|Home | About | Journals | Submit | Contact Us | Français|
Primary cilia are microtubule-based solitary sensing structures on the cell surface that play crucial roles in cell signaling and development. Abnormal ciliary function leads to various human genetic disorders, collectively known as ciliopathies. Outer dense fiber protein 2 (Odf2) was initially isolated as a major component of sperm-tail fibers. Subsequent studies have demonstrated the existence of many splicing variants of Odf2, including Cenexin1 (Odf2 isoform 9), which bears an unusual C-terminal extension. Strikingly, Odf2 localizes along the axoneme of primary cilia, whereas Cenexin1 localizes to basal bodies in cultured mammalian cells. Whether Odf2 and Cenexin1 contribute to primary cilia assembly by carrying out either concerted or distinct functions is unknown. By taking advantage of odf2−/− cells lacking endogenous Odf2 and Cenexin1, but exogenously expressing one or both of these proteins, we showed that Cenexin1, but not Odf2, was necessary and sufficient to induce ciliogenesis. Furthermore, the Cenexin1-dependent primary cilia assembly pathway appeared to function independently of Odf2. Consistently, Cenexin1, but not Odf2, interacted with GTP-loaded Rab8a, localized to the distal/subdistal appendages of basal bodies, and facilitated the recruitment of Chibby, a centriolar component that is important for proper ciliogenesis. Taken together, our results suggest that Cenexin1 plays a critical role in ciliogenesis through its C-terminal extension that confers a unique ability to mediate primary cilia assembly. The presence of multiple splicing variants hints that the function of Odf2 is diversified in such a way that each variant has a distinct role in the complex cellular and developmental processes.
Primary cilia are structures that protrude from the cell surface and play pivotal roles in sensing and transducing environmental cues into intracellular processes. They assemble from basal bodies maturated from mother centrioles. The basal body consists of ciliary rootlets and subdistal and distal appendages. Subdistal appendages position the basal bodies underneath the cell surface by contacting microtubules, and distal appendages anchor them to the plasma membrane. Hence, both structures are considered indispensable for the assembly of primary cilia. The components that play a crucial role in the assembly and function of these appendages are beginning to emerge.
Outer dense fiber protein 2 (Odf2) was originally identified as a component of sperm outer dense fibers that is thought to play an important role in proper sperm tail function.1 Later, Nakagawa et al. suggested that chicken and mouse Odf2s may also localize to mother centrioles.2 Independently of this work, Lange and Gull reported the discovery of a ~96-kDa protein called Cenexin (derived from the Latin, senex for old man, and “C” in Cenexin for centriole) that specifically localizes to matured mother centrioles.3 A monoclonal antibody that detects Cenexin was later used to isolate rat cDNAs (GenBank accession number AF162755.1 for rat Cenexin1 and AF162756 for rat Cenexin2). Comparative analyses of primary amino acid sequences from chicken and murine Odf2s with those of rat Cenexins revealed for the first time a molecular link between sperm tail cytoskeleton and mother centriole. Subsequent studies demonstrated that human Cenexin1 (hCenexin1) (DQ444714; recently renamed as Odf2 isoform 9, NP_002531.3) is a splicing variant of Odf2 that is highly expressed in somatic cells.4 Interestingly, deletion of the ODF2 locus (odf2−/−), which eliminates both Odf2 and its splicing variant Cenexin1, in mouse F9 embryonic carcinoma cells results in complete loss of the appendages of mother centrioles, leading to a failure to generate primary cilia.5 Providing Odf2 to the odf2−/− mutant cells has been shown to remedy the ciliary defect.5 However, whether the Odf2 construct used in these studies encodes the bona fide Odf2 protein (638 residues), whose ~70-kDa product is largely associated with sperm tails, or contains the ~190-residue-long C-terminal extension specifically found in the 93-kDa Odf2-splicing variant (note that Lang and Gull estimated the molecular weight of Cenexin1 at ~96 kDa), called Cenexin1 (805 residues), has not been clearly indicated.
We have demonstrated that Cenexin1, lacking the Odf2-specific 20 residue-long C-terminal sequence (residues 619–638 in hOdf2), but bearing instead the Cenexin1-specific C-terminal extension (residues 614–805) (Fig. 1A), is the major splicing variant expressed in somatic cells and tissues and is localized to centrosomes and basal bodies.6 A recent report shows that Cenexin1 lacking the N-terminal 187 residue is sufficient to induce primary cilia in mouse F9 odf2−/− cells.7 On the other hand, Odf2 has been shown to localize along the ciliary axoneme in serum-starved hTERT-RPE1 cells.6,8 Rab8a, a small GTPase that is important for vesicular trafficking of membrane proteins and primary cilia assembly was suggested to specifically interact with the Odf2-specific 20-residue-long motif.8 These observations suggest that Cenexin1 and Odf2 may play distinct roles in primary cilia assembly/disassembly processes. However, investigation on the exact role of Cenexin1 or Odf2 has been difficult, largely due to the incomplete depletion of these two proteins by siRNA-based approaches.
In this communication, we report that Cenexin1, an Odf2 isoform containing the unique C-terminal extension, but not Odf2, is necessary and sufficient to induce ciliogenesis. These findings resolve the previous puzzlement about the physiological role of Odf2 and Cenexin1 and prove the functional diversification of these two splicing variants in the regulation of distinct cellular processes.
To closely investigate the potential role of human Cenexin1 and/or Odf2 (hCenexin1 and/or hOdf2, respectively) in ciliogenesis, we first re-examined the previously reported interaction8 between Rab8a and hCenexin/hOdf2 after serum starvation, a condition that promotes ciliogenesis. An hOdf2 splicing variant (hODF2 var) containing an N-terminal 19-residue insertion, but lacking the hCenexin1-specific C-terminal extension,8 was also included for comparison. To this end, HEK293T cells were first co-transfected with Rab8a and either one of hCenexin1, hOdf2 and the hOdf2 variant (Fig. 1A). Twenty-four hours after transfection, the resulting cells were cultured under serum-starvation conditions for an additional 30 h before harvest. Since guanosine 5′-triphosphate (GTP) loading onto Rab8a may alter the interaction, prepared total cellular lysates were incubated with 200 μM of guanosine 5′-diphosphate (GDP), GTP or a non-hydrolyzable GTP analog, GTPγS, before co-immunoprecipitation analyses. Results showed that Rab8a efficiently interacted with hCenexin1 in the presence of either GTP or GTPγS (Fig. 1B). This interaction was, however, almost completely annihilated in the presence of GDP (Fig. 1B), suggesting that GTP loading of Rab8a is required for the event. Under the same conditions, regardless of the presence or absence of GTP, GTPγS or GDP, both hOdf2 and its variant lacking the hCenexin1-specific C-terminal extension only marginally interacted with Rab8a (Fig. 1B). This result supports our previous observation that hCenexin1, but not hOdf2, plays an important role in ciliogenesis.6 It is also in line with a recent observation that shows the importance of the C-terminal extension in ciliogenesis.7 However, this finding is in disagreement with the previous observations that Rab8a interacts with hOdf2 and its variant, but not hCenexin1, in yeast two-hybrid analyses,8 and that bacterially expressed Rab8a interacts with the hOdf2 variant through the Odf2-specific C-terminal 20 residues (residues 619–638).8 The apparent discrepancy could be attributed to different experimental systems (mammalian transfection system vs. yeast and bacterial systems) used for interaction studies.
To closely examine the function of Odf2 and Cenexin1 in ciliogenesis, we then took advantage of the murine F9 odf2−/− cells5 lacking both endogenous Odf2 and its splicing variant Cenexin1 (mOdf2 and mCenexin1, respectively), which were expressed from the same ODF2 genetic locus. The cells were infected with lentiviruses to stably express wild-type or mutant forms of hCenexin1 or hOdf2 (Fig. 2A). The hCenexin1 (S796A) mutant, defective in polo-like kinase 1 (Plk1)-binding,6 and the hCenexin1 (1–613) truncation mutant, lacking the entire C-terminal extension,6 were included to examine the importance of hCenexin1-dependent Plk1 recruitment and C-terminal extension-dependent centrosome localization, respectively, in regulating ciliogenesis. To determine the potential dominant-negative effects of the above constructs, cells expressing each of these constructs in the wild-type genetic background were also analyzed. The expression levels of hCenexin1 and the hCenexin1 (S796A) mutant were much lower than those of hOdf2 and hCenexin1 (1–613) (Fig. 2A), likely because of the inefficiency of expressing the full-length protein.
Using the above cell lines, we then investigated the ability of various hCenexin1 and hOdf2 constructs to localize to basal bodies and to induce primary cilia. Using an antibody that specifically detects both Cenexin1 and Odf2 equally by interacting with the common middle region of the protein,4 we observed distinct fluorescent signals (mostly from mCenexin1 because of its high abundance in comparison to that of mOdf2 in somatic cells) associated with the basal bodies in wild-type F9 cells (ODF2+/+ + vector) (Fig. 2B). These signals were not apparent in the odf2−/− cells expressing the control vector. Under these conditions, exogenously expressed hCenexin1 and the hCenexin1 (S796A) mutant defective in Plk1 binding6 efficiently localized to the basal bodies, whereas hOdf2 and hCenexin1 (1–613), lacking the C-terminal extension, did not (Fig. 2B). These results support our previous findings6 that the C-terminal extension is required to target hCenexin1 to basal bodies/centrosomes in a manner independent of Plk1 binding.
Decorating primary cilia with an acetylated tubulin antibody revealed that approximately 10% of ODF2+/+ cells produced primary cilia, whereas less than 2% of odf2−/− cells did so under the same conditions (Fig. 2B and C). Expressing any of the hCenexin1 or hOdf2 construct in the wild-type genetic background did not significantly alter the fraction of cells with primary cilia, suggesting that these constructs do not induce a detrimental dominant-negative effect in primary cilia formation (Fig. 2C, left part). Remarkably, providing either hCenexin1 or hCenexin1 (S796A) fully rescued the ciliary defect associated with the odf2−/− mutation, despite its low expression level in comparison to the level of endogenous mCenexin1. This observation suggests that only a small fraction of endogenous mCenexin1 protein is likely involved in promoting ciliogenesis, and that a low amount of exogenously expressed wild-type hCenexin1 or the hCenexin1 (S796A) mutant is sufficient to remedy the ciliogenesis defect associated with the odf2−/− mutation. In contrast to these findings, expressing hOdf2 or hCenexin1 (1–613) lacking the C-terminal extension, failed to rescue the defect, even though its expression level was far greater than that of the full-length hCenexin1 (Fig. 2A). Thus, hCenexin1 is necessary and sufficient to promote primary cilia assembly.
hCenexin1 distinctively localizes to basal bodies, while hOdf2 localizes along the axoneme of primary cilia.6,8 These observations suggest that the two proteins either cooperate to promote ciliogenesis or function independently to carry out unrelated events at two different subciliary structures. To investigate these possibilities, F9 odf2−/− cells expressing hCenexin1 were superinfected with lentiviruses expressing hCenexin1, hCenexin1 (S796A), hOdf2, or hCenexin1 (1–613). Subsequent examination of the resulting cells revealed that, although these proteins were expressed at levels similar to those in Figure 2A, they failed to significantly increase the capacity of the hCenexin1-expressing odf2−/− cells to generate primary cilia (Fig. 2D and E). These results suggest that hCenexin1 does not cooperate with hOdf2 to promote ciliogenesis. This finding does not eliminate the possibility that hOdf2 contributes to ciliogenesis independently of hCenexin1.
Next, we performed immunoelectron microscopy (IEM) to precisely determine the sub-basal body structures to which hCenexin1 and/or hOdf2 localize. In F9 ODF2+/+ cells, endogenous mCenexin1 (mostly mCenexin1 because mOdf2 is much less abundant) was found to be associated with distal/subdistal appendages (Fig. 3A), which agrees with the previous observation.5,7 To investigate whether exogenously expressed hCenexin1 or hOdf2 has the capacity to localize to this site, we used odf2−/− cells expressing either hCenexin1 or hOdf2 to eliminate any background signals stemming from endogenous mCenexin1 and mOdf2 proteins. The results showed that in ~5–10% of the centrosome-containing cells, out of a total of more than 800 thin-sectioned cells, signals immunoreactive against hCenexin1 were clearly found to be associated with the distal/subdistal appendages (Fig. 3A). In stark contrast, we failed to observe hOdf2 signals at or around basal bodies under the same conditions. Consistent with these results, we observed that odf2−/− cells expressing exogenously introduced hCenexin1, but not hOdf2 or the control vector, were able to generate apparently normal primary cilia (Fig. 3B). This finding is in line with the immunostaining data shown in Figure 2B and C. Taken together, these data conclusively indicate that the 93-kDa hCenexin1, abundantly expressed in somatic cells and tissues,6 is the Odf2 variant that is required for primary cilia assembly. Our data also indicate that the C-terminal extension, found only in hCenexin1 and its homologs in other eukaryotic organisms, is necessary for the function of hCenexin1 in promoting ciliogenesis.
Although the above results showed that hCenexin1, but not hOdf2, was critical for inducing ciliogenesis, the underlying mechanism of how hCenexin1 promotes ciliogenesis remains elusive. A recent report suggests that Chibby localizes to the distal end of mother centrioles and plays an important role in ciliogenesis.9 Chibby appears to interact with and function at a point downstream of hCenexin1,9 suggesting that the hCenexin1-Chibby interaction could be important for proper primary cilia assembly. Therefore, we investigated whether hCenexin1 and/or hOdf2 are capable of recruiting Chibby to the basal bodies of serum-starved F9 cells. In F9 ODF2+/+ cells, both hCenexin1 and Chibby were distinctly localized to the basal bodies. However, in F9 odf2−/− cells (lacking the somatic variant, mCenexin1, and the sperm-tail-associated mOdf2), Chibby localization was severely disrupted (Fig. 4A and B). Providing hCenexin1 to the odf2−/− cells fully restored the Chibby localization to the basal bodies. On the contrary, providing hOdf2 failed to recruit Chibby to these structures under the same conditions (Fig. 4A and B). Both ODF2+/+ and odf2−/− cells appeared to normally recruit γ-tubulin to their respective basal bodies under these conditions. These results suggest that hCenexin1 is specifically required to recruit Chibby to basal bodies and to promote primary cilia assembly at these sites. Similar experiments performed using asynchronously growing F9 cells also showed that hCenexin1, but not hOdf2, was sufficient to recruit Chibby to mother centrioles (Fig. S1).
After the initial isolation of Odf2 as a major component of sperm tail fibers, studies have revealed the presence of several Odf2 splicing variants6,7,10 (see also Fig. S2). However, whether Odf2 and its variants have overlapped and/or distinct functions remains to be clarified. Interestingly, unlike the sperm tail-enriched Odf2, one of the variants, Cenexin1, bears an unusual C-terminal extension and is highly expressed in somatic cells and tissues.6 Moreover, Odf2 localizes along the axoneme of primary cilia, whereas hCenexin1 localizes to basal bodies in mammalian cultured cells.6,8 These observations suggest that Odf2 and Cenexin1 may play either differential roles to mediate distinct cellular events or concerted roles to efficiently promote common cellular processes. As one of the experimental approaches to discern the physiological functions of these two variants, we closely investigated whether and, if so, how Cenexin1 and/or Odf2 contribute to primary cilia assembly by expressing these proteins in mouse F9 embryonic carcinoma cells lacking endogenous Odf2 and its splicing variants, including Cenexin1. Several lines of evidence provided here demonstrated that Cenexin1, but not Odf2, is necessary and sufficient to induce ciliogenesis. First, Cenexin1, but not Odf2, preferentially interacted with GTP-bound Rab8a, a component critically required for proper ciliogenesis. Second, Cenexin1 expression, but not Odf2, fully rescued the primary cilia assembly defect associated with the loss of the ODF2/CENEXIN1 locus. Third, Odf2 co-expression failed to enhance the ability of Cenexin1 to promote ciliogenesis. Fourth, Cenexin1, but not Odf2, was required for proper basal body localization of Chibby, a component shown to be required for basal body formation and proper ciliogenesis.9 Consistent with these findings, Cenexin1, but not Odf2, was found to localize to the distal/subdistal appendages of mother centrioles/basal bodies (Fig. 3A). The differential expression and function of hCenexin1 and hOdf2 are summarized in Table 1.
Notably, there are at least 10 distinct Odf2/Cenexin1 isoforms reported to be expressed in human cells (Fig. S2). Among them, four of them (isoforms 1, 2, 9 and 11) possess the unique C-terminal extension that is demonstrated to play a crucial role in centriole localization and ciliogenesis (Fig. 2). Except the isoform 2, three of them also contain exon3b-encoded peptide fragment in their N termini (Fig. S2). The exon3b fragment is suggested to play a role in the localization of a murine isoform to the centrosomes and to the primary cilium.11 However, close examination of various hCenexin1 truncation mutants revealed that the C-terminal region of hCenexin1 was fully capable of localizing to centrosomes and its localization efficiency was indistinguishable from that of the full-length hCenexin1 (Fig. S3). In stark contrast, the N-terminal fragment (1–249) containing the exon3b-encoded region failed to exhibit any detectable centrosome-associated fluorescent signals under various expression levels (Fig. S3), hence furthering our previous observation.6 Since the exon3b-encode fragment itself is not sufficient to localize to the centrosomes,11 one possibility is that it functions cooperatively with other domains of hCenexin1 to promote this process.
Consistent with a crucial role of the C-terminal region of hCenexin1 in the subcellular localization, the hCenexin1 (1–613) truncation mutant containing the exon3b fragment but lacking the C-terminal extension failed to induce any detectable level of primary cilia (Fig. 2). Moreover, Kunimoto et al. reported that the N-terminal hCenexin1 (1–344) truncation form containing the exon3b fragment failed to induce any detectable level of primary cilia, whereas the C-terminal hCenexin1 (188–805) mutant lacking the exon3b-encoded fragment induced primary cilia as efficiently as the full-length hCenexin1 (1–805).7 These observations suggest that the C-terminal hCenexin1 (188–805) region is sufficient to induce primary cilia, and that the exon3b fragment is not required for this event. Therefore, we propose that the C-terminal region confers most, if not all, of the ability of hCenexin1 to localize to the centrosomes and to promote primary cilia formation. Whether and, if so, how the exon3b-encoded fragment functions in concert with the other region(s) of the Cenexin1, such as the C-terminal extension, to promote the physiological function of hCenexin1 in ciliogenesis or other cellular processes remains to be further investigated.
It should be noted that, although our data demonstrate that Cenexin1 is sufficient for inducing primary cilia assembly, they do not eliminate a potential role for Odf2 in regulating this process. Since Odf2 distinctly localizes along the axoneme of primary cilia, one possibility is that its role could be confined to this subciliary structure. Notably, however, a failure to induce any additive effect on primary cilia assembly by the additional expression of Odf2 in Cenexin1-expressing cells hints that Cenexin1 does not require Odf2-dependent events to mediate ciliogenesis. The exact role of Odf2 in this process remains to be further investigated.
In this study, we demonstrated that one of the splicing variants of Odf2, called Cenexin1, plays a crucial role in ciliogenesis by localizing to basal bodies in somatic cells. These findings are in sharp contrast to the function of its prototype, Odf2, which primarily localizes along the axonemes of sperm tails to protect the tails from shear stress.12 Notably, Cenexin1 was required for normal localization of Chibby not only to basal bodies but also to centrosomes, suggesting that besides its role in primary cilia assembly at basal bodies, Cenexin1 may play a role in other cellular processes during the cell cycle. In support of this view, we have shown that Cenexin1 functions as a scaffold to recruit Plk1 to centrosomes and to promote bipolar spindle formation and mitotic progression in somatic cells.6 Given the presence of multiple Odf2 splicing variants, it is therefore highly likely that each of these variants has its own specific physiological function in distinct cellular and developmental processes. Hence, the multitude of splicing variants and their differential expressions are likely fundamental to increasing the biodiversity of proteins encoded by the same genetic locus, and therefore vital for coping with diverse cellular needs in complex eukaryotic organisms.
PCR-amplified hCenexin1 and hOdf2 fragments were subcloned into a pCI-neo-3xFlag vector (M3133 and M3134). The pCI-neo-3xFlag-hOdf2 isoform 4 (M3132) construct was generated by inserting the N-terminal 19-residue fragment into the hOdf2 coding sequence. Rab8a was amplified with PCR and cloned into a pCI-neo-HA vector (M3143).
Embryonic carcinoma F9 ODF2+/+ and odf2−/− cells were cultured in Dulbecco’s modified Eagle’s medium supplemented with 15% or 10% fetal bovine serum, respectively, at 37°C in the presence of 5% CO2. For the lentivirus generation, pHR′-CMV-ΔR8.20vpr (M891) and pHR′-CMA-VSV-G (M892) were co-transfected into HEK293T cells with pHR-CMV-SV-puro-hCenexin1 (M895), -hCenexin1 (S796A) (M896), -hODF2 (M623) or -hCenexin1 (1–613) (M897). Resulting cells were washed and provided with fresh medium 6–8 h post-transfection, and then viruses were collected two days post-transfection. F9 cells infected with the viruses were selected with 2 μg/ml of puromycin.
Transfected cells were lysed in TBSN buffer [20 mM Tris-Cl (pH 8.0), 150 mM NaCl, 0.5% NP-40, 5 mM EGTA, 1.5 mM EDTA, 20 mM p-nitrophenyl phosphate and protease inhibitors] and total lysates were incubated with 200 μM of GTPγS, GTP, or GDP for 1 h at 4°C. The resulting lysates were incubated with 2 μg of rat anti-HA antibody (Roche) for 4 h at 4°C, and then added with protein G-sepharose beads (Santa Cruz Biotechnologies) and incubated for an additional 2 h before immunoprecipitation. Immunoprecipitates were separated with sodium dodecyl sulfate-PAGE (SDS-PAGE) and transferred to polyvinylidene difluoride (PVDF) membrane for immunoblotting analyses.
Immunoblotting analyses were performed by incubating PVDF membranes with primary antibodies followed by appropriate horseradish peroxidase-conjugated secondary antibodies (GE Healthcare Life Sciences). Immunoreactive signals were detected by the enhanced chemiluminescence detection system (Pierce). Primary antibodies used in this study are anti-Flag (Sigma), rabbit anti-HA (Berkeley Antibody) and anti-Cenexin1/Odf24 antibodies.
Indirect immunostaining was performed as previously described.13 All the appropriate secondary antibodies [Alexa Fluor 488 (green)- and Alexa Fluor 594 (red)-conjugated antibodies] were purchased from Invitrogen. Confocal images were acquired using a Zeiss LSM 510 system mounted on a Zeiss Axiovert 100 M microscope. To visualize primary cilia, F9 cells were cultured on coverslips for 72 h, and then placed on ice for 30 min to depolymerize microtubules before fixation. Antibodies used for immunostaining analyses were anti-acetylated α-tubulin T7451 (Sigma), anti-Cenexin1/Odf2 (which detects both human and mouse Cenexin1 and Odf2 proteins through the common middle region),4 anti-Chibby (Santa Cruz Biotechnologies) and anti-γ-tubulin (Santa Cruz Biotechnologies).
To determine the localization of hCenexin1 at the primary cilia in odf2−/− cells expressing hCenexin1, IEM was performed as previously described.14 Briefly, cultured cells were fixed in 4% formaldehyde and 0.05% glutaraldehyde in phosphate-buffered saline, and were scraped and pelleted. The resulting cell pellets were dehydrated in a series of cold ethanol, infiltrated in the mixture of 100% ethanol and LR White, embedded in LR White resin and cured in a 55°C oven. The thin sections (80 to 90 nm) were mounted on nickel grids and incubated in a serial dilution of anti-Cenexin1/Odf2 (which detects both Cenexin1 and Odf2 equally well), followed by immunogold-conjugated secondary antibody (Electron Microscopy Science). The sections were counterstained with uranyl acetate and lead citrate, and examined in the EM (Hitachi 7600). Images were captured with a digital camera (AMT).
This work was supported in part by National Cancer Institute (NCI)’s intramural grants (K.S.L.), the Global Frontier Project grant (NRF-M1AXA002–2012M3A6A4054949) (K.W.L.) and the World class Institute research program (B.Y.K.) of the National Research Foundation of Korea funded by the Ministry of Education, Science and Technology of Korea, and Korea Basic Science Institute (KBSI)’s International Joint Research program Grant F30601 (J.K.B).
No potential conflicts of interest were disclosed.
Supplemental materials may be found here:
Previously published online: www.landesbioscience.com/journals/cc/article/23585