CCTα and CCTδ Chaperonin Subunits Are Essential and Required for Cilia Assembly and Maintenance in Tetrahymena

Background The eukaryotic cytosolic chaperonin CCT is a hetero-oligomeric complex formed by two rings connected back-to-back, each composed of eight distinct subunits (CCTα to CCTζ). CCT complex mediates the folding, of a wide range of newly synthesised proteins including tubulin (α, β and γ) and actin, as quantitatively major substrates. Methodology/Principal Findings We disrupted the genes encoding CCTα and CCTδ subunits in the ciliate Tetrahymena. Cells lacking the zygotic expression of either CCTα or CCTδ showed a loss of cell body microtubules, failed to assemble new cilia and died within 2 cell cycles. We also show that loss of CCT subunit activity leads to axoneme shortening and splaying of tips of axonemal microtubules. An epitope-tagged CCTα rescued the gene knockout phenotype and localized primarily to the tips of cilia. A mutation in CCTα, G346E, at a residue also present in the related protein implicated in the Bardet Biedel Syndrome, BBS6, also caused defects in cilia and impaired CCTα localization in cilia. Conclusions/Significance Our results demonstrate that the CCT subunits are essential and required for ciliary assembly and maintenance of axoneme structure, especially at the tips of cilia.


Introduction
Cilia are conserved organelles with important sensory and motile functions. Defects in cilia have been associated with a large number of human diseases, collectively known as ciliopathies. Cilia have a microtubule-based axoneme that is anchored to the basal body. The axoneme is typically composed of 9 doublet-microtubules arranged as a peripheral ring. Motile cilia usually have a pair of singlet microtubules at the center of the axoneme. The assembly and maintenance of cilia is dependent on bidirectional trafficking of protein complexes between the cell basal body and the cilia tip, the activity known as intraflagellar transport (IFT) [1]. Kinesin-2 motors move IFT cargo from the cell body to the tip of cilia, while recycled components are returned to the basal body by cytoplasmic dynein 1b motors [2,3,4,5].
The presence of different classes of molecular chaperones has been reported in cilia of diverse organisms. In Chlamydomonas, Hsp40 and Hsp70 were found in flagella [6,7], Hsp40 and the CCTa/TCP-1 subunit of the cytosolic chaperonin CCT were found in cilia of sea urchin embryos [8,9] and Hsp70 and Hsp90 were detected in cilia of Tetrahymena [10]. These chaperones could have a role in ensuring that the ciliary proteins preserve their native functional conformation during and after ciliogenesis, possibly by participating in the assembly or maintenance of large ciliary protein complexes. In fact, Hsp40 is a component of the radial spoke complex in sperm flagella of the ascidian Ciona intestinalis [11] and flagella of Chlamydomonas, where it may be involved in interactions between the radial spoke and central microtubules [12]. A mutation in BBS6, a protein related to CCTa, causes the Bardel Biedl Syndrome, a disease associated with defects in the function of cilia [13].
We have reported that in Tetrahymena, the expression of CCT chaperonin subunit genes is up-regulated during cilia regeneration following deciliation [14,15] and CCTa, d, e and g subunits localize to growing and full-length cilia of Tetrahymena [16]. CCT is a heterooligomeric complex formed by two rings connected back-to-back, each composed of eight distinct subunits (CCTa to CCTf) [17]. Each CCT subunit consists of three domains: an equatorial domain containing an ATP-binding site, an apical domain that interacts with the target protein, and the intermediate domain that connects the apical and equatorial domain. The apical domain contains a helical protrusion [18], which is involved in opening and closing the central cavity of the chaperonin. The full size CCT complex mediates the folding, driven by ATP binding and hydrolysis, of a wide range of newly synthesised proteins including tubulin (a, b and c) [19,20,21] and actin [22,23] as quantitatively major substrates.
In this study, we investigate the role of CCTa and CCTd subunits in Tetrahymena. We show that both CCTa and CCTd subunits are required for survival of Tetrahymena. Cells lacking expression of CCT subunits, before their death, show dramatic alterations in the microtubule cytoskeleton and cilia. An epitopetagged CCTa rescued the gene knockout phenotype and revealed that CCTa is a ciliary protein that is important for the maintenance of cilia tip integrity. We also show that a mutation of a conserved amino acid in CCTa that is also present in BBS6, a cilia-specific CCTa-related protein, affects the cytoskeleton and cilia. Collectively, our data show that CCT components are essential in a ciliated cell type, and that the referred CCT subunits play specific roles in ciliary assembly and maintenance.

CCTa and CCTd are essential in Tetrahymena
Tetrahymena thermophila cells, like most ciliates, have two nuclei, the germline, transcriptionally silent micronucleus (MIC) and the somatic, transcriptionally active macronucleus (MAC). Using DNA homologous recombination, we constructed heterokaryon strains with disruptions of either CCTa or CCTd genes only in the micronucleus using a neo2 gene cassette that confers resistance to paromomycin [24]. To study the consequences of gene disruptions, we allowed pairs of knockout heterokaryons to mate and produce progeny cells with new macronuclei developed from the zygotic micronuclei and expressing the gene knockout phenotype. While control wildtype strain matings produced viable conjugation progeny at the frequency of 95% (n = 200), no viable paromomycin-resistant progeny was recovered from matings of CCTa or CCTd knockout heterokaryons (n = 180 and 107 respectively). Inspection of drop cultures containing isolated pairs of mating CCT (a or d) heterokaryons revealed exconjugant cells that separated but failed to give rise to vigorous clones. These non-viable exconjugants were assumed to be progeny of mating heterokaryons that were expressing the CCT subunit knockout phenotypes. Typically these non-viable exconjugants presumably lacking a zygotic expression of either CCTa or CCTd died after ,50 hpm (hr post mixing of heterokaryons). Within this time, most of the CCTa and CCTd heterokaryon progeny failed to divide even once and about 20% completed a single cell division. The progeny that had divided often produced two daughter cells unequal in size (data not shown). While at 26 hpm, progeny cells of a control cross had a nuclear organization typical of a vegetative cell (1 MIC and 1 MAC) most of the CCT heterokaryon progeny had the pattern of DNA typical of an early exconjugant cell (two MACs and one or two MICs, Figure 1E and L compare with wildtype in D), consistent with an arrest in cell differentiation at an early post-conjugation stage and failure to enter a vegetative cell cycle.
Similar observations were made for CCT knockout heterokaryon progeny that were isolated into MEPP medium that supports growth of cells lacking either a functional oral apparatus [25] or cilia [26,27]. Thus the lethality of CCT heterokaryon progeny is not caused by loss-of-function of cilia or oral apparatus, both organelle types required for phagocytosis. All these observations indicate that both CCTa and CCTd genes are essential.

Cells lacking zygotic CCTap or CCTdp loose cytoplasmic and cortical microtubules and have structural defects in axonemes
Next we analyzed the morphology of the non-viable progeny of mating CCT heterokaryons before their death. These cells were designated as CCTa-KO and CCTd-KO. By immunofluorescence of the CCTa-KO and CCTd-KO cells with antibodies that recognize respectively CCTa and CCTd proteins, we observed a reduction of signal in the KO cells (Figure 2A-D). Typically CCTa-KO and CCTd-KO cells were smaller and more rounded as compared to wildtype ( Figure 1G, J, compare with A, C). Both the CCTa-KO and CCTd-KO cells showed progressive loss of microtubules in the cell body ( Figure 1E and J). At 26 hpm, in the CCTa-KO cells, the cortical longitudinal bundles (LM) and transverse microtubule bundles (TM) were less apparent based on immunofluorescence with an antibody against a-tubulin ( Figure 1E-H, compare with A to C). It appears that in CCTa-KO cells, LMs are thinner, and TMs are shorter, suggesting shortening or loss of individual microtubules within the cortical bundles ( Figure 1E-H, compare with A-C). At 36 hpm the LMs and TMs were no longer detectable in CCT-KO cells (data not shown). The intracytoplasmic microtubules were nearly completely absent at 26 hpm ( Figure 1L, compare to 1D). The CCTa-KO and CCTd-KO cells had fewer cilia, especially in the mid and posterior region of the cell ( Figure 1G and J). In a normal cell, new cilia are inserted primarily within the mid and posterior segment of the cell. Tetrahymena cells assemble new basal bodies in an asymmetric pattern, primarily within the central and posterior region of the cell. The fact that the density of cilia decreases in the central and posterior portion of the cell indicates that CCT KO cells are unable to assemble new cilia but are able, at least for sometime, to maintain pre-existing cilia (that were assembled before the KO induction). In the CCT KO cells, the basal body rows revealed by anti-centrin antibodies were often distorted and tended to be further apart ( Figure 1P-R compare with O). Gaps in the rows of basal bodies were apparent suggesting that the assembly of new basal bodies is also affected by CCT depletion ( Figure 1R).
Since one of the major substrates of the cytosolic chaperonin CCT is tubulin, we compared the phenotypes of CCTa-and CCTd-KO cells with the phenotype of cells entirely lacking zygotic expression of conventional aand b-tubulin (products of ATU1, BTU1 and BTU2 genes). To this end, we mated heterokaryons that carry in their MICs disruptions of all conventional a-tubulin and b-tubulin genes, namely ATU1, BTU1 and BTU2 (J.G., unpublished results). As expected, no viable paromomycin-resistant progeny was obtained from crosses of tubulin knockout heterokaryons (n = 120). Typically exconjugants separated but failed to establish viable clones and died before 48 hpm. At ,26 hpm, the tubulin-KO cells had a spherical shape and lacked most of LMs, TMs, and intracytoplasmic microtubules ( Figure 1S, T and V) and had a dramatically reduced number of cilia. Despite the rapid loss of microtubular structures, some tubulin-KO exconjugants had divided once, in most cases asymmetrically. The tubulin-KO cells had fewer cilia (consistent with a failure of assembly of new units) and among the remaining (pre-exisiting) cilia 67% (n = 426) had splayed tips ( Figure 1U and inset in V). The length of pre-existing cilia was slightly reduced at 26 hpm (5.8760.54 mm (n = 102 cilia), with similar values at 36 hpm (5.6560.47 mm (n = 112), (compare to wildtype cilia length, 6.7760.59 mm (n = 175)). The differences are statistically significant (t-test; p,0.001) ( Figure S1A). Thus, the consequences of loss of CCT subunits and loss of tubulin are similar except that the length of cilia is slightly more affected in the CCTd-KO cells. These data argue that to a large extent, the consequences of loss of CCT activity could be mediated by lack of proper folding of ciliadestined tubulin by CCT.

CCTa and CCTd depleted cells are unable to reciliate
The capacity of the CCTa and CCTd-KO cells to reciliate after deciliation was investigated using a deciliation protocol adapted to a small number of cells (see Materials and Methods S1 and References S1). We have used cells at ,20 h of KO. The same procedure was performed in WT cells as control. Contrarily to WT cells, the CCTa and CCTd-KO cells after 20 h of KO induction are unable to recover their cilia. Very few KO cells were able to reciliate, and in such cases, there was only a partial reciliation, with a random distribution of the new cilia ( Figure S2) (data not shown for CCTa). This observation confirms the CCTa and CCTd are required for assembly of new cilia.

HA epitope-tagged CCTap rescues CCTa-KO cells and localizes to cilia
To address the specificity of the observed CCT gene knockout phenotypes, we tested whether the progeny of mating CCT knockout heterokaryons cells could be rescued by reintroduction of a wildtype CCT gene fragment encompassing the disrupted region. To this end, we mated pairs of CCT heterokaryons, subjected them to biolistic bombardment using a corresponding CCT gene fragment that was designed to replace the disrupted CCT gene sequence by DNA homologous recombination (as described in Material and Methods), and selected progeny cells with paromomycin, to which resistance was conferred by the neo2 cassette. In principle, we attempted to select surviving progeny that had replaced some of the disrupted copies of a CCT gene with wildtype copies in the new MAC. After biolistic transformation, for both CCTa and CCTd mating heterokaryons, ,97% of the wells (n = 480 corresponding to 10 7 mating cells) contained drugresistant growing cells while no such wells appeared in the same number of selected mock-transformed CCT mating heterokaryons. The presence of the CCT transgenes in the rescued cells was confirmed by PCR ( Figure S3). Thus, we confirm that the lethality of CCT gene knockout mating heterokaryons is caused by disruption of CCT loci.
To test whether the lethality in progeny of CCT knockout heterokaryons is caused by a loss of the CCT subunit protein, and not solely by gene targeting, we attempted to rescue the mating CCTa heterokaryons by biolistic bombardment with a fragment that was designed to insert a gene encoding an HA-tagged CCTa under the control of the cadmium-dependent MTT1 promoter into the non-essential BTU1 locus [28]. Rescues were observed at the frequency of ,92% of the wells (n = 480 corresponding to 10 7 mating cells). The genomic DNA extracted from CCTa-HA rescued cells was found to contain the transgene fragment ( Figure  S4A and B). Using antibodies against the HA and CCTa -subunit we also confirmed by western blot that the rescued cells expressed CCTa p-HA protein ( Figure 2E). As expected for a MTT1-driven transgene, the levels of CCTa-HA protein were increased with either the higher dose or longer exposure to cadmium ( Figure 2E). Thus, we have successfully expressed a MTT1-driven copy of CCTa gene in cells that lack the endogenous CCTa gene. Interestingly, polyclonal antibodies that were generated against a CCTa peptide, reacted weakly with the (more slowly migrating) transgene protein in rescued cells as compared to wildtype protein in control cells ( Figure 2E). Since the antibodies were generated against the last 12 amino acids of CCTa [29], addition of HA to the C-terminus could have a steric effect on the epitopes of the polyclonal antibodies. In absence of exogenous cadmium, CCTap-HA was localized primarily to the cell body and was not detected in cilia ( Figure 2H compare with negative control in 2G). When cadmium chloride (2.5 mg/ml) was added to the medium for 76 h, a stronger signal of CCTap-HA was detected and the protein was prominently present in cilia and accumulated at the ciliary tips ( Figure 2I and J). Next, we investigated the consequences of lowering the levels of CCTap-HA expression, by growing rescued cells without exogenous cadmium (in an SPP medium from which the residual cadmium ions were removed by exposure to chelex-100 resin referred as SPPCT, see Material and Methods). Wildtype cells had similar growth rates in SPPCT supplement with exogenous cadmium to the growth shown in SPPCT without addition of cadmium. On the other hand, the rescued CCTa-HA cells had a growth rate slightly lower when grown in SPPCT without cadmium than in SPPCT complemented with cadmium (data not shown). It is worth to mention that while lack of exogenous cadmium has resulted in a dramatic decrease in the levels of CCTap-HA, small amount of the protein was still present, likely because the MTT1 promoter has a noninduced basal level of expression, mimicking a knockdown of CCTa ( Figure 2F). Strikingly, cilia were shorter in cells with reduced levels of CCTap-HA (grown without cadmium), than in wildtype cells grown under the same conditions ( Figure S1B). Furthermore, based on immunofluorescence with an antibody against tubulin, these cells have an increased number of cilia with splayed tips or abnormal spotted tubulin staining pattern at the tips ( Figure 2L and M, compare with O). Also, these phenotypes were not observed in wildtype cells growing in SPPCT (data not shown).
To conclude, we observed that CCTap-HA, when moderately overexpressed localizes to cilia and is enriched at the tips. These data are consistent with our previous observations [16] that CCTa is a ciliary protein. This localization was also confirmed by isolation and fractionation analysis of cilia obtained from wildtype  Figure S5A). To assess the effectiveness of fractionation, we reprobed the same blot with anti a-tubulin antibody. As expected, tubulin, the major protein of the axonemes was weakly detected in the membrane fraction ( Figure S5A). The specificity of the antibody was tested by peptide pre-absorption to the antibody as shown in Figure S5B. Moreover, our results show that the depletion of CCTa affects the structure of axoneme tips.
The G346E mutation in Tetrahymena CCTap leads to a temperature-sensitive growth and affects the function of oral cilia We took advantage of the availability of CCTa knockout heterokaryons to introduce a mutation into CCTap that could affect cilia. Kim and colleagues [13] showed that BBS6, a protein whose mutation causes a ciliopathy, the Bardet-Biedl Syndrome, has amino acid sequence homology with CCTa. The genome of Tetrahymena and many other non-vertebrate eukaryotes lacks an obvious BBS6 sequence. These observations suggest that BBS6 is a vertebrate-specific variant of CCTa that has evolved cilia-specific functions. Consequently, organisms like Tetrahymena that lack BBS6, could be using CCTa for ciliary functions, as is supported by our data so far. To identify amino acids of CCTa that could be important in the context of cilia, we produced a multiple sequence alignment of BBS6 and CCTa proteins from a few ciliated organisms. Overall, the BBS6 and CCTa sequences are 19% identical (30% of similar) ( Figure S6A). We examined amino acids that represent the apical domain of CCTa and could contribute to the substrate-binding site [30]. Within this domain, one amino acid is conserved between BBS6 and CCTa of diverse organisms: glycine 346 from T. thermophila CCTa. Importantly, in humans, a mutation at the corresponding position, G345E, causes Bardet-Biedl Syndrome [13]. Since the mutation occurs in the CCTa apical domain ( Figure S6B) we hypothesised that the mutation G346E could interfere in the interaction between CCTa and folded substrates relevant to cilia assembly/maintenance. To investigate the impact of the mutation in the native functional structure of the CCTa protein, we predicted with the ProModII program [31] and compared the secondary structure of the wildtype and mutated G346E CCTa apical domain ( Figure S6C). The model of the partial CCTa structure obtained was based on the crystal structure of the subunit a of the chaperonin thermosome from Thermoplasma acidophilum [32]. We observed that the replacement of the glycine for a glutamate has led to disruption/disappearance of several asheets (see arrow in Figure S6C) which might interfere with the flexibility of this domain that is required for folding. Indeed, the apical domain contains a helical protrusion [18], which is involved in opening and closing the central cavity of the chaperonin. The remnant of the secondary structure of this apical domain did not suffer any change with the referred substitution.
We used a fragment encoding a CCTa with the ciliopathybased mutation, G346E, in an attempt to rescue mating CCTa heterokaryons. Besides the single mutation, the fragment encoded an otherwise wildtype sequence and was intended to replace the disrupted sequence at the native locus. Rescued cells were isolated, indicating that G346E is not a lethal mutation. The genomic DNA of these transformants was analyzed by PCR and sequenced, revealing the presence of two products corresponding to the neodisrupted and the introduced G346E encoding CCTa allele ( Figure S4C).
The G346E mutant cells grew extremely slowly on the regular SPP medium. Furthermore, the CCTa-G346E cells were temperature-sensitive, growing more slowly at 30uC as compared to 16uC. At 16uC, the G346E population contained mostly normal-looking cells in respect to size and shape, but some of these cells displayed erratic movement patterns including prolonged periods of spinning around the antero-posterior axis (results not shown). At 30uC, 50% of G436E cells had a normal shape (average dimensions 25647 mm, n = 488), 27% were extremely elongated (average dimensions 69650 mm, n = 260), 11% had a drop-like shape (n = 105) and 13% were very large so called monster cells (90660 mm to 50645 mm, n = 127).
We noticed that these cells grew better in MEPP media that stimulates the uptake of nutrients by pathways that do not require phagocytosis in the oral apparatus [25]. We tested their capacity of performing phagocytosis vacuoles adding Indian ink to the medium and quantified the cells that presented food vacuoles containing ink ( Figure 3A-D). Ninety four percent of G436E cells (n = 1448) were unable to uptake ink. Thus, oral cilia may not be fully functional in the G436E strain. Noticeable, the 6% of the mutant cells that were able to ingest ink, and that were designated by ''normal looking cells'' started to prevail in the culture when mutant cells grew at 30uC for long periods (several weeks). This led us to investigate if the introduced mutation in the CCTa coding region gene was still present in these cells. By sequencing analysis we confirmed that the mutation G346E continued to be present in the CCTa gene sequence of these cells, being the only allelic form of CCTa found. Indeed, the mutation G346E was inserted in the region of CCTa coding region that have been removed when constructing the KO heterokaryons strains. Therefore, since these were the cells that were rescued by the introduction of the G346E mutated gene, the observed recovered phenotype could never be a consequence of a recombination event between the wild type CCTa gene and the mutated one. Most probably, these cells constitute a suppressor strain where a second mutation occurred restoring the original phenotype, by reverting the effect of CCTaG346E mutation, and are under natural selection when growth occur over long periods.
The microtubule cytoskeleton of CCTa-mutG346E-carrying cells was analyzed by immunofluorescence using an antibody against a-tubulin ( Figure 4). Mutant cells frequently contained multiple sets of nuclei and multiple cortical domains, e.g oral apparatus, and constitute the typically designated monster cells ( Figure 4B and C compare with wildtype in A), consistent with failures to undergo cytokinesis. However the normal-looking cells were able to divide completely (data not shown). The evident defects in completing cytokinesis and their multiple attempts to divide led mutant cells to exhibit dramatic alterations in the organization of ciliary rows as confirmed using an antibody against centrin ( Figure 4F and G, compare with D and E). In contrast to wildtype cilia ( Figure 4H) that have a spear-like staining of tubulin at the tip, the mutant cells showed an abnormal staining at the most distal part revealed as a strong spotted staining of tubulin ( Figure 4I to K). Since low levels of CCTa lead to abnormalities of cilia tips (Figure 2L and M) and mutant cells also have abnormal tips we decided to investigate if the mutated form of CCTa was targeted to cilia. We observed that the antibody against CCTa did not give any ciliary staining in the mutant cells, even when the body of the cell is clearly labeled (Figure 3 F-H compare with E).
Taken together these data clearly show that the CCTa mutation G346E affects CCTa cilia localization which in turns affects cilia tips. The observed effect of the CCTa mutation G346E supports the previous evidences shown in this paper that CCTa is required for cilia structure maintenance, particularly at the tip level.

Discussion
We have investigated the function of CCTa and CCTd subunits of the eukaryotic cytosolic chaperonin in T. thermophila. To our knowledge, this is the first functional study of CCT subunits in a ciliated model. We show that the CCTa and CCTd genes are essential in T. thermophila, as shown earlier in yeast [33,34]. The essential function of CCT is not unexpected since the CCT complex participates in the folding of essential cytoskeletal proteins (actin and tubulin). Also CCT may mediate the folding of 1000-2000 other proteins that play diverse and critical functions in the cell, as, for example, cell cycle progression, chromatin remodeling, assembly of nuclear pore complex and protein degradation [35,36].
The phenotypes of cells lacking either CCT subunits or tubulin are quite similar: these cells fail to grow within 1-2 generations, loose cytoplasmic and cortical microtubules, fail to assemble new cilia and have defects at the tips of microtubules in pre-existing axonemes. Thus, it is possible that to a large extent, the lethality induced by CCT subunit loss-of-function is caused by failure to fold tubulin. Consistently, in mammalian cultured cells, reduction of CCT levels by 90% (due to siRNA-mediated knockdown) strongly reduced the levels of total and newly synthesized aand btubulin [37]. The observed splayed tips of axonemal microtubules could be explained by increased curvature of protofilaments that depolymerize [38]. It is likely that tips of axonemes are unstable due to lack of addition of new tubulin subunits. The simplest explanation of our observations is that proteins destined to cilia, including tubulin, requires folding by CCT. Despite the fact that the phenotypes of CCT subunit loss-offunction can be explained by the resulting failure in tubulin folding in the cell body, published work and some data presented here continue to support a role for CCT subunits inside cilia. Thus, expression of CCT subunit genes is increased during cilia regeneration in Tetrahymena [14] and Chlamydomonas [39]. While this result alone could be explained by a cell body-restricted activity whose levels increase during ciliation, localization and proteomic studies have detected some CCT proteins in cilia and centrioles/basal bodies in Tetrahymena, sea urchin embryos and Chlamydomonas [9,16,40]. Here we present complementary data showing that CCTa is present in both membrane/matrix and axonemal fractions of cilia (see Figure S5), suggesting that the protein is interacting with the axonemal microtubules while circulating in the ciliary compartment. Also the CCT-depleted cells show a reduced level of this protein in cilia (see Figure 2). Moreover, we show that the epitope-tagged CCTa which rescues the gene knockout lethal phenotype localizes to cilia. Thus, either the folding activity of CCT chaperonin also occurs inside cilia, or the CCT subunits found in the ciliary compartment have other functions. Interestingly, other chaperones not required for tubulin folding have been found inside cilia. For example, Hsp70 and Hsp90 were found in Tetrahymena cilia [10] and Hsp70 was detected in Chlamydomonas flagella and ciliated cells of sea urchin embryos [7,41]. Hsp70 was identified as one of the components of a 17S complex p28-containing inner dynein arms in Chlamydomonas [42]. Noteworthy, both Hsp70 in Chlamydomonas [7], and the epitope-tagged CCTa in Tetrahymena (this study) preferentially localize to the tips of assembled cilia. Since ciliary proteins are subjected to significant mechanical stress, their function may require a relatively high level of turnover to replace damaged proteins. Inside cilia, molecular chaperones could be involved in quality control and turnover of ciliary proteins. In agreement with this model, in Chlamydomonas, Hsp70 and Hsp40 affect flagellar movement possibly by maintaining/transforming protein conformations [12,43]. CCT could be required for the maintenance of axonemal proteins subunits such as tubulin and ciliary actin [44,45], or alternatively, for their assembly and/or turnover. This hypothesis is supported by our observations that show both the localization of overexpressed CCT at the tips of cilia, as well as defects of ciliary tips in cells depleted in CCT activity. Importantly, in Xenopus multi-ciliated cells, CCTa and CCTe were localized in punctuate structures along the ciliary axonemes, and their mislocalization induced by the depletion of an antagonist of Wnt pathway (Fritz) has been correlated with fewer and shorter cilia phenotype [46]. Moreover, the fact that CCTd-KO cells were unable to reciliate indicates CCT activity is important for new assembly of cilia, and this role may not be simply the cytosolic supplier of tubulin. Interestingly, Tetrahymena cells grown in an enriched medium and treated with cycloheximide can partially regrow cilia after deciliation suggesting the presence of a pool of stored tubulin [47] that cells could use for assembly of new cilia.
It is known the distal ends of axonemal microtubules are covered by caps, structures that connect axonemal microtubules to the cilia membrane [48,49]. These structures were suggested to be involved in the assembly and maintenance of cilia, possibly regulating the assembly and disassembly of axonemal microtubules [50,51,52]. We can speculate that CCT subunits are associated with either the distal ends of axonemal microtubules or with caps. It is known that the CCT subunits c, a, f and d bind to in vitro assembled microtubules, and thus behave like microtubuleassociated proteins (MAPs) [53]. Interestingly, CCT subunits bind to F-actin and reduce the filament elongation rate at the plus end in erythrocyte membrane cytoskeletons [54]. It is conceivable that through the ability to behave as end-binding MAPs, CCT subunits affect the assembly and turnover of tubulin on axonemal microtubules known to occur preferentially at the distal end of axonemes [55]. Additionally, CCTs may be involved in interactions between microtubules and the cilia membrane at ciliary tips. There is some evidence that CCT subunits interact with membranes. The adrenal medullary form of CCT (chromobindin A) efficiently binds to chromaffin granule membranes [56]. In human erythrocytes, CCTa is translocated to the plasma membrane following a heat-shock, interacting with the specialized membrane skeleton [57].
We show that CCTa-G346E mutation impairs CCTap localization at cilia tips and those cilia present an abnormal pattern of staining with anti a-tubulin. These observations support a model that this CCT subunit has a direct ciliary role. As the evolutionary related BBS6 [13], CCTa may have a role in assembly of some complexes at cilia tips. Nachury and collaborators [58] have shown that the BBSome, an oligomeric complex of BBS (BBS1-2, BBS4-5, BBS7-9) proteins, was directly implicated in ciliogenesis by promoting vesicle trafficking to the cilia membrane. Very recently, it was shown that BBS6 forms with the other chaperonin-like BBS10 and BBS12 proteins (vertebratespecific BBS genes), a complex with CCT proteins (CCT1-5 and CCT8) that is required for BBSome assembly [59]. Similarly to Tetrahymena CCTa depleted cells and CCTa-G346E mutant where oral and somatic cilia presented functional failures, the respiratory tract cilia of BBS6-/-mice showed structural abnormalities accompanied by functional defects affecting cilia tips and reduction of ciliary beat frequency [60]. Therefore, is tempting to suggest that in Tetrahymena CCT chaperonin does not require BBS6 to interact with BBSome subunits since CCTa evolutionary could be seen as its representative/substitute.
In conclusion, the construction of Tetrahymena CCTa-and CCTd-KO strains has helped to define the role of CCT subunits in a ciliated organism. We show in this study that CCT subunits are needed for assembly of cilia and maintenance of axoneme structure, especially at the tips of cilia.

Cells and culture conditions
Strains used in this study are listed in Table S1. Tetrahymena cultures were grown in SPP [61] supplemented with an antibiotic/ antimycotic mixture at 100 U/ml penicillin, 100 mg/ml streptomycin, and 0.25 mg/ml amphotericin B. In some experiments we used the MEPP medium on which Tetrahymena cells grow in the absence of phagocytosis [25]. The SPPCT (metal-depleted medium; D. Dave and J.G, unpublished) was used in some experiments. SPPCT was prepared by depleting the SPP medium from ions with 5% of Chelex-100 beads (BioRad) followed by complementation with trace metals (100-fold concen- To assay phagocytosis, India ink was added at a final concentration of 1%, cells were incubated for 30 min and were scored for the presence of food vacuoles (filled with black ink).

Germline disruption of CCTa or CCTd genes
To disrupt either CCTa or CCTd genes in the MIC, we introduced neo2 cassette-interrupted targeting fragments into early mating cells using a biolistic gun and produced heterozygous transformants as described [62,63]. To prepare a targeting plasmid for disruption of the CCTa gene, a ,1.6-kb genomic fragment that included ,400-bp of the 59UTR plus the first 1.2kb of the coding region of CCTa (including the translation initiation codon), was amplified with primers Alf-5F and Alf-5R containing restriction sites for SacII and BamHI respectively (primers Alf-5F:59-TCCCCGCGGATGAATGAAAGAGTGA-GATG-39 and Alf-5R:59-CGCGGATCCTTCAACAGCATCA-ACAACGA-39). This fragment was cloned into the plasmid p4T2-1 [24], a neo2 cassette plasmid. The resulting plasmid was digested with ClaI and XhoI and used to insert a ,1.2-kb of 39 UTR of CCTa, with the last 813-bp of genomic fragment of CCTa, including the codon stop. This fragment was amplified with the primers Alf-3F and Alf-3R containing the restriction site of ClaI and XhoI respectively in their flanking regions, (primers Alf-3F:59-CCATCGATGAATGTGCTGAAGTTTACGA-39 and Alf-3R: 59-CGGCTCGAGCCCATTCTACATCTTATCC -39), to create the plasmid pNeo2CCTa.
To prepare a plasmid for the disruption of the CCTd gene, a 262-bp of 59UTR, with the initial ,1.6-kb genomic fragment of CCTd including the first codon, was amplified with addition of SacII and BamHI sites in the primers respectively (primers Delt-5F:59-TCCCCGCGGTATGAATTGTTTTGAAGTGT-39 and Delt-5R: 59-CGCGGATCCTCAAT-CAATTCAGTGTCTTC-39). This fragment was cloned into p4T2-1 using SacII and BamHI sites. The resulting plasmid was digested with ClaI and XhoI and used to insert a ,1.5-kb of 39 UTR of CCTd, with the last 364-bp of the genomic fragment of CCTd, including the stop codon. This fragment was amplified with the primers containing the restriction site of ClaI and XhoI respectively in their flanking regions, (primers Delt-3F: 59-CCATCGATGACTAG-AGAAATGAAGGGTGTT-39 and Delt-3R: 59-CGGCTCGAGTAAGAAGACTGTTGA-TACCG-39), to create pNeo2CCTd.
For germline targeting, each disruption plasmid (pNeo2CCTa and pNeo2CCTd) was digested with SacII and XhoI and used to transform mating CU428.1 and B2086.1 strains by biolistic bombardment. For each transformation, approximately 10 mg of DNA was used to coat gold bombardment particles of 0.6 mm in size (Bio-Rad). Gene replacements mediated by these targeting fragments were designed to remove ,800-bp of regions encoding highly conserved domains of the CCTa and CCTd proteins. Heterokaryons were generated by bringing the micronucleus to homozygosity using a star cross while allowing the disrupted alleles to assort from the macronucleus [62].

Rescues of mating CCT knockout heterokaryons with tagged and mutated CCT-encoding transgenes
To test whether the lethality associated with disruption of the CCTa and CCTd is caused by loss-of-function of these genes, we attempted to rescue mating knockout heterokaryon cells with corresponding fragments of DNA containing the coding sequence of CCTa and CCTd genes, respectively. The genomic fragment of CCTa gene was obtained by PCR with the primers Alf-5F: 59-TCCCCGCGGATGAATGAAAGAGTGAGATG-39 and Alf-3R: 59-CGGCTCGAGCCCATTCTACATCTTATCC -39, and cloned into T-Vector (Promega). In the case of the CCTd gene, the fragment to clone was amplified with the primers Delt-5F (59-TCCCCGCGGTATGAATTGTTTTGAAGTGT-39) and Delt-3R (59-CGGCTCGAGTAAG-AAGACTGTTGATACCG -39) and digested after with SacI and XhoI enzymes for biolistic transformation.
To create the CCTa G346E mutant strain we performed a somatic rescue transformation of CCTa-KO cells with mutated CCTa gene fragment obtained by site-directed mutagenesis [64] with an oligonucleotide: 59-GAAGCTTCCTATCTAGAA-GAAT-GTGCTGAAGTT-39. In all the cases the biolistic transformation and selection of cells were performed as already described [65,66]. The presence of the desired mutation in the CCTa gene of the transformed and rescued CCTa-KO cells was confirmed by PCR, using standard conditions, and analysis of the pattern obtained by restriction enzyme hydrolysis of the PCR products. It was also confirmed by sequencing the entire CCTa gene that no other modification was present.
To express CCTa-HA protein at levels comparable to physiological conditions, we rescued mating CCTa heterokaryon progeny, by introducing a fragment of DNA containing the coding sequence of CCTa-HA, without applying any selective pressure to increase the transgene copy number [66]. The transforming DNA was inserted by homologous recombination in an ectopic locus, the b-tubulin locus BTU1, and its expression was under the promoter MTT1 (metallothionein 1 protein), dependent of cadmium chloride. The CCTa KO heterokaryons strains (CCTA-A1 and CCTA-B5) were allowed to complete conjugation that takes approximately 14 h. Then, 24 h after mixing the heterokaryons (hpm, hours post mixing), the cells were transformed biolistically with the BTU1-MTT1-CCTa-HA-BTU1 cloned fragment. Transformants that integrated the transgene into the BTU1 locus were selected with paromomycin (90 mg/ml) and cadmium chloride (1.5 mg/ml or 2.5 mg/ml).

Indirect immunofluorescence microscopy
For staining KO cells, ,50-100 cells were isolated into 10 ml of 10 mM Tris, pH 7.5, on a coverslip previously coated with poly-L-lysine (Sigma). These cells were generally isolated after 18 hpm that is ,4 h after end of conjugation and consequently should be ,4 h of KO. Coverslips were processed for immunofluorescence labeling as described in Thazhath and co-workers [67]. TO-PRO-3 (Molecular Probes) was used (1:1000) to stain DNA during 90 min, at room temperature. The following primary antibodies were used: mouse 20H5 anti-centrin (1:100, gift of Dr. Salisbury, Mayo Clinic, Rochester, MN), mouse 12G10 anti a-tubulin (1:10, from University of Iowa, Developmental Studies Hybridoma Bank), rat purified (by affinity column) anti-CCTa (1:10) (this work), crude rat serum anti-CCTa (1:50) [29] and crude rat serum anti-CCTd (1:30) [16]. Secondary antibodies were goat anti-mouse Alexa 488 (Molecular Probes) (1:500), goat anti-rat-FITC and goat anti-mouse-TRIC (Sigma) conjugates, both used at dilution of 1:600. For immunolocalization of CCTa-HA protein in KO rescued cells, they were grown in falcon tube overnight without any drug except cadmium chloride, when added, washed, fixed and processed for immunofluorescence as the other slides. The primary antibody used was the mouse monoclonal anti-HA (Sigma) and the secondary antibody was goat anti-mouse Alexa 488 (Molecular Probes), in a dilution of 1:500.
Cells were viewed using a LeicaH TCS SP2 spectral confocal microscope (using 63x oil immersion with 1.40 NA). Images were assembled using Image NIH Image J. and Adobe Photoshop 6.0H software. The length of axonemes either on cells or isolated was determined on Z-project of confocal sections using NIH Image J.

Protein electrophoresis and western blotting
To analyze the expression of the tagged CCTa-HA protein in the rescued cells, total protein extracts from 25000 cells were prepared, as well for wildtype cells, and used per lane. Briefly, cells were pelleted by centrifugation at 16006g for 3 min, suspended in 1 ml of 10 mM Tris-HCl, pH 7.5 and further concentrated into a dense pellet by centrifugation at 16006g for 3 minutes. Cell pellets were resuspended in 10 ml of 10 mM Tris-HCl, pH 7.5 and lysed with same volume of lysing buffer (62.5 mM Tris, pH 6.8, 2% SDS, 10% glycerol, 0.0005% Bromophenol Blue, 5% b-mercaptoethanol, final concentrations). Protease inhibitors were added at a final concentration of 0.5 mg/ml leupeptin, 10 mg/ml chymostatin, 10 mg/ml trans-epoxysuccinyl-L-leucylamido-(4guanidino) butane (E-64), and 15 mg/ml antipain. The mixture was boiled for 3 min at 95uC.

Sequence analysis
For the multiple sequence alignment of CCTa and BBS6 protein sequences, the sequences obtained from NCBI and TIGR database and listed on Table S2 were used. The multiple sequence alignment was produced using the T-Coffee method [68] more appropriate for alignment of proteins with low percentage of identity as BBS6 and CCTa. The alignment was edited with GeneDoc program.
Prediction of the secondary structure of CCTa apical domain in wildtype and in the G346E mutant cells was done using the program ProModII [31], since the inserted mutation was in this protein domain. The model of the partial CCTa structure obtained was based on the coordinates of the subunit a of the chaperonin thermosome from T. acidphilum (Pubmed accession numbers: 1a6e, 1a6d and 1q2v). The visualization of the predicted structure is made by Rasmol program.

Statistical analyses
The experiments were performed at least three times and the results were expressed as means 6 S.D. Differences between the data were tested for statistical significance by t-test. P values less than 0.05 or 0.001 were considered statistically significant.  Figure S3 PCR analysis of strains obtained in rescue experiments to confirm their genotype. A) Analysis by PCR of the CCTa locus in wildtype cells and rescued CCTa-KO cells. For WT strain it was observed only one band (white arrowhead) corresponding to WT allele, whereas in rescued CCTa KO strain (RA+) an additional band (asterisk) corresponding to the disrupted-allele CCTa is visible. To facilitate the interpretation of the bands pattern a heterozygous strain for CCTa disruption was obtained by a cross of one of the CCTa-neo-disrupted heterokaryon strains with a WT strain. In the heterozygous (HZ) two bands were found, one corresponding to the WT allele (2.8 kb) and the other corresponding to the disrupted-allele of CCTa, with the expected size (3.5 kb). B) Analysis by PCR of the CCTd locus in WT cells and in the rescued CCTd-KO cells. Also, PCR analysis revealed two bands in rescued CCTd-KO strain (RD+) confirming the presence of the WT and the disrupted allele. Found at: doi:10.1371/journal.pone.0010704.s003 (0.09 MB TIF) Figure S4 Genotypic analysis of the CCTa-KO cells rescued with a HA tagged CCTa cDNA or genomic CCTamutG346E. A) PCR analysis using genomic DNA from the rescued CCTa-HA strain (RAHA) with the: 1. pair of primers that amplify full cDNA CCTa; 2. Primer-F for initiation codon of CCTa gene and primer-R for 39end of HA sequence; 3. Primer-F for initiation codon of CCTa gene and primer-R for a sequence of BTU1 gene where the fragment was intended to recombine. B) PCR analysis of full coding sequence (using AlfF and AlfR primers that anneal respectively at initiation and termination codons) of CCTa showing the presence of cDNA CCTa (1.6-kb) and a CCTa fragment with size ,3.5-kb corresponding to the neo-disrupted-CCT allele present in the native locus of the rescued CCTa-HA strain. A heterozygous strain (HZ), containing the genomic wildtype (WT) CCTa allele (2.8-kb) and the disrupted allele (3.5-kb), was used to compare PCR band pattern. WT strain and plasmid DNA containing the cDNA of CCTa (C+) were also used as controls. C) PCR analysis of the macronuclear genotype of transformed CCTa-mutG346E strain. PCR products obtained using AlfF and AlfR primers that anneal respectively at initiation and termination codons in WT cells, HZ cells (that have in their macronuclear genotype the wildtype and neo-disrupted CCTa alleles) and the CCTa-mutG346E strain. Found at: doi:10.1371/journal.pone.0010704.s004 (0.09 MB TIF) Figure S5 CCTa is a ciliary protein found in both axonemal and membrane/matrix fraction of cilia. A) Cilia from wildtype cells were isolated and fractionated in axonemal (Ax) and membranar (Mb) fraction which contains the soluble ciliary matrix. Western blot analysis using a serum against CCTa was performed showing the presence of the protein in both ciliary fractions and in total cilia extract. Western blot using anti atubulin supports the effectiveness of the cilia fractionation. B) The specificity of the antibody used above was confirmed by preabsorption of the antibody with the peptide used to elicit it. Western blot analysis of total protein extracts of wildtype cells and purified cilia extracts revealed only one specific band for CCTa that is not detected when antibody is pre-absorb to the peptide. Found at: doi:10.1371/journal.pone.0010704.s005 (0.19 MB TIF) Figure S6 The apical domain of CCTa is related to a domain in BBS6 protein and contains a highly conserved G346 residue. A) Multiple sequence alignment of BBS6 and CCTa protein sequences using T-Coffee method. The multiple sequence alignment was produced with ClustalW2 program. The sequences were obtained from NCBI databases (see table S2). The alignment was edited with GeneDoc program and the aminoacid conserved percentage is indicated using the following shade style identity: red 100%; green 80% blue 60%. The position of the mutated G346 amino acid in this study is indicated by a black arrow. B) Schematic representation of CCTa protein showing its different domains, along with the position of the mutation made in the protein. C) Rasmol representation of the secondary structure of CCTa apical domain in wildtype and mutant cells (mutation G346E) using a ribbon model. The aminoacid residue that was mutagenized is depicted as white space-filling form (indicated with a red arrowhead). Note the mutation has led to the disappearance of b-sheets present in the ribbon model of wildtype cells (white arrow).