Mutations in α-tubulin promote basal body maturation and flagellar assembly in the absence of δ-tubulin

We have isolated suppressors of the deletion allele of δ-tubulin, uni3-1, in the biflagellate green alga Chlamydomonas reinhardtii. The deletion of δ-tubulin produces cells that assemble zero, one or two flagella and have basal bodies composed primarily of doublet rather than triplet microtubules. Flagellar number is completely restored in the suppressed strains. Most of the uni3-1 suppressors map to the TUA2 locus, which encodes α2-tubulin. Twelve independent tua2 mutations were sequenced. Amino acids D205 or A208, which are nearly invariant residues in α-tubulin, were altered. The tua2 mutations on their own have a second phenotype - they make the cells colchicine supersensitive. Colchicine supersensitivity itself is not needed for suppression and colchicine cannot phenocopy the suppression. The suppressors partially restore the assembly of triplet microtubules. These results suggest that the δ-tubulin plays two roles: it is needed for extension or stability of the triplet microtubule and also for early maturation of basal bodies. We suggest that the mutant α-tubulin promotes the early maturation of the basal body in the absence of δ-tubulin, perhaps through interactions with other partners, and this allows assembly of the flagella.

δ-Tubulin is the fourth member of the tubulin superfamily and was first identified in Chlamydomonas (Dutcher and Trabuco, 1998). It is found in the genomes of other organisms with triplet microtubules (Chang and Stearns, 2000;Vaughan et al., 2000;Smrzka et al., 2000) but it is not present in the genomes of D. melanogaster (Adams et al., 2000;Celniker et al., 2002) or C. elegans (The C. elegans Sequencing Consortium, 1998). Cells with the δ-tubulin gene deleted (uni3-1) have flagellar, basal body and cell division defects.

303
We have isolated suppressors of the deletion allele of δ-tubulin, uni3-1, in the biflagellate green alga Chlamydomonas reinhardtii. The deletion of δ-tubulin produces cells that assemble zero, one or two flagella and have basal bodies composed primarily of doublet rather than triplet microtubules. Flagellar number is completely restored in the suppressed strains. Most of the uni3-1 suppressors map to the TUA2 locus, which encodes α 2-tubulin. Twelve independent tua2 mutations were sequenced. Amino acids D205 or A208, which are nearly invariant residues in α-tubulin, were altered. The tua2 mutations on their own have a second phenotype -they make the cells colchicine supersensitive. Colchicine supersensitivity itself is not needed for suppression and colchicine cannot phenocopy the suppression. The suppressors partially restore the assembly of triplet microtubules. These results suggest that the δ-tubulin plays two roles: it is needed for extension or stability of the triplet microtubule and also for early maturation of basal bodies. We suggest that the mutant α-tubulin promotes the early maturation of the basal body in the absence of δ-tubulin, perhaps through interactions with other partners, and this allows assembly of the flagella.
Electron microscopy of uni3-1 cells showed a loss of the C tubule along the length of the basal body (Dutcher and Trabauco, 1998). Using gene silencing, a similar lack of the C tubule was observed in Paramecium tetraurelia (Garreau de Loubresse et al., 2001).
The uni3-1 mutant cells have a heterogeneous flagellar phenotype. 50% of uni3-1 cells are aflagellate, 25% are uniflagellate and 25% are biflagellate. A pedigree analysis of uni3-1 mitotic cells suggests that the ability to assemble a flagellum from a basal body lacking δ-tubulin requires maturation of the basal body through multiple cell cycles (Dutcher and Trabuco, 1998). These observations raise several important questions about the roles of δ-tubulin and the C tubule. First, if the C tubule is needed to assemble flagella, why are uni3-1 cells able to assemble flagella in a sizeable fraction of the cells? Second, if δ-tubulin is needed to assemble the C tubule, why do sperm basal bodies in Drosophila, which lack δ-tubulin, have C tubules?
To address these questions, we screened for suppressors of the uni3-1 deletion allele in Chlamydomonas. We report here the isolation and characterization of 12 tua2 alleles (encoding α2-tubulin) that suppress the flagellar assembly defects of the uni3-1 allele. Only two different mis-sense mutations in TUA2 were found among these 12 independent alleles. We observed that these mis-sense mutations in α-tubulin can partially restore C-tubule assembly. This result suggests that δ-tubulin is not strictly required for C-tubule assembly. We suggest that the mutant α-tubulin alters the ability of the basal body to undergo maturation, perhaps through interactions with other partners.

Materials and Methods
Cell culture and genetic analysis Culture conditions and media were as described previously (Lux and Dutcher, 1991) and nitrite medium was as described by Fernández et al. (Fernández et al., 1989). Standard matings were as described in Dutcher (Dutcher, 1995). Diploid strains were selected as described (Palombella and Dutcher, 1998) using the linked nit2-1 and ac17 mutations in trans for selection. The pyr1and tub2-1 (formerly col R 4) mutant strains were obtained from the Chlamydomonas Genetics Center (Duke University, Durham, NC) and the tua1-1and apm1-1 mutations were obtained from P. Lefebvre (University of Minnesota, Minneapolis, MN). All drugs were added after autoclaving. Stock solutions of colchicine were made in ethanol and stored in the dark. Medium with colchicine was incubated in yellow lucite boxes or yellow Saran Wrap to prevent breakdown of the colchicine (Palombella and Dutcher, 1998). Concentrations of colchicine ranged from 0.10 mM to 4.5 mM. Pyrythiamine was used at 1 µM. 5fluoroindole was used at 7 µM and assayed in yellow lucite boxes to prevent its breakdown (Palombella and Dutcher, 1998). Cells were assayed on medium with oryzalin and vinblastine from 1 µM to 15 µM. Oryzalin was a gift from Eli Lily Laboratories (Indianapolis, IN). Stock solutions of these drugs were solubilized in 100% dimethyl sulfoxide (DMSO). Measurements of cell sizes were performed as described (Preble et al., 2001).
Isolation of suppressors of uni3-1 Two separate screens were used to isolate cells that regained the ability to assemble two flagella on most cells. In the first screen, 180 single uni3-1 colonies were grown for 3 days on R medium and then subjected to ultraviolet (UV) irradiation for 45 seconds (Lux and Dutcher, 1991). The irradiated colonies were allowed to recover in the dark for 24 hours and were then placed individually into 20 ml R medium in 150 mm culture tubes. Cells that swam to the upper 5 ml were transferred. This transfer was repeated seven times. 10 µl of supernatant was plated to solid R medium, 15-20 individual colonies were picked into 2.5 ml R medium and assayed for swimming ability, and 48 suppressed strains were recovered. In the second screen, 200 single uni3-1 colonies were grown for 3 days on R medium and then subjected to γ-irradiation for 30 minutes from a 137 Cs source (498 Roentgen/minute). The same enrichment protocol described above was followed and 15 colonies were recovered.
The first 18 strains were mated to a wild-type parent (CC124) and the segregation of swimming phenotypes was assayed. All strains contained extragenic suppressors that were easily separated from the uni3-1 mutation by recombination. The remainder of the strains were mated to a pyr1 strain because we had determined that most suppressors were linked to this mutation. This provided a quick assay for placing strains into recombination groups. Further classification of the strains was based on complementation and recombination tests using suppression and colchicine supersensitivity as assays. Diploid strains that were homozygous for the uni3-1 allele were used for suppression assays and diploid strains that were homozygous for the UNI3 allele were used for colchicine-supersensitivity assays.

Isolation of revertants of tua2 alleles
Because the tua2 alleles confer supersensitivity to colchicine, it was possible to select revertants on media with low concentrations of colchicine. Revertants of the supersensitivity phenotype of tua2-1, tua2-2, tua2-3 and tua2-4 strains were selected. Approximately 10 8 cells were placed on 1.75 mM and 2.2 mM colchicine plates without mutagenesis and grown for 2-3 weeks. A single revertant was saved from each plate and retested on media with a range of colchicine concentrations. Each strain was mated to a pyr1 parent and the segregation of the reversion was determined. All events in the five strains were inseparable from the supersensitive phenotype. The revertants were named tua2-1R1, tua2-2R1, tua2-3R1 and tua2-4R1. Each revertant was assayed for growth phenotypes and flagellar assembly. Each revertant strain was mated to uni3-1 cells to determine whether suppression of the flagellar assembly phenotype of uni3-1 was also reverted. From each of the crosses tua2-1R1×TUA2, tua2-2R1×TUA2, tua2-3R1×TUA2 and tua2-4R1×TUA2, 12 meiotic progeny were subsequently mated to uni3-1 pyr1 mt + or uni3-1 pyr1 mtstrains to verify the failure to suppress the flagellar phenotype of uni3-1 cells and to verify new synthetic phenotypes.

Drug sensitivity tests
To determine the degree of colchicine supersensitivity, cells were grown to densities of ~2×10 6 cells ml -1 . The cells were diluted 1000fold and then plated on solid R medium to determine the number of cells that could form colonies. The data are displayed graphically as the percentage of viable cells against the concentration of colchicine.

Protein preparations
Cell bodies were separated from axonemes by centrifugation after deflagellation with dibucaine (King, 1995) and were passed twice through a French Press at 4000 psi. Insoluble material was removed by centrifugation at 34,000 g for 60 minutes. Axonemal proteins were prepared as described (King, 1995), with the addition of Nonidet P-40 to 0.1%. Protein concentrations were determined using the BCA reagent (Sigma, St. Louis, MO) with bovine serum albumin (BSA) as a standard. Denaturing electrophoresis was performed on 10% polyacrylamide vertical minigels Mini V 8•10 (Gibco BRL/Life Technologies) according to Laemmli (Laemmli, 1970). Gels were transferred to nitrocellulose by using the Gibco BRL/Life Technologies Mini-V 8•10 Vertical Gel Electrophoresis system for 1 hour at 100 V as recommended by the manufacturer. After transfer, the nitrocellulose blots were first blocked with 3% BSA in TBS (Bollag et al., 1996) and subsequently probed with a rabbit polyclonal antibody to α-tubulin (kindly provided by D. Cole) (Cole et al., 1998) at 1:10,000 in 1.5% BSA in TBS, with a monoclonal antibody to αtubulin (DMA1) (Research Diagnostics, Flanders, NJ) at 1:1000 in 10% goat serum in PBS, and a monoclonal antibody to acetylated αtubulin (LeDizet and Piperno, 1986). A secondary goat anti-rabbit polyclonal antibody or a secondary goat anti-mouse polyclonal antibody (Biorad, Hercules, CA) conjugated to alkaline phosphatase was used at a 1:2000 dilution to detect primary antibody binding.
Threading Chlamydomonas sequence onto the crystal structure of tubulin The amino acid sequences of Chlamydomonas αand β-tubulin (accession number A523928 and UKBM, respectively) were mached to the crystal structure of bovine tubulin (accession number 1JFF) using the Swiss-Model Automated Comparative Protein Modeling Server (http://www.expasy.ch/swissmod/) in the first approach mode. The resulting structure was viewed by SwissPdb Viewer version 3.7 (http://www.expasy.ch/spdbv/).

Electron microscopy
Cell pellets were prepared for electron microscopy by high-pressure freezing followed by freeze substitution as described (Preble et al., 2001;O'Toole et al., 2003). Serial thin sections, of 50-60 nm were viewed in a Philips CM10 electron microscope operating at 80 kV or 100 kV. Using the microscope's goniometer stage, the sections were tilted to an angle that permitted the basal body microtubules to be viewed in cross section. To allow tilting along the appropriate axis, a rotating specimen holder was used.

Suppressors of uni3-1 are linked to TUA2
We generated suppressors of the uni3-1 deletion allele by treating uni3-1 cells with UV or γ-irradiation. Because 25% of the cells in uni3-1 cultures are biflagellate and can swim effectively, suppressors could be identified as cultures with an increased proportion of swimming cells. Sixty-one suppressed strains were isolated from 254 independent cultures. Forty-six of the suppressor alleles are linked to the TUA2 locus (α2-tubulin) on linkage group IV based on mapping to pyr1, which confers resistance to pyrithiamine (Ranum et al., 1988;Harris, 1989) (Table 1). Eleven of the remaining 16 suppressors map to linkage group II and show no phenotypes beyond suppression of uni3-1. The remaining four suppressors do not map to known tubulin loci and also convey supersensitivity to colchicine. Further analysis of the remaining recombination groups will be reported elsewhere. Each of the 46 strains showed complete suppression of the flagellar assembly defect: more than 92% of the suppressed cells had two flagella, 3% of the cells had more than two flagella and the remainder were aflagellate (n=200). Eight of the 46 alleles were tested for dominance and were found to be recessive for suppression of the flagellar defects of the uni3-1 allele (Table 2). These alleles did not suppress the uniflagellate phenotypes of the uni1 or uni2 alleles (data not shown) (Huang et al., 1982;Dutcher, 1986; Dutcher and  Trabuco, 1998). When the 46 suppressors were examined in UNI3 strains, they formed functional, motile flagella. In addition, they showed supersensitivity to the tubulin-binding drug, colchicine (Fig. 1). Colchicine supersensitivity is recessive to the wild-type allele in 15 strains examined at colchicine concentrations ranging from 0.1 mM to 1.0 mM (Table 3). These mutations do not confer supersensitivity to the other microtubule-affecting drugs oryzalin, isophenyl carbamate or vinblastine or confer resistance to Taxol (data not shown). Supersensitivity to antimicrotubule-agents can be conferred by mutations in αor β-tubulin as well as other genes (Umesono et al., 1983;Hoyt et al., 1990;Jung et al., 1990;Stearns et al., 1990;Chen et al., 1994). The genetic map position and colchicine-supersensitivity phenotype conferred by these mutations suggested that these suppressors carried mutations in TUA2.

Suppressors of uni3-1 are missense mutations in TUA2
In order to determine whether these suppressors are mutated in TUA2, we isolated genomic DNA from 12 of the 46 suppressor strains and amplified the TUA2 gene by PCR. The isolated PCR products were then directly sequenced. TUA2 DNA from three control strains (137c mt + , uni3-1 and one unlinked suppressor) was amplified and sequenced.
The TUA2 sequences from mutant and control strains contain 12 base changes in intron 1 (GenBank Accession number AY182002) compared with the sequence deposited in GenBank (Accession number M11448) (Silflow et al., 1985) and modified in James (James et al., 1993). These changes include nucleotide substitutions and insertions. The parental strain used in our laboratory (137c) is derived from a different isolate of Chlamydomonas than that used in the previous studies of TUA2 (Harris, 1989), and so we attribute the intron 1 changes to genetic drift. No additional base changes were found in the three control sequences. However, when we compared the sequences of the pyr1-linked suppressor strains with the parental strain, we found one of two nucleotide substitutions in exon 3.
Nine strains were found to have an A instead of a G at position 1210, which results in the amino acid change D205N. Three strains have an A instead of a G at position 1219, which results in the amino acid change A208T. We verified that these changes were not due to errors made by Taq DNA polymerase by sequencing the PCR products from multiple independent reactions. In addition, we sequenced 125 bps of TUA2 DNA (including nucleotide 1210) from 12 meiotic progeny from a cross of tua2-1×TUA2 and found that the colchicine phenotype cosegregates with the base change at position 1210 in all progeny tested. Because the suppressors are clearly TUA2 alleles, they have been named tua2-1 through to tua2-12 (Table 1).
, diploid cells that grow on 1.0 mM colchicine medium and with recessive supersensitivity phenotype; -, diploid cells that fail to grow on 1.0 mM colchicine medium and with allelic mutations; NT, not tested. the C tubule have been found at the distal end of the basal bodies from uni3-1 cells using dual-axis electron tomography of basal bodies (O'Toole et al., 2003). We observed a modest increase in the frequency and length of C tubules in uni3-1; tua2-6 basal bodies compared with uni3-1 basal bodies. The tendency of uni3-1 basal bodies to form a second stellate fiber array (normally located in the transition zone between the basal body and the axoneme, in the proximal half of the basal body) was retained in tua2-6; uni3-1 basal bodies (O'Toole et al., 2003).
In parallel with the above study, we examined serial sections of tua2-6; uni3-1 cells to quantify the rescue by the tua2-6 suppressor. We found that basal bodies had acquired up to nine triplet microtubules (one example in Fig. 2A-F). One of the 14 serially sectioned basal bodies had nine triplet microtubules whereas the remainder had one (n=2), two (n=3), three (n=5) or four (n=4) triplets. Thus, suppression of the flagellar assembly defect does not appear to arise from the assembly of basal bodies with a full complement of nine triplet microtubules. Suppression might result from the limited restoration of triplet microtubules within a basal body or from the altered properties of the mutant α-tubulin. As observed by tomography (O'Toole et al., 2003), stellate fibers were present in the interior of proximal regions of the basal bodies. We also examined serial sections of 13 tua2-6; UNI3 basal bodies and found no differences from wild-type cells. Serial sections of two tua2-6; UNI3 cells are shown in Fig. 2G-N. Structures along the basal body are indistinguishable from wild-type basal bodies.
The cleavage furrow defect of uni3-1 is not suppressed by the tua2 alleles In wild-type populations, two recently divided daughter cells have nearly equal sizes (Preble et al., 2001). Multiple mutant alleles in the gene for ε-tubulin have defects in the placement of the spindle and cleavage furrow (Ehler et al., 1995;Preble et al., 2001;Dutcher et al., 2002). Previous observations of uni3-1 cells suggested that their cleavage furrow is not placed normally (Dutcher and Trabuco, 1998). To confirm this observation, the areas of pairs of daughter cells that had recently divided were measured for wild-type, uni3-1, tua2-2, tua2-6, tua2-2; uni3-1 and tua2-6; uni3-1 cells, and the ratio between the large and the small cell was calculated. For wildtype cells this ratio is close to 1. The distribution of the ratio of large to small uni3-1 daughter cells is broader than for wildtype cells (Fig. 3A,B). The distribution of the ratio of large to small daughter cells for tua2-2; uni3-1 and tua2-6; uni3-1 is similar to that for uni3-1 cells (Fig. 3B-D). The distributions of the ratios of tua2-2 and tua2-6 cells (Fig. 3E,F) resemble wild-type cells. Although the tua2-2 and tua2-6 alleles suppress the flagellar defects of the uni3-1 allele completely, they do not suppress the defects in cleavage furrow placement.

Isolation of intragenic revertants of tua2
Does reversion of the colchicine-supersensitivity phenotype of the tua2 alleles also result in reversion of the ability to suppress the flagellar phenotype of uni3-1? To answer this question, we isolated four spontaneous revertants of several tua2 alleles with the D205N amino acid change that enables them to form colonies on 1.0 µM colchicine. Each of the revertants was inseparable from the colchicine-supersensitive phenotype in more than 150 tetrads for each of the four revertant strains (Table 4). These results suggested that the reversion event was likely to be intragenic. No obvious flagellar or cell division phenotypes were observed in the haploid revertant strains. The population doubling times for tua2-6 and tua2-12 are similar to that for wild-type cells at 25°C (7.5 hours, 7.6 hours and 7.5 hours, respectively), and the population doubling times for tua2-6; uni3-1, tua2-12; uni3-1 and uni3-1 are similar (8.1 hours, 8.0 hours, and 8.2 hours respectively). Diploid cells that are heterozygous for intragenic revertants with a tua2-1 or a TUA2 allele were constructed and analyzed. The diploid strains heterozygous for TUA2 have wild-type levels of sensitivity to colchicine and no flagellar phenotypes (Table 5), whereas the diploid strains heterozygous for tua2-1 remain supersensitive to colchicine. Thus, the revertant alleles are recessive to the colchicine-supersensitive allele suggesting that they have a loss or a reduction in function. In addition, they do not show any interaction with the tua1-1 allele in α1-tubulin in heterozygous diploid strains. The phenotypes of the revertants in combination with the uni3-1 allele were examined. The tua2-1R1, tua2-2R1 and tua2-4R1 alleles lost the ability to suppress the flagellar phenotype of uni3-1 cells. The tua2-2R1 allele was recessive to the tua2-2 allele for suppression of the flagellar defect of uni3-1, which again suggests that the revertant allele has a loss-of-function phenotype. The tua2-3R1 allele had no phenotype of its own, but did show a phenotype in combination with the uni3-1 allele. The uni3-1; tua2-3R1 double mutant was aflagellate at all temperatures examined when grown vegetatively but was flagellated when grown as gametic cells. Ninety-five percent of the cells had two full-length flagella, and wild-type motility. Thus, tua2-3R1 acts as an enhancer of the Uni3 flagellar phenotype. This interaction with the uni3-1 allele is partially dominant. In tua2-3R1; uni3-1/TUA2; uni3-1 cells, 75% of the cells are aflagellate and the remainder are uniflagellate, which suggests that a mutant protein is present and interfers with the wild-type α-tubulin. The diploid phenotype suggests that the presence of three wild-type copies of α-tubulin allele can compete with the mutant protein.  Sensitive, cells fail to grow on 1 mM colchicine medium; +, cells grow on 1 mM colchicine medium. δ-tubulin suppressors

Journal of Cell Science 117 (2)
The TUA2 genes in the four revertants were sequenced. In the tua2-3R1 allele, we found a single base change in addition to the original lesion. The base change C1115T is predicted to cause the amino acid substitution P173L. We found a complicated molecular lesion in the tua2-2R1 allele in addition to the original tua2-2 mutation. A nine-base insertion accompanied by a two-base deletion appears to have occurred and is probably the result of duplicated adjacent DNA. The tua2-2R1 allele probably encodes an altered and truncated α2tubulin; amino acids 334-339 in wild-type α2-tubulin would be missing and the last 106 amino acids would be replaced by 29 'random' amino acids in addition to the original lesion D205N.
The protein encoded by the tua2-2R1 allele would probably not be functional in microtubule-based structures, even if it was stable in the cell. The predicted molecular mass of this mutant protein is ~36 kDa, which is significantly smaller that the 49 kDa predicted for the wild-type polypeptide. Flagellar proteins from tua2-2R1 cells and wild-type cells were isolated and analyzed by SDS-PAGE followed by immunoblotting and probed with three anti-α-tubulin antibodies (Cole et al., 1998;Blose et al., 1984;LeDizet and Piperno, 1986). For the DMA1 antibody, the epitope is the C-terminus of the protein, that of the 6-11-B1 antibody is the N-terminus of the protein, whereas that of the remaining antibody is unknown. No other than wildtype α-tubulin band was detected with either antibody (data not shown). We repeated the same procedure with cell-body lysates and obtained the same result. Loading as much as 50 µg protein, we were unable to detect any band apart from the one corresponding to wild-type α-tubulin (Fig. 4); wild-type αtubulin was detected with as little as 2 µg total protein loaded (data not shown). We, therefore conclude that, in the revertant tua2-2R1 allele, no detectable α2-tubulin is produced. Two other revertant alleles, tua2-1R1 and tua2-4R1, were examined and no base changes other than the original lesion were found in the coding sequence. Both tua2-1R1 and tua2-4R1 appear to be intragenic (Table 4). We suspect that these revertants represent mutations in the regions that flank TUA2, perhaps altering TUA2-regulatory sequences. This conclusion would be consistent with the absence of protein resulting in the reversion of the colchicine-supersensitivity phenotype. However, we cannot exclude the possibility that these revertants contain a mutation in a gene tightly linked to TUA2.
Treatment with microtubule destabilizers does not rescue the flagellar defects of uni3-1 Colchicine supersensitivity suggests that the microtubules in tua2 strains have decreased stability. We tested the hypothesis that changes in microtubule stability suppress the flagellar phenotypes of uni3-1 cells in two separate experiments. We asked whether low concentrations of colchicine or oryzalin can phenocopy the suppression of the flagellar defects of uni3-1cells. We also tested whether mutations that cause resistance to microtubule-destabilizing drugs obviate the suppression (see following section)? Wild-type and mutant cells (2×10 6 cells Histograms display the number of cells with a given large-/small-cell area ratio. uni3-1 cells produce daughters of unequal sizes, suggesting a defect in the placement of the cleavage furrow. This defect is not suppressed by either of the tua2 alleles. tua2 strains show wild-type ratios, which indicates that the positioning of the cleavage furrow is not affected by the mutations. ml -1 ) were treated for 10 hours with colchicine (0.01 mM to 2.0 mM) and oryzalin (0.1 µM to 15 µM). This range of drug concentration had little or no effect on viability after 10 hours. There was no change in the number of cells with zero, one and two flagella in uni3-1 or wild-type cells (n=300). Thus, decreasing the stability of the microtubules by pharmacological means does not suppress the flagellar defect of uni3-1 cells.
Suppression is unaffected by mutations that confer resistance to oryzalin Double mutants were constructed to determine the effect of other tubulin mutations on the suppression of the uni3-1 allele and on growth on colchicine medium. A mis-sense mutation in β-tubulin (tub2-1) confers resistance to colchicine (Lee and Huang, 1990) and results in hyperstabilization of the microtubules (Schibler and Huang, 1991). Mis-sense mutations in α1-tubulin (tua1-1) and in the APM1 locus confer resistance to oryzalin, a herbicide that destabilizes microtubules (James et al., 1989;James et al., 1992;Lux and Dutcher, 1991). Double mutants were constructed between tua2-3 and tub2-1, tua1-1 or apm1. In all three double mutants, the colchicine-or oryzalinresistance phenotypes were epistatic to the colchicinesupersensitivity phenotype (the cells were resistant rather than supersensitive to the drugs) the triple mutants of tua2-3; tua1-1; uni3-1 and tua2-3; apm1; uni3-1, the Uni3 flagellar phenotype remained suppressed. The tub2-1 mutation was not tested in triple mutants with uni3-1 because the tua2-3; tub2-1 double mutant strain had defects in flagellar assembly.

Discussion
We have isolated point mutations in TUA2 that suppress the flagellar assembly phenotype caused by a deletion of δ-tubulin, but they fail to suppress the cleavage furrow placement phenotype. We characterized 12 independently isolated suppressors and found that they have one of two amino acid substitutions in α2-tubulin: either D205N or A208T. These residues are highly conserved in α-tubulins across a wide range of species. With two exceptions, they are virtually invariant in the 204 α-tubulin sequences available in GenBank. One exception is the α-tubulin of the Bermuda land crab that has E instead of D at the position equivalent in Chlamydomonas (D205) (GenBank Accession number U92646), the other is the α2-tubulin from Schizosaccaromyces pombe, which has S at the position equivalent in Chlamydomonas (A208) (GenBank Accession number K02842). Alanine scanning mutations in Saccharomyces cerevisiae that change the positions equivalent to D205 and E206 to alanine have a cold-sensitive lethal phenotype and benomyl supersensitivity (Richards et al., 2000). The D205N mutation is presumably not deleterious because Chlamydomonas has a second gene encoding an identical α-tubulin protein.
In the crystal structure of α-tubulin, these residues lie either before or within (D205 and A208, respectively) α-helix 6 (Fig.  5). The N-terminal end of helix 6 contacts the guanine of the nucleotide (Nogales et al., 1998). This helix is proposed to act as a hinge in β-tubulin that rotates after GTP hydrolysis does not occur (Erickson, 1998;Amos and Löwe, 1999). However, this rotation does not occur in α-tubulin because GTP hydrolysis. In addition, the position of this helix differs between zinc sheets and microtubules (Li et al., 2002). Zinc ions induce flat sheets of microtubules with antiparallel protofilaments, whereas microtubules have parallel protofilaments. In zinc sheets, helix 6 contacts helix 4 of the adjacent molecule and part of the M loop (Löwe et al., 2001), whereas in microtubules, there is no interaction across the protofilament and a loss of density (Li et al., 2002). Li and coworkers suggest that the loss of density might result from differences in the behavior of helix 6 in αand β-tubulin (Li et al., 2002). The difference in the placement of helix 6 in αand β-tubulin suggests that this region of α-tubulin can assume different conformations and is flexible. Structural comparisons of axonemal/basal body microtubules to singlet microtubules would help to address this question. Neither of the amino acid substitutions (D205N and A208T) act by making α-tubulin more like δ-tubulin, which has a E at the position equivalent to D205 and an A at the position equivalent to A208.
What are the roles of δ-tubulin and the C tubule? Careful examination of basal bodies from uni3-1 cells by dual-axis electron tomography (O'Toole et al., 2003) and serial sections reveal that C tubules are present at the distal end of the mutant basal bodies but are missing along the remaining length of the basal bodies. These results show that δ-tubulin is not required for initiation of the C tubule but it is probable that δ-tubulin is important for either polymerizing or stabilizing the C tubule. Furthermore, full-length C tubules are not required for the assembly because flagella can be assembled on basal bodies that lack a C tubule (Simpson et al., 1997;Preble et al., 2001;Marshall and Rosenbaum, 2003). δ-tubulin is needed to stabilize or extend the C tubule. Another important question is whether δ-tubulin or the presence of an extended C tubule Fig. 4. Western blot of proteins from cell bodies of wild-type and tua2-2R1 cells. Increasing amounts of protein were loaded to detect the predicted 39 kD truncated protein of tua2-2R1 cells. Lane 1: 7 µg of protein from wild-type cells. Lanes 2-5: 7 µg, 15 µg, 30 µg and 50 µg of protein from tua2-2R1 cells. Truncated tubulin protein is not detectable in tua2-2R1 extracts, only wild-type tubulin. δ-tubulin suppressors allows a nascent centriole to become a mature basal body that can assemble a flagellum.
Several proteins/epitopes are present on old centrioles and absent from nascent or young centrioles during the G1 and S phase of their first cell cycle in mammalian cells. The acquisition of these proteins is referred to as maturation and occurs throughout the G2 phase of the cell cycle. These proteins include cenexin or outer dense fiber 2 (Lange and Gull, 1995;Piel et al., 2000;Nakagawa et al., 2001), ninein, CEP110 (Ou et al., 2002), ε-tubulin (Chang and Stearns, 2000), the dynactin complex (Quintyne and Schoer, 2002) and centriolin (Gromley et al., 2003). Ninein, CEP110 and εtubulin have been localized by electron microscopy to the subdistal appendages located at the distal end of centrioles. The appendages are only present on older centrioles in mammals (Ou et al., 2002;Chang et al., 2003). An older centriole differs functionally from a younger centriole by being able to assemble a primary cilium. This suggests a role for one or more of these proteins and for maturation in promoting the assembly of the cilium. Of those, only ε-tubulin has been identified to date in Chlamydomonas (Dutcher et al., 2002). In addition to the presence of the above proteins, the level of polyglutamylation on αand β-tubulin is increased on older centrioles (Bobinnec et al., 1998) and on older basal bodies (Lechtreck and Geimer, 2000). Unlike centrioles, daughter basal bodies must mature early in their first cell cycle in order to assemble flagella or cilia and so for example, all basal bodies in the ciliated epithelium of the trachea have the ODF2 epitope (McKean et al., 2003). Therefore, there is a need for an additional maturation event in basal bodies which we refer to as early maturation and which must occur at M/G1 phase of the cell cycle rather than in G2/M phase, as for centrioles.
The heterogeneous flagellar assembly phenotype in uni3-1  cells suggests that δ-tubulin might be a signal for early maturation (Fig. 6). Half of the uni3-1 cells were unable to assemble flagella and the remaining cells assembled one or two flagella. Cells without a flagellum produce one daughter cell without flagella and one with a single flagellum (Dutcher and Trabuco, 1998). There is no difference in the morphology of basal bodies that assemble or do not assemble flagella (O'Toole et al., 2003). A subset of basal bodies in uni3-1 cells must mature and recruit proteins that are needed for flagellar assembly.
Based on the phenotype of the suppressor mutations, we propose two hypotheses for the role of δ-tubulin and the C tubule. The tua2 suppressors completely suppress the flagellar assembly defect but only partially suppress the C-tubule defect. We found that 13 of the 14 basal bodies have fewer than five extended C tubules, two of which have only a single extended tubule. Our first model proposes that the presence of extended C tubules is sufficient to promote early maturation and that δtubulin is required for C-tubule stability or extension. In this case, a single extended C tubule must be enough to ensure early maturation, and the tua2 mutations must act by promoting limited extension of C tubules. The second model suggests that δ-tubulin plays two independent roles. First, it is required for stability or extension of the C tubule and, second, it is required for early maturation of the nascent basal body (Fig. 6). In this model, the two functions of δ-tubulin are suppressed to different degrees by the α-tubulin mutations. The mutant αtubulin permits all basal bodies to assemble a flagellum but only partially suppresses the ability to assemble the C tubule. The altered α-tubulin might interact with a protein complex that promotes early maturation more successfully than it does with proteins that affect the stability or extension of the C tubule. Early maturation might involve the recruitment of εtubulin, the intraflagellar transport machinery, which concentrates around the distal region of the basal bodies (Deane et al., 2001), or the Uni1 protein, which is needed for early maturation (Huang et al., 1982). Stability or extension of the C tubule might require unknown microtubule-associated proteins. The uni3-1; tua2-3R1 double mutant is intriguing in that vegetative cells lack flagella but gametic cells are flagellated. We speculate that different proteins are required for maturation and assembly in gametes and in vegetatively growing cells.
The failure to promote the assembly of the C tubule at the proximal end of the basal body also explains why the cleavage furrow defect (Fig. 3) is not suppressed. The C tubule in Chlamydomonas plays a role in the orientation of rootlet microtubules, the extension of centrin fibers of the nucleus/basal-body connector and the attachment of striated fibers between the two basal bodies. Many of these fibers are attached or located at the proximal end of the basal body, which lacks extended C tubules in the suppressed strains. Defects in this fiber system result in defects in cleavage furrow and nuclear placement (reviewed in Dutcher, 2003) (Fig. 6). The cleavage furrow placement is similar to the one observed in bld2; rgn1 cells, where singlet, doublet and triplet microtubules are present (Preble et al., 2001).
The colchicine-supersensitivity phenotype per se is not necessary for suppression of the uni3-1 lesion. If destabilized microtubules alone were sufficient, we would have expected colchicine or oryzalin to phenocopy the suppression, but neither colchicine nor oryzalin suppressed the Uni3 flagellar phenotype. In addition, the triple mutants tua1-1; tua2-3; uni3-1 and apm1; tua2-3; uni3-1 conferred resistance to oryzalin, but showed suppression of the Uni3 flagellar phenotype, therefore providing evidence that the altered α-tubulin is needed for suppression rather than supersensitivity to colchicine.
It is somewhat surprising that none of the isolated suppressors of uni3-1 show mutations in TUA1, another αtubulin gene. TUA1 is tightly linked to UNI3 on linkage group III (250:0:0 PD:NPD:TT), and the uni3-1 allele represents a 27 kb deletion. A 7 kb clone containing only the UNI3 gene rescues the phenotypic defects of uni3-1 (Dutcher and Trabuco, 1998), which demonstrates that the additional sequences removed in the deletion are apparently not important. However, it is possible that the deletion has altered the DNA structure in the region and thus made the TUA1 gene less accessible to UV or γ-irradiation mutagenesis. It is also possible that there are slight differences in the expression patterns of the two tubulin genes that favor TUA2 as a site of mutation for suppression.