α- and β-tubulin C-terminal tails with distinct modifications are crucial for ciliary motility and assembly Free

A tubulin molecule consists of an ∼400-amino-acid core and an ∼20-amino-acid unstructured C-terminal tail (CTT) that is abundant in glutamate residues (Nogales et al., 1998). In microtubules, the C-terminal region protrudes from protofilaments and affects various aspects of microtubule function: tubulin polymerization (Serrano et al., 1984), microtubule severing (Roll-Mecak and Vale, 2008), kinetochore dynamics (Miller et al., 2008), and interactions with motor proteins (Wang and Sheetz, 2000; Sirajuddin et al., 2014) and microtubule-associated proteins (Hinrichs et al., 2012). However, the function of tubulin CTTs in cilia and flagella (interchangeable terms), abnormalities of which cause various human diseases collectively called ‘ciliopathies’, is still obscure.

Several kinds of post-translational modifications (PTMs), including polyglutamylation and polyglycylation, occur in the tubulin CTT. Polyglutamylation, a modification on the γ-carboxyl group of glutamate residues, forms side chains of up to 20 glutamate units (Eddé et al., 1990; Rüdiger et al., 1992; Audebert et al., 1994; Redeker et al., 2005). We and others have previously shown that this modification regulates ciliary motility (Kubo et al., 2010; Suryavanshi et al., 2010; Ikegami et al., 2010) by affecting an axonemal structure called the nexin–dynein regulatory complex (Kubo et al., 2012; Kubo and Oda, 2017). Another kind of tubulin modification, polyglycylation, competes with polyglutamylation for the same glutamate residues and generates side chains of up to 40 glycyl units (Redeker et al., 1992, 2005; Wall et al., 2016). Polyglycylation was shown to be involved in the stability and maintenance of ependymal cilia (Bosch Grau et al., 2013) and primary cilia (Rocha et al., 2014). Recently, this modification was also found to affect the motility of mouse sperm by modulating the power generation of axonemal dyneins (Gadadhar et al., 2021). In mammals, seven out of the 13 members of the tubulin tyrosine ligase-like (TTLL) protein family possess glutamylation activity (Janke et al., 2005; Van Dijk et al., 2007) and three other members possess glycylation activity (Wloga et al., 2009; Ikegami and Setou, 2009). In cilia, the levels of glutamylation and glycylation affect each other; the lack of glutamylation causes the increase in glycylation and vice versa (Redeker et al., 2005). However, the physiological significance of this anticorrelation between the two PTMs for ciliary functions is still unclear.

We used Chlamydomonas reinhardtii, which possesses only two α-tubulin genes (TUA1 and TUA2) and two β-tubulin genes (TUB1 and TUB2), each pair encoding an identical protein (Silflow et al., 1985) (Figs 1A, 5A). Such a small number of tubulin genes and species variation is a feature not seen in mammalian cells, which normally have multiple genes for both α- and β-tubulins. The simple gene organization of tubulins in Chlamydomonas might enable complete replacements of the wild-type tubulin with mutated versions by gene editing. In this study, using CRISPR/Cas9, we generated strains harboring mutations in the CTT of either α- or β-tubulin. For α-tubulin, we generated mutant strains in which the four glutamate residues in the CTT were replaced by alanine. For β-tubulin, we produced mutant strains lacking variable lengths of the CTT region where multiple glutamate residues are clustered. The α-tubulin mutant strain, named TUA1(4A), without four glutamate residues (E445, E447, E449 and E450) was found to exhibit nearly complete absence of polyglutamylation and ciliary motility. In contrast, the β-tubulin CTT mutant strain TUB2(Δ6E) without six glutamates (E435, E437, E439, E440, E441 and E442) was found to lack glycylation and assemble only short cilia without the central-pair microtubules. Our study thus reveals distinct and important roles of α- and β-tubulin CTTs for ciliary function.

We set out to produce an α-tubulin CTT mutant lacking the four glutamate residues (E445, E447, E449 and E450; Fig. 1B) predicted as glutamylation and glycylation sites (Redeker et al., 2005). Because either one of the two α-tubulin genes of Chlamydomonas, TUA1 (Cre03.g190950) and TUA2 (Cre04.g216850) (Fig. 1A), can complement the lack of the other (Fromherz et al., 2004; Kato-Minoura et al., 2020), we used a strain lacking the TUA2 gene and replaced all the four glutamates with alanines in the TUA1 protein.

First, a TUA2-knockout strain was generated. A paromomycin-resistant gene cassette was inserted into intron 1 of TUA2 in the wild-type strain by CRISPR/Cas9-mediated gene editing (Fig. 1C, upper panel). Insertion was detected by PCR and verified by sequencing (Fig. S1). Semi-quantitative reverse-transcription PCR (RT-PCR) showed that TUA2 was not expressed, whereas TUA1 expression was nearly doubled in a transformant named tua2(int1) (not shown), as in other α-tubulin mutant strains generated by insertional mutagenesis (Kato-Minoura et al., 2020). This indicates that α-tubulin was expressed exclusively from TUA1 in tua2(int1) cells. Next, we introduced a specific donor DNA into the TUA1-coding sequence in tua2(int1). The donor DNA encodes a C-terminal sequence of the TUA1 protein with or without the mutations (Fig. 1B), the 3′ untranslated region (UTR) of TUA1 and a hygromycin-resistance gene cassette (Fig. 1C, lower panel; Fig. S2). PCR screening identified several candidates for both the control and the TUA1 mutant strains, which were subsequently selected by sequencing the TUA1 gene in the control and in a mutant strain with four ‘GAG’ to ‘GCG’ mutations, reflecting glutamate to alanine replacements (Fig. 1D). Representative strains were chosen and named ‘TUA1(WT)’ for the control and ‘TUA1(4A)’ for the TUA1 mutant gene with the glutamate replacements.

We performed indirect immunofluorescence microscopy of the nuclear-flagellar apparatus (NFAp) isolated from wild-type (cc124), tpg1, TUA1(WT) and TUA1(4A) strains to examine the level of polyglutamylation. The mutant tpg1 strain lacks TTLL9 polyglutamylase and thereby has a reduced level of axonemal tubulin polyglutamylation (Kubo et al., 2010). In this study, we used two rabbit polyclonal antibodies, polyE#1 and polyE#2, to specifically target glutamate side chains containing three or more glutamates (Kubo and Oda, 2017). Despite their similar specificity, we employed both antibodies to enhance result reproducibility. Although both antibodies exhibit high specificity, preliminary experiments demonstrated that polyE#2 exhibited superior performance in immunofluorescence experiments. PolyE#2 showed uniform staining of both wild-type and TUA1(WT) axonemes along the entire length (Fig. 2A). The tpg1 axoneme had a signal only in the proximal region, consistent with our previous study (Fig. 2A; Kubo et al., 2010). In contrast to wild-type and tpg1 NFAps, the TUA1(4A) NFAps displayed very faint fluorescence (Fig. 2A). GT335, an antibody that recognizes glutamate side chains of any length, including monoglutamylation, showed a staining pattern different from the polyE#2 pattern (Fig. 2B). The wild-type and TUA1(WT) axonemes showed a signal along the length with decreasing intensity towards the tip (Fig. 2B). The tpg1 axoneme displayed a signal only in one-third of the proximal region. In contrast, the TUA1(4A) axoneme showed essentially no signal (Fig. 2B). However, for an unknown reason, the nuclei had intense staining in TUA1(4A).

Western blotting of the axonemes supported the immunostaining observations (Fig. 3B). In contrast to the wild-type axonemes, which were intensely polyglutamylated, the tpg1 and TUA1(4A) axonemes were much less glutamylated. Absence of both polyE and GT335 signals in the TUA1(4A) axonemes suggested that polyglutamylation occurred mostly on the α-tubulin CTT. This is consistent with our previous conclusion that β-tubulin is poorly glutamylated in the Chlamydomonas axoneme (Kubo et al., 2010).

Indirect immunofluorescence microscopy with an antibody (Gly-pep1) recognizing mono- and bi-glycylated tubulin demonstrated that glycylated tubulin was present along the full length of both wild-type and TUA1(WT) axonemes but was absent in the basal bodies and cytoplasmic microtubules (Fig. 3C). Interestingly, the level of glycylation significantly increased by ∼1.2 times in the TUA1(4A) axoneme compared to that in the TUA1(WT) axoneme (Fig. 3B,C). This result is in accordance with a previous report that showed that mutation in the α-tubulin CTT causes an increase in β-tubulin glycylation, implicating a ‘cross-talk’ between α- and β-tubulin polymodification (Redeker et al., 2005). It must be noted, however, that no signals were detected by western blotting using an anti-polyG antibody, which recognizes tubulin with long polyglycylated side chains (Xia et al., 2000), in any of the axonemal samples used (not shown). As in human cilia (Bré et al., 1996; Rogowski et al., 2009), glycylated side chains most likely do not significantly elongate in Chlamydomonas cilia. Indeed, the Chlamydomonas genome does not have a gene encoding a protein homologous to TTLL10, an enzyme that elongates glycyl side chains (Kubo and Oda, 2019).

Although the TUA1(4A) axoneme had an apparently normal protein composition (Fig. 3A) and structure (Fig. 3D), the cells of this strain were severely defective in motility and showed only some irregular movements (Fig. 4A). In contrast, the tpg1 cells could swim at a reduced velocity (Fig. 4A). This suggests that residual polyglutamylation or the glutamate residues of the α-tubulin CTT in the tpg1 axoneme is important for ciliary motility.

To further explore the motility of TUA1(4A) cilia, we performed a sliding disintegration assay, wherein fragmented axonemes are treated with a protease in the presence of ATP to induce microtubule sliding (Okagaki and Kamiya, 1986; Kurimoto and Kamiya, 1991). Strikingly, non-motile TUA1(4A) axonemes did display sliding disintegration and the velocity of microtubule sliding was much higher than the velocity of wild-type, TUA1(WT) or tpg1 axonemes (Fig. 4B). The tpg1 mutation has also been shown to produce faster microtubule sliding in the axoneme lacking outer-arm dyneins (Kubo et al., 2010, 2012). Therefore, the TUA1(4A) results support the idea that the negative charges on the tubulin C-terminal region increase the inter-microtubule friction (or some inter-doublet binding force) occurring in the axoneme (Kubo et al., 2010, 2012). Taken together with the fact that TUA1(4A) cells are unable to swim, it is most likely that the polyglutamylation or glutamates in the α-tubulin CTT are necessary for proper force generation of dynein required for the regular ciliary beating.

We next investigated the function of the β-tubulin CTT by truncating its glutamate-rich region. Chlamydomonas has two β-tubulin genes, TUB1 (Cre12.g542250) and TUB2 (Cre12.g549550) (Fig. 5A), encoding identical polypeptides with nine glutamate residues in their CTTs (Fig. 5B). We first generated a TUB1-knockout mutant by inserting a paromomycin-resistance gene cassette into the TUB1 gene of the wild-type strain (Fig. 5C, upper panel). Genotyping and semi-quantitative RT-PCR revealed that TUB1 expression was undetectable in the novel tub1(int2) mutant strain, indicating that it expressed β-tubulin only from TUB2 (not shown). Because amino acid substitutions for all nine glutamate residues using gene editing were expected to be difficult, we then introduced a donor DNA in the TUB2 coding sequence of tub1(int2) to completely or partially truncate the glutamate-rich region of the β-tubulin CTT (Fig. 5C, lower panel). We generated mutants lacking four or six glutamates by inserting a stop codon in the TUB2 coding sequence near the 3′ UTR (Fig. 5B,C). PCR and sequencing identified an insertion of the donor DNA encoding a stop codon, the 3′ UTR of TUB2 and the hygromycin-resistance gene for the three transformants, designated TUB2(WT), TUB2(Δ4E) and TUB2(Δ6E), respectively (Fig. 5D). Despite repeated trials, we could not isolate a mutant lacking nine glutamates of β-tubulin CTT in the background of tub1(int2) for some reason. The mutants TUB2(Δ4E) and TUB2(Δ6E) both demonstrated a somewhat reduced rate of cell proliferation compared to that of the controls, with the latter exhibiting a more pronounced effect (Fig. S3). Tetrahymena mutants in which β-tubulin glutamates were substituted in three out of the five positions in the CTT (E437, E438, E439 E440, and E442) showed a hypomorphic or lethal phenotype (Xia et al., 2000; Thazhath et al., 2002; Redeker et al., 2005). Despite the necessity to distinguish the effect resulting from amino acid substitutions and truncations in the β-tubulin CTT, the viability of the TUB2(Δ6E) strain was surprising based on the lethality of Tetrahymena mutants.

Both TUB2(Δ4E) and TUB2(Δ6E) cells were severely impaired in motility (Fig. 6A), suggesting that the β-tubulin CTT is also involved in ciliary function, similarly to the α-tubulin CTT. Interestingly, TUB2(Δ4E) cells exhibited three distinct patterns of motility. These patterns were characterized by nearly full motility, partial motility and non-motile behavior. The reason for this variability in motility is currently unknown, but it is likely attributed to the unstable interaction between the axonemal dynein and the adjacent outer doublet. Most severely impaired was the TUB2(Δ6E) mutant, which did not display swimming at all (Fig. 6A). This motility defect might well be due to abnormal ciliary structure as TUB2(Δ6E) cilia were about half the length of wild-type cilia (Fig. 6B). TUB2(Δ6E) cells failed to regenerate cilia after pH-shock-induced ciliary detachment, whereas TUB2(Δ4E) cells showed almost normal ciliary regeneration (Fig. 6C). In addition, intraflagellar transport (IFT) particle proteins were found to be abnormally accumulated in the cilia of β-tubulin mutant strains, especially those of TUB2(Δ6E) (Fig. S4).

Western blotting using polyE and GT335 antibodies showed that the level of polyglutamylation is increased or unchanged in TUB2(Δ4E) and TUB2(Δ6E) axonemes (Fig. 7A). In contrast, the glycylation levels of axonemes in these mutants were significantly decreased; in particular, the signal was almost completely missing in TUB2(Δ6E) (Fig. 7A). Therefore, it is likely that glycylation mainly occurs on the β-tubulin CTT in Chlamydomonas.

To determine the localization of axonemal glutamylation and glycylation, we performed immunostaining of NFAps. As expected, both TUB2(Δ4E) and TUB2(Δ6E) axonemes showed slightly increased polyE#2 staining (Fig. 7B). Also as expected, although the NFAp of wild-type and TUB2(WT) strains showed intense axonemal staining with the antibody recognizing glycylated tubulin (Gly-pep1), the NFAps of TUB2(Δ4E) and TUB2(Δ6E) strains had significantly decreased levels of axonemal staining (Fig. 7C).

To explore the reason for the paralyzed cilia of TUB2(Δ6E) cells, we examined the cross-section images of the isolated axonemes by transmission electron microscopy (TEM). The wild-type (cc124), TUB2(WT) and TUB2(Δ4E) axonemes had a normal ‘9+2’ configuration (Fig. 8A; Table S1). Interestingly, however, almost all of the TUB2(Δ6E) axoneme images lacked the two central-apparatus microtubules and, instead, contained electron-dense materials in the central lumen (Fig. 8A; Table S1). Several Chlamydomonas mutants are also known to lack the central apparatus. It is interesting to note that two central apparatus-lacking mutants, pf15 and pf19, have mutations in the subunits of katanin (Dymek et al., 2004; Dymek and Smith, 2012), a protein complex that severs microtubules (McNally and Vale, 1993) depending on its interaction with the β-tubulin CTT (Zehr et al., 2020).

The presence of central-apparatus proteins was examined in the axonemes from wild-type, TUB2(WT), TUB2(Δ4E) and TUB2(Δ6E) strains by a silver-stained SDS-PAGE gel. Although TUB2(Δ4E) cilia showed a normal band pattern, TUB2(Δ6E) cilia lacked several bands, reflecting the deficiency of the central apparatus (Fig. 8B, left). To assess the presence of the proteins that constitute the central-apparatus projections, polyclonal antibodies were raised against CPC1 (C1b projection of the central apparatus; 205.1 kDa), FAP221 (C1d projection; 104 kDa), FAP101 (C1a projection; 85.7 kDa), KLP1 (C2c projection; 83 kDa), PF20 (C2 projection; 53 kDa) and FAP7 (C1a projection; 54.7 kDa) (Fig. S5), and used for western blotting of axonemal samples. The results showed that TUB2(Δ6E) cilia almost completely lacked all the proteins examined, similar to the pf19 mutant axoneme deficient in the central-pair apparatus (Fig. 8B, right; Fig. S5).

Even though we need to differentiate between the functions of the glutamate residues of tubulin CTTs and the post-translational modifications that occur there, our study suggests that glutamylation specifically occurs on α-tubulin, whereas glycylation specifically occurs on β-tubulin in Chlamydomonas cilia. More specifically, some or all of the four glutamate residues (E445, E447, E449 and E450) in the α-tubulin CTT are likely the main sites for glutamylation, whereas some or all of the six residues (E435, E437, E439, E440, E441 and E442) in the β-tubulin CTT are likely the main sites for glycylation. This feature is the same as that of Tetrahymena (Redeker et al., 2005) but different from that of sea urchin sperm (Multigner et al., 1996), mammals (van Dijk et al., 2007) and Drosophila (Rogowski et al., 2009), in which the two species of tubulin likely undergo both glutamylation and glycylation.

We found that TUA1(4A) is almost completely immotile in contrast to tpg1, which displays motility, albeit reduced. This might be owing to the different degrees of tubulin polyglutamylation inhibition in their axonemes; tpg1 retains polyglutamylated tubulin in the ciliary basal region, presumably catalyzed by TTLL enzyme(s) other than TTLL9, whereas TUA1(4A) almost completely lacks polyglutamylation from the base to tip of the axoneme (Fig. 2). Previously, we speculated that polyglutamylation increased inter-doublet friction of the outer-doublet microtubules. This speculation was based on the observation that microtubule sliding in the axonemal fragments of tpg1 oda2 double mutant, which lacks outer-arm dynein, was faster than that in the axonemal fragments of oda2. We thought that the putative friction (strong dynein–microtubule interaction) might be important for the generation of ciliary motility (Kubo et al., 2010, 2012). The presence of motility in tpg1 but not in TUA1(4A) cilia suggests that, if the inter-doublet friction was, in fact, important for motility, friction in the proximal region of the cilium might function as a trigger of the ciliary beating cycle. Another possible reason for the loss of motility in TUA1(4A) is that axonemal dyneins cannot generate force in the absence of the four glutamates in the α-tubulin CTT. Combining outer-arm dynein- or inner-arm dynein-lacking mutations with TUA1(4A) might provide further insights into the potential role of these four glutamates in force generation. Whether the loss of polyglutamylation per se is the cause of the motility loss in TUA1(4A) awaits further studies.

Motility analyses of tpg1 combined with various dynein-lacking mutants led to the conclusion that polyglutamylation is important for the function of inner-arm dynein but not that of outer-arm dynein, because inner-arm-dynein-deficient mutants with normal outer-arm dynein, but not mutants lacking outer arm, remain motile in the absence of polyglutamylation (Kubo et al., 2010, 2012). A later detailed study strongly suggested that polyglutamylated tubulin directly associates with the nexin–dynein regulatory complex (Kubo et al., 2012, 2015), a structure that bridges the adjacent outer doublets and regulates the function of inner-arm dynein. However, our present study suggests that the polyglutamylated side chains or the glutamate residues of the α-tubulin CTT might also affect the function of outer-arm dynein, as their loss resulted in a non-motile phenotype even when outer-arm dynein was intact. The true importance of the α-tubulin CTT needs to be examined using various dynein-lacking mutants in the background of mutations that inactivate polyglutamylation while preserving the four glutamates of the α-tubulin CTT.

Many TUB2(Δ4E) cells failed to swim (Fig. 6A), even though their axonemal structure was normal (Fig. 8A; Table S1). This suggests that the negative charges of glutamate residues of the β-tubulin CTT are also involved in the ciliary motility. However, because the lack of glutamate residues in TUB2(Δ4E) coincidentally caused decreased glycylation (Fig. 7), there is also the possibility that glycylated tubulin affects the ciliary motility. Indeed, tubulin polyglycylation has been reported to regulate the function of axonemal dyneins (Gadadhar et al., 2021). Although ‘poly'glycylation likely does not occur in the Chlamydomonas axoneme, glycylation might be important for ciliary motility. This can be clarified by generating a novel Chlamydomonas mutant lacking the enzyme(s) responsible for glycylation.

In the TUA1(4A) axoneme, polyglutamylation was almost completely missing, whereas glycylation increased (Fig. 3C). In the TUB2(Δ4E) axoneme, in contrast, glycylation was missing, whereas polyglutamylation slightly increased. This is consistent with the previous reports that there is an inverse correlation between the levels of glutamylation and glycylation (Redeker et al., 2005; Wloga et al., 2009; Rogowski et al., 2009). In accordance with these studies, TTLL3 (monoglycylase) and TTLL7 (glutamylase) were found to compete for the same sites on β-tubulin in vitro (Garnham et al., 2017). This competition for the same substrate is interpreted to be a regulatory mechanism to control the level of tubulin modification, but we still do not understand how the level of α-tubulin glutamylation affects the level of β-tubulin glycylation. Because polyglycylation was also found to modulate ciliary motility (Gadadhar et al., 2021), one idea is that polyglycylation can partially complement the motility defect caused by the lack of polyglutamylation. How these two modifications affect ciliary motility in a coordinated way still needs to be explored. The generation of the double-mutant TUA1(4A)×TUB2(Δ6E) might offer intriguing insights in future studies.

Unexpectedly, we found that TUB2(Δ6E) cells are viable even though they have a slow cell proliferation rate (Fig. S3). This is different from Tetrahymena mutants with three mutations in five glycylation sites of β-tubulin (E437, E438, E439, E440 and E442), which show a hypomorphic or lethal phenotype (Xia et al., 2000; Thazhath et al., 2002; Redeker et al., 2005). Although Chlamydomonas cells without the nine glutamates in the β-tubulin CTT appear to be inviable (Fig. 5B), our results indicate that the six glutamate residues (E435, E437, E439, E440, E441 and E442) corresponding to the five glycylation sites of Tetrahymena are dispensable for cell survival in Chlamydomonas.

The Tetrahymena mutants harboring a mutation in the glycylation sites of the β-tubulin CTT partially lack the B-tubule outer doublets and the central-pair microtubules, and abnormally accumulate IFT particle proteins between the ciliary membrane and outer doublets (Thazhath et al., 2002; Redeker et al., 2005). Similarly, the TUB2(D6E) cilium completely lacks the central apparatus, although it has normal outer doublets (Fig. 8A). In addition, IFT particle proteins, particularly IFT-B proteins, were increased in the cilia (Fig. S4). This observation suggests that IFT particle proteins accumulate within the central lumen of TUB2(Δ6E) cilia, similar to strains that lack the central apparatus (Lechtreck et al., 2013).

TUB2(Δ6E) was unable to regenerate cilia after pH-shock-induced deciliation (Fig. 6C). During the assembly of cilia, vast amounts of tubulins are transported into cilia by IFT (Craft et al., 2015). In vitro pull-down assays suggested that the N-terminal domains of the IFT particle proteins IFT74 and IFT81 form a tubulin-binding module. More specifically, the calponin homology domain of the IFT81 N-terminus binds with the globular cores of the tubulin dimer and the basic IFT74 N-terminus likely associates with the acidic tail of β-tubulin (Bhogaraju et al., 2013). In accordance with this idea, Chlamydomonas mutants deficient in either the IFT81 or the IFT74 N-terminus assemble flagella with reduced speed (Kubo et al., 2016; Brown et al., 2015). Furthermore, in vitro imaging demonstrated that IFT of β-tubulin with an abolished E-hook is significantly reduced by approximately 90% (Craft van de Weghe et al., 2020). The inability of TUB2(Δ6E) to assemble full-length cilia is consistent with these previous results, indicating that the β-tubulin CTT is involved in the IFT of tubulin. To confirm this idea, the frequency of IFT in the TUB2(Δ6E) cilia should be investigated.

Interestingly, the phenotype of TUB2(Δ6E) – characterized by non-motile cilia lacking the central apparatus – bears resemblance to a zebrafish morphant that lacks glycylation and glutamylation of the axoneme (Pathak et al., 2011). One possibility is that tubulin glycylation plays a role in central-apparatus assembly. However, as we cannot differentiate the effects of the lack of glycylation and lack of the β-tubulin C-terminus using TUB2(Δ6E), it is necessary to investigate this hypothesis with a mutant carrying intact β-tubulin with decreased glycylation. We are currently working on generating a Chlamydomonas strain deficient in TTLL3, a gene that encodes tubulin monoglycylase, to address this.

The phenotype of TUB2(Δ6E) is also identical to that of Chlamydomonas mutants pf15 and pf19, each with a mutation in a gene encoding a katanin subunit (Dymek et al., 2004; Dymek and Smith, 2012). Katanin, an AAA ATPase (ATPases associated with diverse cellular activities), is a microtubule-severing enzyme (McNally and Vale, 1993). Katanin was shown to be involved in the formation of the central apparatus by severing peripheral microtubules to produce central microtubule seeds (Sharma et al., 2007; Liu et al., 2021). Furthermore, tubulin polymodifications (Sharma et al., 2007; Lacroix et al., 2010; Waclawek et al., 2017; Shin et al., 2019; Joachimiak et al., 2020; Szczesna et al., 2022) and glutamate residues in the β-tubulin CTT have been shown to be important for katanin to sever microtubules in vitro (Zehr et al., 2020). Specifically, Zehr et al. (2020) have shown that glutamate residues in β-tubulin increase the negative charges of microtubules to activate the katanin ATPase. Therefore, the loss of the central apparatus in TUB2(Δ6E) lacking the β-tubulin CTT could be due to the deficiency of katanin function. It is important to identify specific glutamate residues of the β-tubulin CTT that are involved in central-apparatus assembly in the future. Katanin has also been suggested to be involved in cell mitosis (Vale, 1991; McNally and Thomas, 1998; McNally et al., 2006). In fact, we noticed that TUB2(Δ6E) cells proliferate significantly slower compared to control cells (Fig. S3), suggesting an abnormality in cell mitosis. Therefore, various aspects of katanin function dependent on the β-tubulin CTT still need to be explored. Our system to introduce a tubulin mutation in Chlamydomonas must contribute to clarifying these important microtubule-based functions in the future.

The cells (cc124 and generated mutants, Table S2) were maintained on Tris-acetate-phosphate (TAP; Gorman and Levine, 1965) agar plates for long-term storage. For the experiments, the cells were cultured in liquid TAP medium at 25°C with a light/dark cycle of 12 h:12 h and aeration.

The target sequences were searched using CRISPRdirect (http://crispr.dbcls.jp/). The crispr RNAs (crRNAs) (Table S3) and trans-activating crispr RNAs (tracrRNAs) (Fasmac) were dissolved in nuclease-free duplex buffer (Integrated DNA Technologies; IDT buffer), dispensed in micro-8-tube strips and cryo-preserved at −80°C. The crRNA (1 µl, 40 µM) was annealed with tracrRNA (1 µl, 40 µM) by heating at 95°C for 2 min and then cooling down gradually to 25°C over 90 min. Then, 2 µl of annealed guide RNA was incubated with 0.5 µl Cas9 protein (10 µg/ml; Fasmac) with 7.5 µl IDT buffer at 37°C for 15 min to generate Cas9/gRNA ribonucleoprotein.

The cells were grown on a TAP agar plate for 5 days under constant light and treated with ∼6 ml autolysin produced in house as described previously (Picariello et al., 2020) for 1 h to remove cell walls. The cells were then agitated gently at 40°C for 30 min. The cells were collected by centrifugation (2000 g, 3 min, room temperature) and washed by 2% sucrose in TAP medium (TAP sucrose). After centrifugation (2000 g, 3 min, room temperature), the cells were resuspended to a concentration of 2–7×10⁹ cells/ml. Cas9/gRNA ribonucleoprotein (10 µl), donor DNA (∼10 µl; up to 2 µg; Fig. S2) and the cells (∼110 µl) were mixed to give a final volume of 125 µl. The mixture was transferred to a cuvette (0.2 cm gap; Bio-Rad) and electroporated (ECM 630 Electroporation System, BTX) immediately at 350 V, 25 Ω and 600 µF. The cuvette was then incubated at 15°C for 60 min. The transformed cells were suspended into 10 ml of TAP sucrose and gently rocked for 24 h under dim light. The cells were collected by centrifugation at 2000 g, plated on TAP agar containing 10 µg/ml hygromycin or 10 µg/ml paromomycin, and cultured for 5–7 days under constant light.

Western blotting was performed according to Towbin et al. (1979) with some modifications. SDS-protein samples were separated by electrophoresis using a handmade SDS-polyacrylamide gel (7.5 or 9% polyacrylamide) and transferred to a PVDF membrane (Immobilon-P, pore size 0.45 µm; Merck-Millipore). The membrane was probed with primary antibodies (Table S4), followed by incubation with secondary antibodies: goat anti-mouse IgG (H+L) (1:2000, Invitrogen, 31430) or goat anti-rabbit IgG (H+L) (1:2000, Invitrogen, 31460). After treatment with Chemi-Lumi One Super (Nakalai Tesque, 02230), signals on the membrane were detected using the Fusion Solo S system (Vilver Bio Imaging). Full images of western blots using anti-tubulin antibodies are shown in Fig. S6.

The cDNA sequences encoding partial CPC1 (Cre03.g183200; amino acids 395–801), FAP221 (Cre11.g476376; amino acids 1–762), FAP101 (Cre02.g112100; amino acids 652–835), KLP1 (Cre02.g073750; amino acids 451–633), PF20 (Cre04.g227900; amino acids 9–606) and FAP7 (Cre12.g531800; amino acids 269–507) were inserted into the pGEX vector (Amersham) and the polypeptides were expressed in BL21 (DE3) competent Escherichia coli (New England Biolabs). Rabbit polyclonal antibodies were raised against the purified polypeptides. Antisera without any purification were found to be sufficient for western blotting for all of the generated antibodies. Full images of western blots using these antibodies are shown in Fig. S5.

Cilia were isolated according to Witman et al. (1972) with some modifications. Briefly, fully grown cells were collected with centrifugation (2000 g, 5 min) and treated with 1 mM dibucaine-HCl (Wako) to amputate their cilia. Cilia were washed and collected by centrifugation (32,000 g, 20 min, 4°C). Occasionally, flagella were demembranated by 0.1% Igepal CA-630 (Sigma-Aldrich) to prepare axonemal samples.

Immunostaining was carried out following Sanders and Salisbury (1995). Cells from 5–10 ml culture were collected by gentle centrifugation and treated with 6 ml autolysin for 60 min to remove cell walls. The cells were washed with NB buffer [6.7 mM Tris-HCl (pH 7.2), 3.7 mM EGTA, 10 mM MgCl₂ and 0.25 mM KCl] and placed on an eight-well-slide glass (8 mm well; Matsunami) treated with polyethylenimine. The cells were then fixed with −20°C methanol and −20°C acetone for 5 min, respectively. The cells were first treated with blocking buffer (1% bovine serum albumin and 3% fish skin gelatin in PBS) and incubated with primary antibodies (Table S4), followed by incubation with secondary antibodies (goat anti-rabbit IgG Alexa Fluor 488, 1:200, Invitrogen; goat anti-mouse IgG Alexa Fluor 594, 1:200, Invitrogen). The cells were treated with antifade mountant (SlowFade Diamond, Thermo Fisher Scientific) and encapsulated with a glass coverslip. The sample was examined with a microscope (BX53, Olympus) equipped with an objective lens (100× UPlan FL N oil objective, Olympus). Images were acquired with a CCD camera (ORCA-Flash4.0 sCMOS, Hamamatsu).

TEM observation of the isolated axonemes was performed as described previously (Huang et al., 1979). Briefly, the isolated axonemes were fixed for 1 h with 2% glutaraldehyde in 10 mM phosphate buffer (pH 7.0) containing 1% tannic acid. After three washes with the buffer, the axonemal pellets were fixed on ice for 45 min with 1% osmium tetroxide in phosphate buffer. The pellets were dehydrated with graded series of ethanol. The pellets were treated with propylene oxide and then embedded in epoxy resin. Thin sections were made by an ultra-microtome (ULTRACUT S, Reichert Leica) and were stained with uranyl acetate and lead citrate. Images were obtained using a JEM-2100F transmission electron microscope (JEOL, Tokyo, Japan).

To assess ciliary regeneration kinetics, cells were deflagellated by pH shock (Rosenbaum et al., 1969). After the deflagellation, aliquots of the cells were isolated and fixed by 1% glutaraldehyde at 15 min intervals up to 180 min. The fixed cells were concentrated by centrifugation at 2000 g for 1 min. Small aliquots (∼5 µl) of the cells were then embedded in a slide and observed using a microscope (BX53, Olympus) equipped with a 20× UPlan FL N objective lens (Olympus). Images were acquired with a CCD camera (ORCA-Flash4.0 sCMOS, Hamamatsu). One cilium each from at least 30 cells was measured by ImageJ to obtain the average length of each time point.

Swimming velocity was acquired by tracking images of the moving cells. Briefly, the cells under the dark-field microscope equipped with a 40× objective were recorded using a digital camera with a frame rate of 30 fps, and the obtained movies were processed with ImageJ.

Microtubule sliding velocity during axonemal disintegration was measured as described previously (Kurimoto and Kamiya, 1991). Briefly, fragmented axonemes (∼5 µm in length) were placed in a perfusion chamber under a dark-field microscope. Microtubule sliding was induced by a solution containing 0.5 µg/ml of Type VII bacterial protease (Sigma-Aldrich) and ATP. This process was recorded using a 100× objective, an oil-immersion dark-field condenser, a light source of a mercury lamp (Olympus, U-RLF-T), and a CCD camera with a frame rate of 30 fps (Olympus, M-3204C).

We greatly appreciate Dr Ritsu Kamiya (Chuo University, Tokyo) for critically reading the manuscript and constructive discussion. We thank Mrs Natsuko Maruyama (University of Yamanashi) for technical assistance. Most of the experiments were conducted at University of Yamanashi Medical School.

The peer review history is available online at https://journals.biologists.com/jcs/lookup/doi/10.1242/jcs.261070.reviewer-comments.pdf.