PF19 encodes the p60 catalytic subunit of katanin and is required for assembly of the flagellar central apparatus in Chlamydomonas

Since the discovery of katanin nearly two decades ago, this microtubule severing protein has been implicated in a variety of cellular processes in a diverse range of eukaryotes, from unicellular organisms such as Chlamydomonas to humans (McNally and Vale, 1993). Katanin is comprised of two subunits: p60, the catalytic subunit, and p80, a regulatory subunit (Hartman et al., 1998). The p60 subunit is a member of the AAA ATPase family of proteins and possesses microtubule severing activity (Hartman et al., 1998; McNally et al., 2000). In its ATP-bound form katanin assembles into hexameric structures that bind to the microtubule lattice (Hartman et al., 1998; Hartman and Vale, 1999). In one model, interaction of katanin with the C-terminal tails of tubulin results in destabilization of tubulin–tubulin interactions, thus causing the microtubules to sever. The p80 subunit appears to play a role in both targeting p60 and enhancing its microtubule severing activity (McNally et al., 2000). Importantly, katanin is the founding member of a family of microtubule severing proteins that includes spastin, fidgetin, and VPS4 (reviewed in Roll-Mecak and McNally, 2010).

Numerous studies in a variety of cell types have revealed important roles for katanin in organizing centrosomal and/or non-centrosomal microtubule arrays. For example, katanin has been shown to play a role in spindle organization in meiosis (Caenorhabditis elegans) and mitosis (several organisms and cell types) (reviewed in Roll-Mecak and McNally, 2010). A recent study in C. elegans indicated that the role of katanin in spindle assembly during meiosis was independent of microtubule severing activity (McNally and McNally, 2011). Yet, the microtubule severing activity of katanin contributes to mitotic spindle length scaling among Xenopus species (Loughlin et al., 2011). In neurons, katanin appears to regulate axonal growth as well as the length and number of microtubules in mature neurons (Karabay et al., 2004; Yu et al., 2005). In plants, katanin-mediated microtubule severing is required for generating cortical microtubule arrays. Microtubule branches are initiated by gamma-tubulin rings which bind along pre-existing microtubules; katanin-mediated microtubule severing releases these branches allowing for generation of parallel arrays of microtubules (Stoppin-Mellet et al., 2002; Wasteneys, 2002; Stoppin-Mellet et al., 2003; Stoppin-Mellet et al., 2007; Nakamura et al., 2010). More recently, katanin has been shown to act as a depolymerase involved in regulating microtubule dynamics during cell migration (Díaz-Valencia et al., 2011; Zhang et al., 2011).

In the green alga Chlamydomonas, several studies have suggested that katanin plays a role in the flagellar autotomy response to stress and in the release of basal bodies from the transition zone prior to mitosis (Lohret et al., 1998; Lohret et al., 1999; Rasi et al., 2009; Parker et al., 2010). We have shown that Chlamydomonas mutants with defects in the p80 subunit of katanin have no defects in mitosis or flagellar autotomy; rather, these mutants lack the central pair of microtubules (Dymek et al., 2004). One possibility is that the central pairless phenotype is uniquely associated with defects in the p80 regulatory subunit of katanin. However, this same central pair defect has also been observed in Tetrahymena mutants which do not express either subunit of katanin (Sharma et al., 2007). In these studies both the p60 and p80 subunits of katanin are required for assembly of motile, normal-length cilia. Therefore, in at least two protozoa katanin plays an essential role in cilia biogenesis and specifically in formation of the central pair of microtubules. It is not known if katanin serves a similar role in other ciliated/flagellated cell types.

Unlike the axonemal doublet microtubules that comprise eukaryotic cilia and flagella, the central microtubules are not nucleated from the basal bodies. In conventional transmission electron microscopy the proximal ends of these microtubules (their minus ends) do not appear to have connections to other ciliary or cellular structures. To gain insight into a mechanism for central apparatus assembly, we have undertaken an analysis of the Chlamydomonas pf19 mutation. Chlamydomonas pf19 mutants are completely viable with normal cell division and flagellar autotomy responses. However, they have paralyzed flagella that lack the central apparatus. Here we report that PF19 corresponds to the KAT1 gene encoding the p60 subunit of katanin and provide evidence that in Chlamydomonas katanin plays a critical role in assembly of the central pair of microtubules.

We previously reported that the PF15 gene encodes the p80 subunit of katanin (Dymek et al., 2004). Given that both the pf15 and pf19 mutants have paralyzed flagella that lack the central apparatus and that the pf19 mutation maps very close to the Chlamydomonas KAT1 gene encoding the p60 subunit of katanin (Kathir et al., 2003), we suspected that pf19 may carry a mutation in the KAT1 gene. In our initial attempts to sequence the KAT1 gene amplified from genomic DNA isolated from the pf19 mutant, we were unable to detect any mutations (Dymek et al., 2004). However, since Chlamydomonas DNA is extremely GC rich, we suspected that there may have been technical difficulties in sequencing this region of genomic DNA. Therefore, we recently re-sequenced the KAT1 gene using different primers and DNA sequencing reagents (see Materials and Methods). We discovered a single mutation, a T to C transition, which converts leucine 277 to serine (Fig. 1A). To ensure that this mutation was not introduced during amplification of the KAT1 gene, we confirmed the presence of this mutation using multiple primer sets and PCR products were sequenced directly; in addition we isolated RNA from pf19 cells and used RT-PCR to confirm the presence of the mutation in the KAT1 transcript produced in these cells. This residue is highly conserved from Chlamydomonas to Arabidopsis to humans and is within the AAA domain (Fig. 1B). Therefore, we suspected that pf19–p60 is non-functional.

To confirm that the central pairless and flagellar paralysis phenotypes are caused by the mutation in KAT1, we co-transformed pf19 cells with a BAC clone containing the full length KAT1 gene as well as a plasmid with a selectable marker (see Materials and Methods). Clone 27D3 rescues both the motility and structural phenotypes in the pf19 mutant (data not shown). Out of 96 paromomycin-resistant transformants, five independent isolates were obtained with restored motility. This result is consistent with a co-transformation efficiency of 5% as we and others have previously reported (Tam and Lefebvre, 1993; Smith and Lefebvre, 1996; Smith and Lefebvre, 1997; Smith and Lefebvre, 2000; Dymek et al., 2004). In all of these strains, the presence of the central apparatus was confirmed by western blot using the anti-PF20 antibody (Smith and Lefebvre, 1997) (data not shown). We then subcloned the KAT1 gene from BAC 27D3. This 6kb subclone (pBCp60) includes ∼600 bp upstream of the translational Start and ∼1200 bp downstream of the translational Stop (see supplementary material Fig. S1 for details). The coding sequence of katanin p60 is the only coding sequence predicted within this 6 kb span of genomic DNA. The pBCp60 plasmid rescues both the motility and structural defects in the pf19 mutant upon transformation (co-transformation efficiency of 5%). These results indicate that PF19 corresponds to KAT1 encoding the p60 subunit of katanin and that the pf19 mutation is recessive.

In an effort to localize p60 we generated a plasmid construct encoding p60 fused with a GFP tag (see Materials and Methods and supplementary material Fig. S1). We engineered the p60–GFP construct to express GFP at the amino terminus of the p60 protein. This construct rescued motility and central apparatus assembly upon transformation into pf19 mutant cells (co-transformation efficiency of 5%, Fig. 1C). For axonemes prepared from pf19 cells, 8.7% of transverse sections have a central apparatus. In pf19 mutants rescued with the GFP–pBCp60 plasmid, greater than 90% of transverse sections have a central apparatus, comparable to wild type (Fig. 1C). Expression of the GFP construct was confirmed by western blots of cell extracts isolated from rescued strains (Fig. 1D). We could not detect p60–GFP in isolated flagella indicating that either all p60–GFP is located in the cell body or that the amount in flagella is below the limit of detection by western blot. Attempts to localize GFP–p60 by fluorescence microscopy were not successful. These experiments included visualizing GFP directly as well as using anti-GFP antibodies and fluorescently labeled secondary antibodies. Based on western blots as well other studies of p60, we suspect that p60 is not an abundant protein. It is also possible that the GFP is not accessible to the anti-GFP antibodies.

As a second method to potentially localize p60 in wild-type cells and to determine if p60 protein is expressed in the pf19 mutant, we obtained anti-p60 katanin antibodies provided by the Quarmby lab (Rasi et al., 2009). The anti-p60 serum did not recognize the tagged p60–GFP protein in whole cell extracts (supplementary material Fig. S2B). Based on results shown in supplementary material Fig. S2A we concluded that this antibody is not of sufficiently high titer or specificity to recognize p60 in cell extracts. Therefore, we did not use this p60 antibody in any additional studies. To date, no commercially available antibodies recognize Chlamydomonas p60.

One possibility is that the central pairless phenotype of the pf19 mutation is due to an incomplete loss of p60 function. To determine if the pf19-p60 protein retains severing activity, we obtained cDNAs encoding either wild-type p60 or pf19-p60 and expressed either of these two proteins in bacteria. For both wild-type p60 and pf19-p60, the proteins were expressed as fusions with a GST tag at the amino terminus of the protein (see Materials and Methods and supplementary material Fig. S2A). Importantly, to obtain full length protein that is soluble under non-denaturing conditions, these proteins were expressed in Rosetta-gami bacteria (Novagen) and induction of expression was carefully controlled (see Materials and Methods). The expressed protein was affinity purified and then assessed for microtubule severing activity using microtubules assembled in vitro and dark-field microscopy (see Materials and Methods).

In the severing assay, microtubules are added to a flow-through chamber constructed as described in the Materials and Methods. Microtubules that do not adhere to the slide are washed through the chamber. In control experiments with no added expressed p60 protein, flow of buffer through the chamber results in the detachment of a certain small percentage of microtubules from the coated slide which then flow out of the chamber. This loss of microtubules was quantified from multiple control experiments using ImageJ (see Materials and Methods) as a 16% reduction in microtubule density. In stark contrast, 20 minutes after bacterially expressed Chlamydomonas wild-type p60 and ATP are added to the chamber, the density of microtubules was reduced by 40% (Fig. 2). This significant reduction (P = 0.0005, Student’s t-test) in microtubule density was not seen in experiments using expressed pf19-p60. In experiments with mutant protein, the small reduction in microtubule density observed 20 minutes after addition of the pf19-p60 protein (∼16%) was not significantly different (P = 0.98, Student’s t-test) from control experiments with no added p60 protein. The simplest interpretation is that the pf19 mutation abolishes microtubule severing activity of the p60 protein.

To explore the difference between the phenotype for the pf19 mutant (this study) and p60 mutants generated by RNAi-mediated reduction of KAT1 expression (Rasi et al., 2009), we used an artificial microRNA (amiRNA) approach (Molnar et al., 2009; Zhao et al., 2009) to reduce KAT1 expression. This technique takes advantage of the cell's own microRNAs that function to regulate endogenous gene expression. Two different gene-specific amiRNA vectors targeting KAT1 were constructed as described previously (Molnar et al., 2009) and in the Materials and Methods, and these were separately transformed into wild-type Chlamydomonas cells (Kindle, 1990). After selection of transformants on paromomycin, colonies were picked into liquid media and screened for abnormal cell division or swimming phenotypes. A total of 2,112 transformants were picked into liquid media from three independent transformations for each construct. Since no anti-p60 antibodies are available for Chlamydomonas (described above), we continued our screen based on the known pf19 phenotype, paralyzed flagella which lack the central apparatus.

Of the transformants picked into liquid media, 90 strains showed motility defects. This is not uncommon since integration of the transforming plasmid may generate a mutant phenotype (Tam and Lefebvre, 1993). These strains were then screened for central apparatus defects by western blot using antibodies generated against the central pair protein, PF20. Of these strains, three (A5, G3, H6) had reduced PF20 protein in isolated axonemes and were chosen for additional studies. All three of these transformants were generated using the same amiRNA construct targeting nucleotides 79–98 of the KAT1 coding sequence. Importantly, this region of the KAT1 gene is unique and shares no similarity with other members of the microtubule severing family of proteins.

Based on western blots of axonemes isolated from these three strains, the central pair protein PF20 is reduced to 50% (A5) or to nearly 0% (H6) of wild-type levels (Fig. 3A). Thin section transmission electron microscopy revealed that these amiRNA mutants lacked the central apparatus (Fig. 3B). The numbers of axoneme transverse sections which lack the central apparatus or in which only electron dense material was found are quantified in (Fig. 3C). As predicted by western blots, approximately 50% of axonemes retained the central apparatus in transverse sections of A5 axonemes. However, in both G3 and H6, only about 10% of transverse sections of axonemes retained the central pair of microtubules. By comparison, in axonemes prepared from pf19 cells, 8.7% of transverse sections have a central apparatus.

To confirm that these strains have reduced levels of the KAT1 transcript, we isolated RNA from each amiRNA strain, as well as wild-type and pf19 cells, and examined transcript levels by northern blot analysis and subsequent densitometry (Fig. 4A). RNA was isolated both before and 45 minutes post-deflagellation. The pf19 mutant produces a KAT1 transcript as expected since pf19 is a missense mutation. Northern blots prepared from cells prior to deflagellation revealed that the KAT1 transcript was reduced to between 40% and 60% of wild-type levels. Northern blots prepared from cells 45 minutes following deflagellation revealed that the KAT1 transcript was reduced to between 30% and 80% of wild-type levels. In the absence of antibodies to detect p60 in whole cell extracts, we could not assess the relationship between transcript levels and p60 protein levels.

To determine if these three transformants represent independent transformation events, we tested for restriction length polymorphisms (RFLPs) at the integration site. Southern blots of DNA isolated from the three amiRNA strains with reduced KAT1 expression and probed with the vector sequence from the plasmid revealed RFLPs in these three strains (Fig. 4B). These results indicate the plasmid integrated into different regions of the genome for each strain and demonstrated that the three strains represent independent isolates.

It is possible that since we screened for transformants with motility and central apparatus defects, we isolated insertional mutants in which the amiRNA construct integrated into a gene required for central apparatus assembly. These transformants were originally identified for their absence of PF20 protein in isolated axonemes. The only insertional mutants isolated to date which lack PF20 are the central pair mutants pf20 and pf15. The only other central apparatus defective mutants available which lack PF20 are pf18 and pf19; to date, no insertional alleles of these mutations have been reported. To determine whether the KAT1 (PF19), PF20 or PF15 genes were affected by the integrated plasmid in the amiRNA strains, we probed southern blots of genomic DNA isolated from each strain with the full length KAT1, PF15 and PF20 genes. No RFLPs were observed for any of these genes (data not shown). Therefore, the amiRNA construct did not disrupt any of these genes. As an important note, no central pairless mutants have been reported in Chlamydomonas in which 50% of axoneme transverse sections lack the central apparatus, such as we see for strain A5.

To determine if p60 katanin plays a role in flagellar excision, we performed deflagellation experiments in response to pH shock. For all three amiRNA strains as well as the original pf19 mutant, 100% of the cells deflagellate in response to acidic pH (Fig. 5A). As is true for wild-type cells, these strains immediately shed their flagella when the pH of the media is dropped to pH4.5. We have also constructed a pf15pf19 double mutant which carries mutations in both the regulatory p80 and catalytic p60 subunits of katanin. This pf15pf19 strain, like both parental strains, has paralyzed flagella which lack the central apparatus (not shown). This double mutant is also wild type for the flagellar excision response (Fig. 5A).

Rasi et al. proposed that the microtubule-severing function of katanin is required not only for flagellar autotomy to sever axonemal microtubules distal to the transition zone, but also to release the basal bodies from the transition zone so that as centrioles, they may participate in spindle positioning during mitosis (Rasi et al., 2009). This severing activity would occur proximal to the transition zone. Parker et al. recently demonstrated that this severing activity occurs following flagellar resorption and prior to cell division (Parker et al., 2010). Using anti-acetylated α-tubulin antibodies, immunofluorescence microscopy revealed the presence of two bright dots near recently divided cells, but which had not yet hatched from the mother cell wall. These dots were presumed to be the flagellar remnants left by excision from the basal bodies. Parker et al. then showed by electron microscopy that these bright dots are in fact the flagellar remnants that result from this excision event. To test whether KAT1 is required for excision between the basal bodies and the transition zone, we tested for the presence of flagellar remnants in newly divided wild-type, p60 amiRNA strains, pf15, and pf19 cells using anti-acetylated α-tubulin antibodies (Fig. 5B). In all cases the two dots which correspond to flagellar remnants were present. Therefore, excision proximal to the transition zone appears to be wild-type in p60 amiRNA strains as well as for pf19 and pf15 cells.

We also assessed whether pf19 or any of the amiRNA strains exhibited cell division defects by assessing culture density over several days. Cultures were inoculated with the same number of cells and then cells were counted over the course of several days. We could discern no significant differences between the amiRNA strains or the pf19 mutant compared to wild-type cells (Fig. 6). We also did not observe any clumpy phenotypes or cells with apparent cell division defects. Therefore, neither the pf19 mutation nor the knockdown of KAT1 expression results in defects in cell division or loss of viability. The double mutant pf15pf19 also has no cell cycle defects (not shown).

For motile eukaryotic cilia and flagella the central apparatus plays a crucial role in generating wild-type waveforms and beat frequency. While important and significant advances have recently been made in the field of ciliogenesis, we still do not know how the central pair of microtubules and associated projections are assembled in the axoneme. In Chlamydomonas several gene products have been identified which affect assembly/stability of specific projections on the central tubules; PF6 (Dutcher et al., 1984), Cpc1 (Mitchell and Sale, 1999), Hydin (Lechtreck and Witman, 2007), and FAP74 (DiPetrillo and Smith, 2010). However, mutations resulting in a complete failure of the central tubules to assemble are arguably the most likely to provide important information about a mechanism for the genesis of the central apparatus.

In Chlamydomonas, only four mutations have been reported that result in a central pairless phenotype (pf15, pf18, pf19, and pf20), all of which have paralyzed flagella (Adams et al., 1981; Dutcher et al., 1984). Of these mutations the sequence and localization of only the PF15 and PF20 gene products has been reported (Smith and Lefebvre, 1997; Dymek et al., 2004). The amino acid sequence of PF20 did not provide significant clues into the mechanism of central pair assembly. However, PF15 encodes the Chlamydomonas homologue of katanin p80. Data reported in this study reveal that the pf19 mutation resides within the KAT1 gene encoding the p60 subunit of katanin. These Chlamydomonas mutants imply that katanin plays a critical role in central apparatus assembly.

The conclusion that PF19 corresponds to KAT1 is based on discovery of a T to C transition in the KAT1 coding sequence, which converts leucine 277 to serine. This mutation was observed using multiple, different, primer pairs for amplification of genomic DNA as well as in cDNA generated from transcripts isolated from the pf19 mutant; furthermore, this was the only mutation observed for the entire coding sequence. Therefore, it is highly unlikely this same mutation occurred multiple times during PCR amplification. In addition, both the motility and central apparatus defects in the pf19 mutant were rescued upon transformation of pf19 cells with a plasmid containing only the KAT1 coding sequence. Therefore, wild-type PF19 corresponds to KAT1.

Our data argue that the pf19 mutation L277S renders p60 nonfunctional. L277 is highly conserved in p60 from Chlamydomonas to mammals and is within the AAA domain, only 28 amino acids upstream from the Walker A nucleotide binding motif. This AAA domain is required for nucleotide binding, oligomerization, nucleotide hydrolysis, and microtubule severing activity (Hartman and Vale, 1999). In addition, we have demonstrated using in vitro assays that the expressed pf19-p60 protein cannot sever microtubules. It is possible that the pf19-p60 protein is nonfunctional when expressed in bacteria. However, a wild-type construct has severing activity when generated, expressed and purified by the same methods. Therefore, it is highly unlikely that a technical problem occurred for the pf19-p60 construct that was not observed for wild-type p60.

It has been proposed that katanin is involved in microtubule severing at a position distal to the transition zone during stress-induced flagellar excision (Lohret et al., 1998; Lohret et al., 1999). Our studies indicate that the microtubule severing activity of katanin is not required for flagellar autotomy. When expressed in vitro, the pf19-p60 protein does not have microtubule severing activity. Yet, the pf19 mutant, like the pf15 mutant in which katanin p80 is mutated, retains the flagellar excision response to pH shock. A pf15pf19 double mutant is also wild type for the deflagellation response. We obtained the same phenotype when we reduced KAT1 expression using an amiRNA approach. These strains have central pairless flagella with no defects in flagellar autotomy. Collectively, these data strongly support the conclusion that the microtubule severing activity of p60 is not required for deflagellation and appear to be at odds with data previously published by others (Lohret et al., 1998; Lohret et al., 1999). These studies reported that an antibody generated against human katanin prevented flagellar excision. It is not clear why these particular antibodies prevent flagellar excision. There are two other potential microtubule severing proteins in Chlamydomonas, Kat2 (C_620017) and a predicted spastin-like protein. It is possible that the anti-human katanin antibodies recognize one of these potential severing enzymes in Chlamydomonas and that one of them is required for flagellar excision. Previous studies have indicated that centrin plays a role in the flagellar autotomy response (Sanders and Salisbury, 1994). Cells with mutations in centrin do not excise their flagella in response to low pH and antibodies generated against centrin also block flagellar excision.

It has also been proposed that katanin is required to sever axonemal microtubules proximal to the transition zone and distal to the basal bodies (Parker et al., 2010; Rasi et al., 2009). This activity results in the appearance of flagellar remnants that can be detected by immunofluorescence and electron microscopy (Parker et al., 2010). Since we show here that these remnants are present in the p60 amiRNA strains and pf19 mutant, and that the pf19 mutation abolishes microtubule severing activity, we conclude that microtubule severing by p60 is not the source, or at least not the sole source, of this severing activity.

Rasi et al. recently reported that katanin function is essential in Chlamydomonas (Rasi et al., 2009). This conclusion was based in part on the observation that genetic screens failed to yield mutants in the catalytic subunit of katanin and in part from their efforts to knock down KAT1 expression using an RNAi approach. In these experiments they were unable to obtain stable knockdown of KAT1 expression in wild-type cells, yet KAT1 expression was reduced in strains which lacked flagella. They concluded that reduction of KAT1 expression is lethal in wild-type flagellated cells, but that katanin is not essential for cells which lack flagella. Their working hypothesis is that the microtubule severing function of katanin is required to release the basal bodies from the transition zone to participate in spindle positioning in mitosis. This severing activity occurs between the basal body and transition zone (see also Parker et al., 2010). In cells lacking flagella, katanin would presumably not be required, although admittedly, the flagella-less cells used in their study still possess basal body attachments to the transition zone (in contradiction to their working model). As noted above, in this study we have shown that flagellar remnants that result from microtubule severing proximal to the transition zone are present in the pf19 and p60 amiRNA mutants. Therefore, this severing is likely not the result of p60 function.

The pf15, pf19, and pf15pf19 double mutants are all viable and show no sign of cell cycle defects. While these mutations are not null mutations, we have shown that the pf19 mutation abolishes microtubule severing activity. Therefore, the microtubule severing activity of katanin is not required for cell cycle progression. Our KAT1 knockdown strains are also viable with no cell cycle defects. One explanation for the difference in phenotype for the KAT1 knockdown strains produced by Rasi et al. compared to those reported here is the difference in method for reducing transcript levels (Rasi et al., 2009). We have tried numerous RNAi methods for reducing protein expression (unpublished results), and all of these methods produced a significant number of off-target effects. We have found that the amiRNA approach of Molnar et al. (Molnar et al., 2009) generates strains with stable knockdown of protein expression without these off-target effects (reviewed in DiPetrillo and Smith, 2010; Dymek et al., 2011; and unpublished observations for three additional axonemal proteins). It is also possible the antibody used by Rasi and colleagues (Rasi et al., 2009) for screening knockdown of expression is not specific (supplementary material Fig. S2).

Particularly noteworthy, a mutation which abolishes severing activity in the catalytic subunit of katanin (MEI-1) in C. elegans does not affect spindle assembly (McNally and McNally, 2011), whereas a null mutation causes defects in female meiotic spindle assembly (Mains et al., 1990). These authors propose a role for katanin in spindle assembly that does not require microtubule severing activity. This conclusion is in agreement with the data reported here that show p60 microtubule severing activity is not required for cell division in Chlamydomonas. Until a null mutation in KAT1 is identified, we cannot rule out the possibility that some additional function of katanin p60 unrelated to microtubule severing is required for cell cycle progression.

In addressing the question of whether p60 function is essential in Chlamydomonas it is important to recognize that there are two other potential microtubule severing proteins in Chlamydomonas, Kat2 (C_620017) and a predicted spastin-like protein. One possibility is that in the absence of p60, either of these two severing proteins may substitute for a specific p60 function, with the exception of central apparatus assembly. In Arabidopsis, the katanin most related in amino acid sequence to Chlamydomonas katanin does not appear to be essential. Multiple mutations have been isolated in the Arabidopsis KAT1 gene and all of these mutants are viable (Bichet et al., 2001; Burk et al., 2001; Webb et al., 2002; Bouquin et al., 2003; Nagawa et al., 2006). Moreover, Nagawa et al. generated a katanin mutant using a gene trapping technique; in this case the gene trap tag inserted in the second exon of the katanin coding sequence, most likely generating a null mutant (Nagawa et al., 2006). In Tetrahymena, null KAT1 mutants complete nuclear division but have defects in cytokinesis, forming chains of attached cells (Sharma et al., 2007). For these cells, cilia assemble which are shorter than normal and which lack the central apparatus. Also in Tetrahymena, Sharma et al. demonstrated that heterokaryon progeny cells lacking either KAT2 or SPA1 (spastin) had a normal cell morphology and gave rise to clones with normal vegetative growth rates (Sharma et al., 2007). In Trypanosoma bruceii, knock down of KAT1 expression does not result in defects in either nuclear or cell division (Casanova et al., 2009); however, it was not reported in these studies if the flagellum retained wild-type structures.

Based on our finding that mutations within either subunit of katanin result in paralyzed, central pairless flagella, it is clear that katanin function is required for assembly of the central apparatus in Chlamydomonas. The question now is, what is the mechanism by which microtubule severing activity contributes to the genesis of the central pair of microtubules? Particularly relevant to this question are studies in plant cells and in neurons. Comparisons have been made between the genesis of the parallel cortical array of microtubules in plants and those found in the axons of neurons (Gardiner and Marc, 2011). In both cell types, katanin appears to sever microtubules to produce shorter pieces which then translocate. In neurons, the severed microtubules translocate into the growing axon (Ahmad et al., 1999; Karabay et al., 2004; Yu et al., 2005; Yu et al., 2008; Baas and Sudo, 2010). In plant γ-tubulin rings bind to the side of existing microtubules to form branches, these microtubule branches are severed by katanin, which then form parallel microtubule arrays (Stoppin-Mellet et al., 2002; Wasteneys, 2002; Stoppin-Mellet et al., 2003; Stoppin-Mellet et al., 2007; Nakamura et al., 2010). Presumably the severed microtubules are stabilized by post-translational modifications or microtubule binding proteins to avoid complete disassembly from newly freed minus ends. The net result in both cases is the rapid reorganization of non-centrosomal microtubule arrays.

Perhaps central microtubule assembly in cilia is analogous to the assembly of these non-centrosomal microtubule arrays. In trypanosomes, McKean et al., 2003 have demonstrated that γ tubulin is specifically required for central apparatus assembly (McKean et al., 2003). In Chlamydomonas, γ tubulin has been localized to the transition zone (Silflow et al., 1999) where central apparatus assembly is believed to initiate. Based on work in Arabadopsis combined with studies in Chlamydomonas, trypanosomes, Tetrahymena and neurons, it is tempting to propose a model of central microtubule assembly where γ tubulin initiates microtubule growth from the lateral surface of an existing microtubule; katanin would then sever the new microtubules to form the central pair in the center of the axoneme. One prediction of this model in light of our results is that the pf15 and pf19 mutants may still have un-severed, nascent central tubules located near the base of the flagellum. To our knowledge there has not been a complete structural characterization of this region for either of these mutants.

Future studies of katanin function in Chlamydomonas and its role in central apparatus assembly will require additional information about the localization of both the p80 and p60 subunits during flagellar assembly as well as the identification of potential katanin interacting proteins. In addition, it will be important to know how this activity is coordinated with post-translational modifications known to occur on axonemal microtubules (reviewed by Gaertig and Wloga, 2008) and which may affect katanin activity, for example (Sudo and Baas, 2010). Finally, it would be of great interest to know in what gene the pf18 mutation resides. This is the only mutation which yields a central pairless phenotype and for which we have not yet identified the mutated gene. Most likely identification of the wild-type PF18 gene and gene product will reveal additional insights into central apparatus assembly.

Chlamydomonas reinhardtii strain A54-e18 (nit1-1, ac17, sr1,mt+) has wild-type motility and was obtained from Paul Lefebvre (University of Minnesota, St Paul, MN). The pf19 and pf15 strains were obtained from the Chlamydomonas Genetics Center (University of Minnesota, St. Paul, MN). A pf15pf19 double mutant strain was isolated from non-parental ditype tetrads resulting from crosses using pf19 and an insertional allele of pf15 (Dymek et al., 2004). Cells were grown in constant light in TAP media (Gorman and Levine, 1965). Generation of the pf19 rescue strains and p60 amiRNA strains is described below.

All cloning procedures are described in supplementary material Fig. S1; primers are listed in supplementary material Table S1. The PF19 gene was subcloned from BAC 27D3 (the BAC library (CRCCBa) was constructed by Paul Lefebvre and purchased from CUGI (Clemson University Genomics Institute). A 4 kb ClaI–ClaI fragment and a 2 kb ClaI–BamHI fragment were ligated into pBluescript (Stratagene) digested with ClaI and BamHI (final clone consists of −542 bp to +1248 bp of the KAT1 gene). This clone, pBCp60, was used to construct a GFP–p60 clone. Our goal was to engineer the GFP coding sequence at the most 5′ location in the p60 gene (supplementary material Fig. S1). To do this, we first used a series of PCR reactions to engineer convenient restriction sites for cloning the gene encoding GFP into the 5′ end of the gene encoding p60. A first set of primers engineered an XhoI site 1022 bp upstream of the translational start site and an EcoRI site immediately upstream of the start site (primers P60Xho and P60NEco). These primers were used to amplify the 5′ end of the gene by PCR using BAC 27D3 as a template. The resulting product (PCR1) was cloned into pCR2.1 (Invitrogen). A second set of primers engineered an XbaI site immediately downstream of the start site and a NotI site just 3′ to the endogenous MluI site (primers P60NXba and P60MluNotI). These primers were used to amplify this region by PCR using pBCp60 as a template. The amplified product (PCR2) was cloned into pCR2.1. A third set of primers was used to engineer an EcoRI site 3′ of the GFP start and an XbaI site just 5′ of the GFP stop (primers P60GFPNEco and P60GFPXba) using pCR-GFP (generously provided by Karl Lechtreck) as a template. The amplified product, PCR3, was cloned into pCR2.1. For all plasmids, PCR1, PCR2, and PCR3, the PCR product inserts were excised using the restriction enzymes corresponding to the engineered sites and gel purified. pBluescript was digested with XhoI and NotI, and in a trimolecular reaction, PCR1, 2 and 3 were ligated together into the digested vector; the resulting plasmid was named 5′GFP-p60pBlue. This ligation created two additional amino acids, serine and arginine, between GFP and the amino terminus of the p60 protein. After confirming the entire sequence and reading frame, 5′GFP-p60pBlue was digested with XhoI and MluI to remove the insert; this fragment is named 5′GFP. 5′GFP was then ligated into pBCp60 (digested with XhoI and MluI) and transformed into TOP10 cells (Invitrogen); the resulting construct is named GFP-pBCp60. The plasmid insert was sequenced to confirm the GFP insertion and reading frame. All sequencing was performed using Big Dye Terminator 3.1 (Applied Biosystems) and analyzed on an ABI Prism 3100 Genetic Analyzer (Applied Biosystems).

For bacterial expression of wild-type and pf19 p60, reverse transcription was performed using Superscript III (Invitrogen) using the p60-reverse primer and RNA isolated from wild-type or pf19 cells as template. After two rounds of amplifications using nested primers (round 1: p69 forward to p60 reverse; round 2: p75 forward to p60 reverseN), the final PCR product was cloned into pCR2.1 (Invitrogen) and named p60cDNA-pCR. To express the cloned cDNA as a GST fusion, we engineered a BamHI site at the 5′ end of p60; p60cDNA-pCR was used as a template and amplified using M13forward and p60-BaM primers. The resulting product was cloned into pCR2.1 (p60Bam-pCR). The insert was removed from p60Bam-pCR by digestion with BamHI and EcoRI and ligated into pGex2T (GE Healthcare), thereby tagging p60 with GST at the 5′ end. The insert was sequenced for both wild-type and pf19 clones to confirm reading frame and were then transformed into Rosetta Gami B (DE3) pLysS cells (Novagen) for expression.

High efficiency transformation was achieved using the glass bead procedure of (Kindle, 1990). To test BAC clones and plasmids for rescue of mutant phenotypes, pf19 cells were co-transformed with 1–3 µg of DNA and 1 µg of the APHVIII gene, conferring paromomycin resistance. Cells carrying integrated DNA were selected on 10 µg/ml paromomycin-TAP plates. Colonies were picked into TAP and screened for motility.

Artificial microRNAi constructs were made according to Zhao and colleagues, and others (Zhao et al., 2009; Molnar et al., 2009; DiPetrillo and Smith, 2010) with the following modifications. The Web MicroRNA Designer (http://wmd3.weigelworld.org/cgi-bin/webapp.cgi (Ossowski et al., 2008) identified possible amiRNA sequences forp60 and designed forward and reverse oligonucleotides flanked by SpeI sites. Two amiRNA sequences were chosen that target p60 cDNA; no. 1: 5′-TAAGAAGGTGCACGATGTCTC-3′ targets nt961–980, no. 2: 5′-TTCATAGTACGTGTATCCCGC-3′ targets nt79–98. Annealed oligonucleotides were ligated into SpeI-digested pChlamiRNA3int plasmid, which contains the APHVIII gene. Colony PCR was used to screen bacterial transformants to confirm the correct orientation of the insert. Note regarding amiRNA in Chlamydomonas: the frequency with which we obtained knockdown of p60 expression  =  1/704. We typically use two to three constructs for each gene of interest. The frequency of knockdown ranges from 0 per 2000 to 1 per 38 transformants (DiPetrillo and Smith, 2010; Dymek et al., 2011) and four other genes (unpublished data). The source of this variability is unknown. However, once knockdown of expression is achieved, the mutants are stable and exhibit no obvious off-target phenotypes.

For each strain, 5 µg of genomic DNA was digested with KpnI and size fractionated on a 0.7% agarose gel. The DNA was transferred to MagnaGraph nylon membrane (GE Healthcare) and UV crosslinked using the Stratagene Stratlinker. The membrane was probed with the APHVIII, PF20, PF15 or KAT1 genes which were each labeled with 50 µCi [α-³²P]dCTP using the Random Primed DNA Labeling Kit (Roche). The membrane was exposed to a phosphor screen and scanned with a Typhoon 9200 Imager (GE Healthcare).

Total RNA was isolated from cells at 0 minutes and 45 minutes post-deflagellation. Poly(A)⁺ RNA was isolated from 1 mg of total RNA using the Oligotex mRNA Midi Kit (Qiagen). Poly(A)⁺ RNA (10 µg) was fractionated on a 1% agarose gel containing formaldehyde and transferred to Hybond-N+ nylon membrane (GE Healthcare). The membrane was probed with wild-type p60cDNA which was labeled with 50 µCi of [α-³²P]dCTP using the Random Primed DNA Labeling Kit (Roche). The ribosomal S14 gene was also labeled and used to probe the membrane as a loading control. The membrane was then exposed to a phosphor screen and scanned with a Typhoon 9200 Imager (GE Healthcare). Densitometry on the scanned image was performed with ImageJ (Rasband, W.S., ImageJ, US National Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2011). Relative expression levels were normalized using S14 transcript levels. Northern blots were repeated for a minimum of three different experiments and we obtained consistent results.

Axonemes were isolated as described previously (Dymek and Smith, 2007). Axonemes (10 µg) or whole cells (1×10⁶ cells) were resolved by SDS-PAGE using a 7% polyacrylamide gel. Following electrophoresis, the samples were transferred to PDVF membrane (Millipore). Membranes were blocked in 5% milk in TBST for 1 hour at room temperature and then incubated with primary antibody diluted in TBST for two hours at room temperature [anti-PF20 1:10,000, affinity purified anti-CaM-IP3 1:1000, anti-RSP1 (generously provided by Joel Rosenbaum, Yale University) 1:10,000, anti-GFP290 (AbCam) 1:1000, anti-p60 serum (generously provided by Lynne Quarmby, Simon Fraser University) 1:1000 or affinity purified anti-p60 (generously provided by Lynne Quarmby) 1:500]. Blots were washed and incubated with secondary antibody diluted in TBST for 30 minutes at room temperature [anti-rabbit-HRP (GE Healthcare) 1:30,000 or anti-mouse-HRP (Thermo Scientific) 1:10,000]. Blots were washed and developed using the ECL Plus Western Blotting Detection System (GE Healthcare).

A loopful of cells was resuspended in 1 ml TAP medium and counted with a hemocytometer. An equal number of cells (9.4×10⁷ cells) were added to 250 ml TAP bubblers (3.76×10⁵ cells/ml). Cells were bubbled in constant light for 3 days. At 24, 48, and 72 hours, samples were collected and counted using a hemocytometer. The results were plotted in MS-Excel.

For analysis of flagellar defects, axonemes from WT and mutants of interest were prepared for thin-section electron microscopy. Specimens were fixed with 1% glutaraldehyde and 1% tannic acidin 0.1 M sodium cacodylate, postfixed in 1% osmium tetroxide, dehydrated in a graded series of ethanol, and embedded in LX112 resin. Uniform silver-gray sections were mounted on Formvar-coated, carbon-stabilized copper grids, stained with uranyl acetate and Reynolds lead citrate, and examined at 100 kV in a transmission electron microscope (JOEL JEM-1010). To quantify central pair defects, transverse sections of axonemes were classified as having two, one or no central microtubules present. Axonemes with electron dense material were considered a separate category.

Wild-type and pf19 p60cDNA in the pGex2T plasmid were transformed into Rosetta Gami B (DE3) pLysS cells (Novagen) and grown to an OD₆₀₀ of 0.6 at 37°C, then induced overnight at 30°C with 1 mM IPTG. Wild-type or pf19 p60-GST was purified using GST-Bind resin using non-denaturing conditions according to the manufacturer’s methods (Novagen). Protein concentration was determined spectrophotometrically. Protein in peak fractions was resolved by SDS-PAGE using a 7% polyacrylamide gel and visualized by Coomassie Blue staining. For microtubule severing assays, ATP was added to a final concentration of 1 mM to both wild-type and pf19 p60–GST peak fractions. Microtubules were polymerized in vitro (Cytoskeleton) and resuspended to a concentration of 0.25 mg/ml in NaLow (10 mM Hepes, 5 mM MgSO₄, 1 mM DTT, 0.5 mM EDTA, and 30 mM NaCl, pH 7.5). Flow chambers were made on microscope slides coated with poly-lysine (1:600, Sigma) using double stick tape with coverlips placed on top. Microtubules (10 µl) were pipetted into the chamber and rinsed with 40 µl of NaLow. Then 40 µl (∼10 µg) of WT p60–GST, pf19 p60–GST or NaLow was flowed through the chamber, followed by 40 µl of WT p60–GST+ATP, pf19 p60–GST+ATP or NaLow+ATP (1 mM ATP). For all samples, 40 µl of NaLow/ATP was then flowed through the chamber. Images were observed on an Axioskope 2 microscope (Zeiss, Inc.) equipped for dark-field optics including a Plan-Apochromate 40× oil immersion objective with iris and ultra dark-field oil immersion condenser. Images were captured at 0 and 20 minutes using a silicon-intensified target camera (VE-1000 SIT; Dage-MTI, Inc.) and converted to digital images using Labview 7.1 software (National Instruments, Austin, TX). Image J (Rasband, W.S., ImageJ, US National–Institutes of Health, Bethesda, Maryland, USA, http://rsb.info.nih.gov/ij/, 1997–2011) was used to quantify the amount of microtubules in each image. For each set of 0- and 20-minute images, the threshold levels were set to differentiate microtubules from background and then equalized between the two images. Mean grey values were then measured and compared between corresponding sets of 0- and 20-minute time points. In all cases the mean grey value decreased after 20 minutes. The difference in mean grey value is expressed as a percent decrease. For each experimental group, at least 10 image sets were analyzed.

Immunofluorescence was performed according to Dymek and colleagues (Dymek et al., 2006). Briefly, dividing cells were adhered to poly-lysine-coated coverslips, fixed with methanol for 15 minutes and allowed to dry for 15 minutes. Coverslips were blocked for 1 hr in 5% BSA/PBS and incubated overnight in anti-acetylated α-tubulin antibodies (Sigma; clone 611B1, 1:100 in 1% BSA/PBS). Coverslips were washed with PBS and incubated for 1 hour in anti-mouse (Texas Red) secondary antibody (Vector Labs, 1:500 in 1% BSA/PBS) and washed with PBS. Coverslips were placed on microscope slides with a drop of prolong anti-fade (Invitrogen) and allowed to dry overnight in the dark.

The authors are grateful for expert technical support for this work from Natalie Vajda, and thank Pete Lefebvre, Win Sale and Roger Sloboda for critically reading the manuscript.