|Home | About | Journals | Submit | Contact Us | Français|
The primary cilium is a microtubule-based organelle that senses extracellular signals as a cellular antenna . Primary cilia are found on many types of cells in our body and play important roles in development and physiology. Defects of primary cilia cause a broad class of human genetic diseases called ciliopathies. To gain new insights into ciliary functions and better understand the molecular mechanisms underlying ciliopathies, it is of high importance to generate a catalog of primary cilia proteins. In this study, we isolated primary cilia from mouse kidney cells by using a calcium shock method and identified 195 candidate primary cilia proteins by MudPIT (multidimensional protein identification technology), protein correlation profiling, and subtractive proteomic analysis. Based on comparisons with other proteomic studies of cilia, around 75% of our candidate primary cilia proteins are shared components with motile or specialized sensory cilia. The remaining 25% of the candidate proteins are possible primary cilia specific proteins. These possible primary cilia specific proteins include Evc2, Inpp5e and Inversin, several of which have been linked to known ciliopathies. We have performed the first reported proteomic analysis of primary cilia from mammalian cells. These results provide new insights into primary cilia structure and function.
To identify the proteins that localize to mammalian primary cilia, we isolated primary cilia from the mouse IMCD3 cell line, which are derived from kidney collecting ducts, by using a modified calcium shock method (see Supplemental Experimental Procedures). The calcium shock method is conventionally employed for olfactory cilia isolation  and has been used for proteomic analysis of human bronchial and rat olfactory cilia [3, 4]. Immunofluorescence imaging confirmed that our modified calcium shock method efficiently removed primary cilia from cultured IMCD3 cells and isolated fraction includes primary cilia (Figure 1).
Not surprisingly for such a method, we found that the isolated primary cilia fraction included non-ciliary proteins when analyzed by Western blot (data not shown). To separate primary cilia from non-ciliary proteins, we fractionated the isolated primary cilia fraction by velocity sedimentation through a 30–45% continuous sucrose gradient. Western blot analysis of these fractions showed that the peak of acetylated-tubulin (fraction 13–17), a marker for primary cilia, was displaced from the peak of actin (fraction 6–9), which was used as a non-ciliary marker (Figure 2A, B). We confirmed that actin is not enriched in the primary cilium by immunofluorescence microscopy (data not shown). A recent study also showed that F-actin is excluded from cilia and the area around the base of cilia .
Since the actin and acetylated-tubulin rich peaks overlapped extensively, it was not possible to simply pick one fraction and consider it a pure ciliary fraction. Instead, we analyzed each of the fractions 5–22 from the continuous sucrose gradient, which included both the actin and acetylated-tubulin peaks, by MudPIT, a mass-spectrometry-based method in which complex mixtures of proteins can be analyzed without prior electrophoretic separation . MudPIT analysis of the fractions 5–22 detected 2,937 proteins in total (based on a cutoff of 10 spectrum counts), which included known ciliary proteins as well as known non-ciliary proteins (Table S4).
To distinguish ciliary proteins from non-ciliary proteins we employed protein correlation profiling . We calculated the relative abundance of each protein in each of the sucrose gradient fractions using the spectrum count of MudPIT data (see Supplemental Experimental Procedures). Once the abundance profiles were calculated for every protein, we next identified possible candidate primary cilia proteins by correlating their profiles to profiles of known ciliary proteins. The profiles of several known intraflagellar transport (IFT) proteins, plotted in Figure 2C, showed that they all had very similar profile patterns. These IFT protein abundance profiles all peaked between fractions 13–17, which coincided with the cilia peak as judged by Western blotting analysis for acetylated-tubulin (Figure 2A, B). In contrast, non-ciliary proteins had a fractionation profile that was very different from that of IFT proteins (Figure 2D). We performed protein correlation profiling as described in Supplemental Experimental Procedures, comparing the profile of every detected protein with the IFT-derived consensus profile. This analysis identified 379 candidate proteins with similar profile patterns to the IFT proteins (Table S4). These proteins were selected as our initial candidate set for the primary cilia proteome.
Protein correlation profiling is a purely statistical tool. While it can indicate which proteins are more likely than others to be ciliary proteins it will always generate false positives. We therefore augmented our analysis with a subtractive proteomic strategy for discriminating ciliary from non-ciliary proteins independent of profile correlations. We generated IMCD3 cells lacking cilia by treating cells with nocodazole and incubating at a cold temperature (Figure 3). We then performed cilia isolations using the calcium shock method from both normal ciliated IMCD3 cells and non-ciliated nocodazole/cold treated IMCD3 cells. The isolated fractions from ciliated and non-ciliated cells were individually analyzed by MudPIT and compared. Any proteins detected twice or more in the fraction from non-ciliated cells were classified as likely non-ciliary proteins and subtracted from the original list of 379 candidates.
As a final step, we corrected for multiple isoforms and redundant protein identification. (Table S4). The final result of our combined correlation profiling and subtractive proteomics analysis was a list of 195 candidate primary cilia proteins (Table S1).
To validate our candidate list, we compared our candidate primary cilia protein list with other systematic studies, including proteomic, comparative genomic, and promoter-analysis based studies of cilia (Table S2) [3, 4, 8–19]. We found that more than 80% of proteins in our candidate list were also identified in one or more prior systematic studies of cilia and around 75% of candidate proteins were also identified in other proteomic studies of cilia. Thus we infer that our proteomic analysis of primary cilia is likely to contain many bona fide ciliary proteins. We note that 138 of our candidate proteins were also identified in proteomic studies of specialized sensory cilia [3, 8, 9]. Significantly, out of our list of 195 candidates, 46 were never identified in other proteomic studies of cilia.
To more directly verify our candidate list, we randomly cloned a subset of candidate proteins, and stably expressed C-terminal green fluorescent protein (GFP)-tagged and N-terminal FLAG-tagged candidate proteins in IMCD3 cells (Table S3). 8 of 18 candidate proteins were localized to primary cilia in these stably expressing lines as judged by GFP fluorescence (Figure 4) or FLAG-tag staining (data not shown). Thus only about half of the proteins tested showed localization in cilia, and while some of these could well be false negatives due to the peptide tags, it is clear that not all of the proteins in our candidate list are bona fide ciliary proteins. As with any complex organelle proteome, our results are best viewed as giving an enrichment of ciliary proteins, rather than an absolute purification.
During the course of our studies, some of these candidate proteins have already been confirmed their ciliary localization in other tissues or cell lines by other recent studies [20–22]. Interestingly, among these candidates we found that Gtl3 (a homologue of Chlamydomonas BUG22)  is localized to both primary cilia and nuclei (Figure 4B), and it has a domain consistent with it being a possible transcription factor. Wdr11 is mainly localized to centrosomes but is also weakly localized to cilia (Figure 4D). One particularly interesting candidate that we localized to primary cilia, Tsga14 (CEP41, Figure 4C), had been identified in the proteomic analysis of the human centrosome  but was not previously localized to cilia.
We identified 195 proteins as potential primary cilia proteins by using the calcium shock isolation method, MudPIT correlation profiling, and subtractive proteomic analysis. This study is the first proteomic analysis of generic mammalian primary cilia. Previous proteomic analyses of non-motile mammalian cilia focused on specialized sensory cilia, namely the connecting cilium of photoreceptors and olfactory cilia [3, 8, 9], which differ in appearance from “generic” primary cilia but share the same fundamental structure of a microtubule axoneme surrounded by a membrane.
Our primary cilia candidate list includes almost all known IFT proteins except Ift43 (Table S5). We also identified kinesin and dynein IFT motor proteins. Notably absent were any BBS proteins, which are the causative gene products of Bardet-Biedl syndrome and are localized to primary cilia . We do not understand why most of the IFT proteins, but none of the BBS proteins, were detected in our analysis. Nevertheless, the fact that we recovered almost every known IFT proteins, including IFT proteins not used to construct the reference protein profile, confirms that our calcium shock method and protein correlation profiling were effective for identifying bona fide ciliary proteins. Our candidate list also includes proteins that potentially regulate protein import into the cilium. We identified the septins (Sept2, Sept7 and Sept9) in our ciliary proteome. Septins have been reported to form a diffusion barrier that controls protein entry into cilia and regulates ciliogenesis [25, 26]. We also detected the small GTPase Ran as well as importin, which regulate both nuclear and ciliary entry of proteins . However, we failed to detect many known ciliary membrane proteins, such as polycystin-1 and polycystin-2. This may reflect the reduced solubility of such proteins under extraction conditions used for protein separation in MudPIT.
Around 75% of our candidate primary cilia proteins are shared components with motile or specialized sensory cilia (Table S2). Especially, most of IFT proteins and Nme7 were also identified in comparative genomics and X-box promoter analysis (Table S2). We speculate that Nme7 is a part of IFT complex or related with IFT. However, Nme7 may regulate protein transport or signal transduction rather than assembly of cilia . This indicates these proteins are necessary for all types of cilia and flagella. On the other hand, prior to our analysis we had expected to recover two major classes of axonemal proteins in our candidate list: tektins and pf-ribbon proteins [29, 30]. As these proteins are known to be stable and abundant components of the axoneme, we assumed that they would be among the most abundant hits in our list, just as they have been reported to be common in proteomic analysis of motile cilia/flagella and basal bodies [4, 10, 23]. Structural studies have suggested that tektins and pf-ribbon proteins form filamentous structures tightly associated with the outer doublets of the axoneme, which are conserved among all animal cilia, and may therefore be critical determinants of outer doublet architecture in all cilia . Surprisingly, neither tektin nor pf-ribbon proteins were found in the candidate list from our primary cilia proteome. Tektin and pf-ribbon proteins have also not been identified in the proteomes of specialized sensory cilia [3, 8, 9]. Moreover, knock out or knock down of tektin induces defects in ciliary motility but not in ciliogenesis per se [32, 33]. Therefore, it is possible that tektin and pf-ribbon proteins are only components of basal bodies and motile cilia, but not primary cilia.
The remaining 25% of our candidate proteins are possible primary cilia specific proteins because these proteins have not been identified in proteomic studies of motile or specialized sensory cilia. However, some primary cilia specific proteins might be also localized to sensory or motile cilia because Inpp5e was also identified in inner segment of photoreceptor . Some of the primary cilia specific proteins are apparently involved in signal transduction, such as, Evc2, Inpp5e, Inversin, Broad-Minded (Bromi) and Nphp3. Evc2 and Bromi function in hedgehog signal transduction [20, 35], while Inpp5e is involved with phosphatidylinositol and PDGF signaling [34, 36].
Because ciliary defects are linked to many genetic disorders, we were particularly interested to see if any ciliary disease loci would turn up in our candidate list. In fact, multiple candidate proteins were found to correspond to known causative gene products of genetic disorders (Table S6). We identified several known ciliopathy gene products, including Inpp5e, Evc2, Nphp3, Nek8 and Lca5 (see referenes in Supplemental Information). A recent study reported that TSGA14 is a plausible causative gene of autism spectrum disorders . Moreover, WDR11 has been also recently reported as a causative gene of idiopathic hypogonadotropic hypogonadism and Kallmann syndrome . In light of our results that Tsga14 and Wdr11 are present within primary cilia (Figure 4C, D), we hypothesize that the autism and abnormal genital development in such patients might illustrate new symptoms of ciliopathy.
In this study, we isolated primary cilia from cultured mammalian cells and utilized this material to generate the first reported proteome of mammalian primary cilia. Our analysis revealed that the majority of primary cilia components are shared with motile and specialized sensory cilia. However, around 25% of the candidate proteins were only identified in this analysis, including Evc2, Inpp5e, Inversin, Nphp3 and Broad-Minded. These novel candidates, a number of which have already been implicated in ciliopathies and signal transduction, are potentially specific to primary cilia. Tsga14 and Wdr11 are new candidates of causative genes of ciliopathy. The results of this study should serve as a starting point toward a greater understanding of primary cilia functions and mechanistic insights into ciliary diseases.
We thank J. Azimzadeh, M. Chan, P. Crofts, B. Engel, J. Feldman, E. Kannegaard, L. Keller, W. Ludington, S. Rafelski, M. Slabodnick and K. Wemmer for helpful discussions and comments. This work was supported by the US National Institutes of Health grants GM077004 (W.F.M), P41 RR011823 (J.R.Y), the W. M. Keck Foundation (W.F.M) and the Japan Society for the Promotion of Science (H.I).
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.