublets at the tip are reduced to just the A-tubule of the doublet, it is with this part of the doublet that FAP256 is associated.
About 70% of the FAP256-null cells do not grow flagella, and the flagella that are assembled are blunt at their tips rather than tapered as seen with the wild-type cells. The bluntness of the tips in FAP256 mutant flagella is due to the fact that the outer doublets are of the same length as the central pair MTs. This structural abnormality could arise because the outer doublet MTs grew to the tip of the flagellum instead of ending proximal to the central pair or because the central pair MTs were unstable and depolymerized at the tip region. The overall length of the flagellum in the mutant cells is shorter than in wild-type cells, thereby ruling against the possibility of the overgrowth of the outer doublet MTs. A third explanation is that in the absence of FAP256, the flagellum fails to initiate tip formation, never forming the specialized trademarks of the wild-type tip. In support of this third mechanism, the loss of FAP256 had pleotropic effects on the overall structure of the flagellar tip, including loss of the end plate, extension of outer-doublet and central-pair microtubules to the flagellar membrane, attachment of radial spokes to the outer doublet MTs in the tip and loss of the membranous cone at the tip of the flagellum. These diverse effects of the FAP256-null mutants suggest that FAP256 might be required for the structural integrity of the specialized structures seen at the flagellar tip and the loss of this protein affected the overall formation of these apical structures.
CEP104 regulates ciliary assembly by interacting with proteins that suppress cilia formation
The vertebrate homolog of the newly identified tip protein, CEP104 was previously reported as a SXIP-motif-containing EB1-interacting protein. CEP104 localizes on both mother and daughter centrioles in non-ciliated RPE1 cells, but only on the daughter centriole in ciliated cells (Jakobsen et al., 2011; Jiang et al., 2012). In addition, we report here that CEP104 moves from the mother centriole to the tip of the primary cilium during ciliogenesis. Consistent with previous results, our data showed that siRNA-mediated depletion of this protein significantly reduced the ability of RPE1 cells to form primary cilia (Jiang et al., 2012).
During ciliogenesis, the mother centriole functions as a basal body by nucleating cilia when the cells are in the G0 stage of the cell cycle, and, upon re-entry to the division phase, cells resorb their cilia allowing the centrioles to form spindle poles (Sorokin, 1968, Rieder et al., 1979, Wheatley et al., 1996). The molecular basis for the conversion of the mother centriole into the ciliary basal body is not understood. Studies focused on the transition between the mother centriole and the ciliary basal body identified an important centriolar protein complex CP110–CEP97 that inhibits ciliary assembly. Loss of these proteins promotes cilia formation (Spektor et al., 2007). In ciliated cells, these proteins were seen only on the daughter centriole, not on the mother centriole or in the cilium proper (Spektor et al., 2007; Kuhns et al., 2013). The mechanism by which CP110 and CEP97 are removed from the mother centriole during ciliogenesis remains unknown (Bettencourt-Dias and Carvalho-Santos, 2008). Based on previous reports and the results presented in this study, we envisage that CEP104 might be involved in the removal of the CP110–CEP97 inhibitory complex from the mother centriole. Recent studies from Jiang and co-workers, showed that CEP104 interacts with CP110 and CEP97 and recruits CEP97 to the plus ends of cytosolic MTs in HEK293T cells, perhaps through EB1 binding (Jiang et al., 2012). Likewise, the interaction of CEP104 with CP110 and CEP97 might be holding these proteins at the mother centriole before the onset of ciliogenesis. At the very initial stages of ciliogenesis, as shown in this report, CEP104 moves from the mother centriole to the tip of the cilium. Unlike CEP104, neither CP110 nor CEP97 was found in the cilium (Spektor et al., 2007) or at the ciliary tip (our data, not shown). Dissolution of a CEP104–CP110–CEP97 complex might release CP110 and CEP97 from the mother centriole, thereby allowing the centriole to participate in ciliogenesis. Once the complex is broken, CEP104 would be free to assume its function at the tip of the nascent cilium. Furthermore, recent reports showing the role of centriolar kinases MARK4 and Ttbk2 in the removal of CP110 and CEP97 from the mother centriole, and in turn in ciliogenesis (Kuhns et al., 2013; Goetz et al., 2012), raise the possibility that a phosphorylation event might be involved in the removal of CP110 and CEP97 from the mother centriole. Neither MARK4 nor Ttbk2 were shown to directly interact with the proteins of the cilia suppression pathway (CP110–CEP97) or CEP104. MARK4 interacts and phosphorylates mother centriolar protein ODF2 (Kuhns et al., 2013) and Ttbk2 was recently identified as an SXIP containing EB1 binding protein (Jiang et al., 2012) suggesting the occurrence of a complex cascade of events involving MT binding proteins and other centriolar proteins at the mother centriole during initial stages of ciliogenesis. Hence, further studies are needed to elucidate how CEP104, CP110 and CEP97 interact at the onset of ciliogenesis.
Through proteomic analysis of Chlamydomonas flagella, FAP256 was identified as a protein required for ciliogenesis in Chlamydomonas and vertebrate cells. It is essential for the formation of specialized structures at the flagellar tip in Chlamydomonas. The mammalian homolog of this protein CEP104 interacts with centriolar proteins that inhibit ciliogenesis, detaches from these proteins at the onset of cilia formation and moves to the tip of the nascent cilium. These results suggest that CEP104/FAP256 has dual functions: one at the centriole to initiate ciliogenesis and one at the ciliary tip, where it is required for the formation or stability of the specializations characteristic of the tip. Because of its integral role in cilia formation, CEP104 can be viewed as a new candidate ciliopathy protein.
Materials and Methods
Immortalized human retinal pigment epithelial cells (RPE1-hTERT) cells were cultured in DMEM and F12 nutrient mix 1∶1 (Gibco) supplemented with 10% heat-inactivated FBS (Gibco) and 1% penicillin-streptomycin (Penstrep, Sigma). Cultures were grown at 37°C, with 5% CO2 and passaged upon confluency (4–5 days) by trypsinization (0.25% Trypsin-EDTA, Gibco). Cultures up to 20 passages were used for the experiments.
Wild-type C. reinhardtii cells (cc125mt+) were obtained from Chlamydomonas Genetics Center (Durham, NC). Roc22 insertional mutants and CBR34mt+ cultures were obtained from Masahiro Ishiura, Nagoya University, Japan. All the strains except Roc22 insertional mutants were maintained in solid media supplemented with 1.5% agar. Roc22 cells were grown in similar culture conditions as described above supplemented with 10 µg/ml hygromycin. For experimentation, cultures were either grown in liquid minimal medium M1 or Tris acetate phosphate (TAP) medium (Harris, 1989) at 18°C with 12 hour light and dark cycles and constant aeration.
Cilia assembly and disassembly
For the induction of primary cilia in RPE1 cultures, 3×105 cells/well were seeded on sterile coverslips in six-well culture plates (BD Falcon) and allowed to grow in DMEM+F12 nutrient mix supplemented with 10% FBS to ∼80% confluency. The cells were then subjected to serum starvation in DMEM+F12 nutrient mix containing 0.5% FBS for 24 hours. For ciliary assembly studies, cells were fixed at various time intervals to 24 hours after serum removal and processed for IFM. For ciliary disassembly, RPE1 cells growing on coverslips at ∼50% confluency were subjected to serum starvation for 24 hours for cilia formation. Cells with fully grown cilia were then induced for ciliary disassembly by adding the serum back, fixed at various time intervals after serum addition and processed for IFM. The growth of full-length cilia after 24 hours of serum starvation was checked by staining the cells on duplicate coverslips for the ciliary membrane marker Arl13B.
For the transfection of siRNAs in RPE1 cells, ∼3×105 cells were plated on sterile coverslips on six-well plates 1 day before the transfection in DMEM+F12 nutrient mix containing 10% FBS without antibiotics. Next day, the growth medium was replaced with fresh medium without antibiotics 1 hour before transfection. The cells were then transfected using 50 nM CEP104 siRNA (ON-TARGET plus SMART pool, Dharmacon) using 5 µl per well Lipofectamine2000 (Invitrogen) or Dharmafectduo (Dharmacon) transfection reagents as described in the manufacturer’s protocol. For control transfections, 50 nM of nontargeting siRNA pool were transfected in RPE1 cells in similar conditions. The next day, the culture medium was replaced with fresh medium containing 10% FBS and the cells were allowed to grow at 37°C till confluency. Upon reaching confluency, the cells were subjected to serum starvation in DMEM+F12 nutrient mix containing 0.5% FBS for 24–48 hours. Cilia formation and CEP104 depletion was assessed by IFM and western blotting using mouse anti-CEP104 antibodies.
Antibodies and affinity purification
For the generation of rabbit anti-FAP256 antiserum, a synthetic gene fragment of 480 bp from the internal region of C. reinhardtii FAP256 gene (CTGGGCAAG………AGCGGCGGT) without a stop codon (synthesized at Genscript, NJ) was cloned into NdeI and HindIII sites of E. coli expression vector pET21A (Novagen). The cloned fragment was expressed with a C-terminal Hexa-His tag in pET21A in E.coli BL21 (DE3) Gold strain (Agilent Technologies). Purification of recombinant protein was carried out as described (Tammana et al., 2008) with minor modifications. FAP256-transformed E.coli cells were grown at 37°C up to 0.5 OD and induced with 0.5 mM IPTG. Cells were then harvested after 5 hours of growth and resuspended in lysis buffer (50 mM sodium phosphate, 300 mM sodium chloride, 1 mg/ml lysozyme and protease inhibitors) for 30 minutes at 4°C and lysed by sonication. Lysates were further clarified by centrifugation at 14,000 rpm at 4°C for 30 minutes. The pellet was resuspended in solubilization and wash buffer (8 M urea, 0.5 M sodium chloride, 20 mM sodium phosphate, 20 mM imidazole) and applied to Ni-NTA agarose affinity columns. The columns were washed with wash buffer thoroughly and eluted by increasing imidazole concentration to 500 mM. Final elutes were then concentrated, subjected to buffer exchange using centrifugal filters of 10,000 MW (Amicon Ultra, Millipore) and analyzed by SDS PAGE (12% resolving) for purity. Purified recombinant FAP256 protein fragment was used for raising antiserum in rabbits (Pocono Rabbit Farm and Laboratory). Finally anti-FAP256 antibodies were purified from the test bleeds by the blot-strip method using the purified recombinant FAP256 protein fragment. The specificity of the antibodies was tested by western blotting and immunofluorescence analysis. The sources and dilutions of all other primary antibodies used in mammalian and Chlamydomonas immunofluorescence analysis are listed in supplementary material Table S3. Secondary antibodies for western blotting analysis, alkaline-phosphatase-conjugated goat anti-rabbit IgG antibodies and alkaline-phosphatase-conjugated goat anti-mouse IgG antibodies (Invitrogen) were used at 1∶10,000 dilution.
For western blotting, flagella were isolated from Chlamydomonas cc125mt+, Roc22mt+ and CBR34mt+ cells growing in 8L TAP liquid medium according to (Cole et al., 1998). For preparing RPE1 cell lysates, cells growing on tissue culture flasks (BD Labware) were scraped using a cell scraper and washed twice in cold PBS containing protease inhibitors. Lysates from the cell pellets or isolated flagella were prepared by resuspending and boiling them for 5 minutes in 1×SDS sample buffer. Samples were resolved on SDS-PAGE (12% resolving) and electro-blotted onto nitrocellulose membrane. Rabbit anti-FAP256 antibodies and mouse anti-CEP104 antiserum were used for detecting Chlamydomonas FAP256 and mammalian CEP104, respectively. Blots were developed by using colorimetric or chemiluminescent methods.
For immunofluorescence analysis, RPE1 cells growing on coverslips were washed three times in 1×PBS and fixed using cold methanol and acetone 1∶1 mix for 5 minutes at −20°C. The coverslips were air dried and rehydrated in PBS and then blocked for 1 hour at room temperature in blocking solution (0.5% BSA, 10% normal goat serum and cold water fish skin gelatin in 1×PBS). Immunofluorescence analysis on Chlamydomonas strains was done essentially as described previously (Pedersen et al., 2003). Primary antibodies against specific proteins and markers for cilia (mouse anti-acetylated α-tubulin and rabbit Arl13b antibodies) and basal bodies (rabbit γ-tubulin antibodies) were used at dilutions given in supplementary material Table S3. Unless otherwise noted, secondary antibodies for IFM, Alexa-Fluor-488-conjugated goat anti-mouse antibodies and Alexa-Fluor-594-conjugated goat anti-rabbit antibodies (Molecular Probes) were used at 1∶1000 dilution. Finally, the coverslips were mounted on clean glass slides using Prolong gold anti-fade mounting medium containing DAPI (Invitrogen) and the images were captured on a Zeiss LSM510 META confocal microscope using a 100× 1.4 NA (oil) plan apochromate lens.
Chlamydomonas flagellar growth and resorption
For flagellar growth, Chlamydomonas cells grown on TAP-agar plates were scraped into 10 mM HEPES and allowed to regenerate flagella. Cells were fixed at various time intervals and processed for IFM. Flagellar resorption was induced in cells with full-length flagella by treating cells with 20 mM sodium pyrophosphate (Lefebvre et al., 1978); cells were fixed at various time intervals for IFM.
Chlamydomonas flagellar axoneme splaying
Axoneme splaying for IFM was carried out as described (Johnson, 1998) with minor modifications. Briefly, Chlamydomonas cells adhered on the coverslips were treated with HMDEK buffer (10 mM HEPES, pH 7.2, 5 mM MgSO4, 1 mM DTT, 0.5 mM EDTA and 25 mM KCl) containing 0.01% NP40 and 1 mM ATP for 3 minutes, fixed and permeabilized with ice-cold methanol. The slides were then processed for IFM using antibodies against acetylated α-tubulin and FAP256 as described above.
Flagellar and ciliary length measurements
For flagellar length measurements from different Chlamydomonas strains, cells were fixed with 1% glutaraldehyde in M1 medium and were attached to coverslips coated with 0.1% poly-L-lysine (Sigma-Aldrich). Images were captured with a Nikon microscope (Eclipse TE2000; Nikon) equipped with plain Apo 40× objective lens and a camera (Cascade 512B; Photometrics). The lengths of at least 100 flagella were measured for each time point using the MetaMorph software package (MDS Analytical Technologies). For measuring the lengths of primary cilia in siRNA transfection experiments, control and siRNA-treated cells (after 24 hour serum starvation) were subjected to staining with mouse anti-acetylated α-tubulin antibodies and rabbit anti-γ-tubulin antibodies for labeling of ciliary axoneme and basal bodies, respectively. Images were captured using the above microscope at 100× magnification and length measurements were made using MetaMorph software as described above.
Immunogold labeling of whole mounts
Whole mounts of axonemes were prepared as described by Johnson and Rosenbaum (Johnson and Rosenbaum, 1992). Concentrated drops of cells were allowed to settle on formvar-coated nickel grids for 10 minutes. Flagellar axonemes were splayed by inverting the grids onto HMDEK buffer containing 0.01% NP40 and 1 mM ATP for 3–5 minutes. The grids were then treated sequentially with drops of fixative containing 2% PFA and 0.5% glutaraldehyde (15 minutes), PBS (3×5 minutes), permeabilizing solution (PBS and 0.5% Triton X-100, 2×5 minutes), wash buffer (PBS and 0.05% Tween20, 3×5 minutes) and blocking buffer (wash buffer with 0.5% BSA, 10% normal goat serum and 0.02% NaN3, 30 minutes). After blocking, the grids were inverted onto drops of affinity purified anti-FAP256 antibodies (1∶50) in blocking buffer for 3–4 hours at room temperature, followed by washing with wash buffer (4×5 minutes). The grids were then incubated with 12 nm gold conjugated to goat anti-rabbit IgG (Jackson Immunologicals, PA) diluted to 1∶100 in blocking buffer for 2 hours. Grids were then rinsed in wash buffer (4×5 minutes), PBS (2×5 minutes), water (2×5 minutes) and then negatively stained with 2% aqueous uranyl acetate for 1 minute. Finally, uranyl acetate was rinsed from the grids with drops of water.
Transmission EM of flagellar tips
EM of flagellar tips was carried out as described (Pigino et al., 2009) with minor modifications. CBR34 (wild-type) and Roc22 cells were allowed to adhere on poly-L-lysine-coated coverslips for 5 minutes. The cells were fixed on the coverslips with 2.5% glutaraldehyde in MI medium for 30 minutes at room temperature. Tannic acid was added to 0.1% and fixation continued for another 30 minutes. The coverslips were then washed in 10 mM HEPES pH 7.4 and fixed for 30 minutes at 4°C in 1% osmium tetroxide in 50 mM HEPES. After washing in water, the cells were stained with 2% uranyl acetate for 1 hour at room temperature. The samples were dehydrated through ethanol and propylene oxide and embedded in Epon 812. Following polymerization of the resin, the coverslips were dissolved from the Epon with hydrofluoric acid. Desired areas of the resin were cut out and glued to blocks of Epon with 5 minute two-part epoxy (Devcon). Sections were stained with lead citrate (Reynolds, 1963) and uranyl acetate, and were viewed with a JEOL 1230 microscope equipped with a Hamamatsu Orca HR digital camera.
Chlamydomonas flagella isolation for iTRAQ analysis
For iTRAQ analysis, we used a method to isolate short and full-length flagella from C. reinhardtii (cc125mt+) cells growing on air-agar interphase (Lewin, 1953; Marshall et al., 2005) to avoid multiple steps of deflagellation at acidic pH. Chlamydomonas cells grown on agar plates lose their flagella in a few days and will synchronously and rapidly regenerate them when transferred to liquid medium. Briefly, cells growing on solid medium containing TAP with 1.5% agar for 5–6 days were gently scraped into an ice cold solution of 10 mM HEPES and allowed to regenerate flagella at 18°C with constant aeration. Following this, flagella were isolated from aliquots of cells by pH shock at 10 minutes and 120 minutes when the flagella were ∼3.45±1.56 µm (n = 103) and 9.8±1.1 µm (n = 108) long, respectively. Isolation of flagella was carried out essentially as described previously (Cole et al., 1998; Witman et al., 1972).
iTRAQ (Isobaric tag for relative and absolute quantification, Applied Biosystems) is a mass-spectrometry-based protein-profiling method that allows simultaneous identification and comparative quantification of peptides by covalent labeling with reporter ions (Ross et al., 2004). iTRAQ analysis for comparing the relative abundance of individual proteins in short and full-length flagella isolated from C. reinhardtii (CC125mt+) cells was done at the Center for Mass Spectrometry and Proteomics, University of Minnesota. Briefly, the total protein content in the flagellar preparations was estimated, and equal quantities (∼100 µg) of each sample were subjected to protein reduction and cysteine blocking, followed by trypsin digestion resulting in the generation of large number of peptides. These peptides were then labeled with iTRAQ reagents containing signature ions 114, 115, 116 and 117 separately: 114 and 115 for peptides generated from short flagella; 116 and 117 for peptides generated from long flagella. This labeling strategy ensures that the samples will be replicated and internal comparison can be done using 114∶115 and 116∶117 ratios. The samples were then pooled, followed by the removal of trypsin and hydrolyzed iTRAQ reagents on a mixed bed resin (MCX) cartridge. The samples were then subjected to strong cation exchange chromatography for fractionation followed by LC/MS/MS analysis. Mass spectrometry data was collected on 4800 TOF/TOF MALDI analyzer. Protein identification was performed using the C. reinhardtii NCBI Refseq gene database and Augustus version 9.0 database searches. The relative abundance of signature ions among the two samples determines the abundance of a particular protein present in the sample with respect to the other sample. Data analysis was carried out using Protein Pilot software (Applied Biosystems) and peptides were filtered at ≥95% confidence cut-off. To calculate the relative abundance of individual proteins, ratios from different tags were averaged. Proteins were categorized into groups based on their relative abundance in short flagella or full-length flagella. Proteins with average fold abundance of ≥1.4 were considered to be enriched in each category.
We are grateful to Dr Masahiro Ishiura, Nagoya University, Japan for sharing Chlamydomonas mutant strains and Dr Caspary Tamara, Emory University, USA for providing anti-Arl13b antibodies.
The project was conceived and designed by J.L.R.; experiments were designed by T.V.S.T and J.L.R.; experiments were carried out and analyzed by T.V.S.T., D.T. and D.R.D.; manuscript was prepared by T.V.S.T., D.R.D. and J.L.R.
This work was supported by the National Institutes of Health [grant number GM014642 to J.L.R.]. Deposited in PMC for release after 12 months.