The gene (GTU1) encoding Tetrahymena thermophila γ-tubulin was cloned and analyzed. GTU1 is a single-copy, essential gene encoding a conventional γ-tubulin. HA-tagged GTU1p localizes to four microtubule-organizing centers (MTOCs) in vegetative cells: basal bodies (BBs), macronuclear envelopes, micronuclear envelopes, and contractile vacuole pores. γ-Tubulin function was studied by placing the GTU1 gene under control of an inducible–repressible promoter. Overexpression of GTU1 had no detectable effect on cell growth or morphology. Depletion of γ-tubulin resulted in marked changes in cell morphology and in MT bundling. MTOCs showed different sensitivities to γ-tubulin depletion, with BBs being the most sensitive. γ-Tubulin was required not only for the formation of new BBs but also for maintenance of mature BBs. BBs disappeared in stages, first losing γ-tubulin and then centrin and glutamylated tubulin. When GTU1 expression was reinduced in depleted cells, BBs reformed rapidly, and the normal, highly organized structure of the Tetrahymena cell cortex was reestablished, indicating that the precise patterning of the cortex can be formed de novo.
Abbreviations used in this paper: BB, basal body; cs, cold sensitive; CVP, contractile vacuole pore; KF, kinetodesmal fiber; Mac, macronucleus; Mic, micronucleus; MT, microtubule; MTOC, MT-organizing center; OA, oral apparatus; OP, oral primordium; SPP, super proteose peptone; TM, transverse microtubule.
The discovery of γ-tubulin was a major breakthrough in analyzing MTOCs (Oakley and Oakley, 1989). γ-Tubulins are conserved in eukaryotes (Burns, 1991) and are structurally similar to α- and β-tubulins (Inclán and Nogales, 2001) to which they are ∼35% identical. γ-Tubulin functions in MT nucleation. During interphase, it localizes in MTOCs such as the centrosomes of animal cells and the spindle pole bodies of yeast, and it accumulates at spindle poles in most cell types examined (Horio et al., 1991; Stearns et al., 1991; Zheng et al., 1991). γ-Tubulin localizes to basal bodies (BBs), the MTOCs for axonemes (Fuller et al., 1995), and is required for BB duplication in Paramecium (Ruiz et al., 1999). γ-Tubulin is present in the cytoplasm of Xenopus and Drosophila eggs as open-ring complexes (γTuRCs) of ∼25 nm diameter (Moritz et al., 1995; Zheng et al., 1995) that can nucleate MTs and bind to MT minus ends. Immunodepletion of γ-tubulin in vitro (Felix et al., 1994; Moritz et al., 1998) or microinjection of anti–γ-tubulin antibodies in vivo (Joshi et al., 1992) inhibits centrosomal MT nucleation. Disrupting γ-tubulin genes in fungi (Oakley et al., 1990; Spang et al., 1996) or flies (Sunkel et al., 1995) produces phenotypes, suggesting a role for γ-tubulin in MT nucleation in vivo. These studies argue that γ-tubulin is an essential component of MTOCs and has a direct role in MT nucleation. Mutational analyses in fungi (Paluh et al., 2000; Jung et al., 2001; Hendrickson et al., 2001; Prigozhina et al., 2001) indicate that γ-tubulin also plays a role in regulating MT dynamics.
Tetrahymena thermophila cells contain at least 17 differentiated MT systems (Gaertig, 2000), including BBs, axonemes, transverse and longitudinal MTs, contractile vacuole pores (CVPs), and mitotic and meiotic spindles. To study how these systems are organized and regulated, we cloned the Tetrahymena γ-tubulin (GTU1) gene. It encodes a conventional γ, ∼65% identical to the γ-tubulins of multicellular eukaryotes. GTU1 is essential for growth. An epitope-tagged γ-tubulin that rescues a GTU1 knockout localized to oral and somatic BBs, to the poles of the mitotic micronucleus (Mic), in envelope-associated particles of macronuclei (Macs) and Mics, and in the CVPs.
We analyzed the effects of depleting γ-tubulin, either by disrupting the GTU1 gene (Hai and Gorovsky, 1997; Hai et al., 2000) or by placing it under control of the inducible-repressible MTT1 promoter (Shang et al., 2002). Different MTOCs showed different rates of loss of γ-tubulin. New BBs failed to replicate and already formed BBs disappeared. BBs disappeared in stages, first losing γ-tubulin and then centrin and glutamylated tubulin. These results show that γ-tubulin is required not only for BB duplication, but also plays a role in maintenance of normal BB structure. Basal bodies could also reform rapidly, and normal cell morphology could be reconstituted by reinducing γ-tubulin expression.
Sequence of the
T. thermophila GTU1 gene
The major products of genomic and RT-PCR reactions using two degenerate primers derived from highly conserved regions found in γ- and β-tubulin were cloned. The complete sequence of these clones identified a putative γ-tubulin coding gene (GTU1) that shared ∼60% identity with the known consensus sequence. Southern blotting showed a single band (Materials and methods), indicating that GTU1 was a single copy gene in the Tetrahymena thermophila genome, which was confirmed by the genetic analyses described below. It encodes a 50.9-kD protein of 449 amino acid residues (Fig. 1)
. The coding region contains five introns and 11 TAA or TAG glutamine codons. The cDNA sequence indicates that polyadenylation occurs 208 nt downstream of the TGA termination codon. As in other Tetrahymena genes, there is no properly positioned, canonical polyadenylation site. Tetrahymena γ-tubulin shares ∼65% identity with most γ-tubulins. A phylogenetic analysis (Fig. 2)
places it with other ciliate tubulins at a position consistent with the phylogeny of ciliates. Tetrahymena γ-tubulin is 35–45% identical to the divergent, nonconventional γ-tubulins of Saccharomyces cerevisiae, Caenorhabditis elegans, and Entamoeba Histolytica (Burns, 1995). Thus, Tetrahymena contains a conventional γ-tubulin.
GTU1 is essential in T. thermophila
To study the effect of GTU1 depletion, we created GTU1 knockout heterokaryon strains with both copies of the GTU1 gene disrupted in the diploid Mics but with the 45 WT copies still present in the polyploid Macs (Materials and methods). We then created ΔGTU1 cells (whose Mics and Macs lacked any GTU1 genes) by crossing two heterokaryon strains to each other. The progeny of this mating separated (exconjugated) at the proper time, indicating that zygotic expression of the GTU1 gene is not required for completion of conjugation. When refed, ΔGTU1 cells regenerated oral apparatuses (OAs) (unpublished data) and initiated growth, which slowed ∼5 h after refeeding, and almost stopped between 7.5 and 12.5 h. The apparent growth of the culture after 12.5 h is likely due to the proliferation of contaminating, unmated parental cells. Progeny of a cross between WT control strains CU428 and B2086 grew steadily over the same time course (unpublished data).
The function of γ-tubulin was also assessed by depleting it from growing cells. GTU1 heterokaryon progeny were rescued with a COOH-terminal HA-tagged GTU1 gene (Fig. 3
A) under control of the cadmium-inducible MTT1 promoter (Shang et al., 2002), producing strain cTTMG-HA. These cells grew normally in the presence of cadmium, indicating this tagged gene functions normally and can be overexpressed without causing an obvious phenotype. Log phase cTTMG-HA cells grown in super proteose peptone (SPP)/ CdCl2 (0.5 μg/ml) were shifted to CdCl2-free SPP to repress expression of the GTU1 gene. Growth slowed at ∼11 h and stopped 16 to 17 h after depletion, whereas similarly treated WT cells continued to grow to maximal density (unpublished data). These studies demonstrate that the GTU1 gene is essential for growth.
γ-Tubulin localizes to four major MTOCs in vegetative cells
The only functional GTU1 gene in both tG2N-HA and cTTMG-HA strains is the COOH-terminal HA-tagged GTU1 gene (Materials and methods), either under control of the endogenous GTU1 or MTT1 promoter, respectively. The growth and mating of the cTTMG-HA strain were indistinguishable from that of WT (unpublished data), indicating that the chimeric gene functions normally.
Anti-HA antibodies detected a protein of the expected size (∼60 kD) in tG2N-HA and cTTMG-HA (Fig. 3 A). To study the localization of γ-tubulin, we stained vegetative cells with anti-HA antibodies (Fig. 3 B). In WT cells, no specific staining was detected (unpublished data). In GTU1-HA transformants, four organelles, likely to be MTOCs based on their associated MTs, stained strongly. First, all BBs, including those associated with somatic and oral cilia and those associated with the newly forming OA known as the oral primordium (OP) before they became ciliated. Staining intensities changed during the cell cycle. The newly formed OP was stained more strongly than the old OA in dividing cells and the somatic BBs in cells containing an OP were stained more strongly than those in interphase stage. Second, numerous small patches that were closely associated with Mac envelopes were stained and showed changes during the cell cycle. Mics and Macs divide by different mechanisms and at different times in the cell cycle. Mics divide by a closed mitosis before Macs, which divide amitotically, without chromosome condensation or segregation of sister chromatids. Macs lack distinct spindles but contain an elaborate array of MTs (Fujiu and Numata, 1999). Strong signals could only be observed on dividing MACs. The stained spots were originally distributed relatively evenly across the surface of Macs (unpublished data) and then lined up, forming short rows. Third, small particles were observed associated with interphase Mic envelopes, the likely MTOCs for Mic spindle MTs. During mitosis, staining occurred along the spindle axis with preferential accumulation at the poles of the dividing Mics. In telophase, we could observe a string of staining between the two daughter Mics, indicating that the separation spindle also stains. And fourth, the two posteriorly located CVPs stained strongly and were the only MTOCs in Tetrahymena that showed no obvious cell cycle changes in staining throughout the cell cycle (Fig. 3 B). CVPs are ring structures marking the sites at which Tetrahymena cells eliminate liquids (Cameron and Burton, 1969). They contain circular MTs and are sites from which other MTs radiate (Elliott and Bak, 1964). Although not rigorously established as such, we will refer to the (nonspindle) structures containing γ- tubulin as MTOCs for convenience.
MTOCs show different sensitivities to depletion of γ-tubulin
To determine the effects of depletion of γ-tubulin on the function of the four MTOCs, we stained GTU1 knockout progeny with an antibody to α-tubulin (Fig. 4)
. Loss of OAs could be observed within 7.5 h after refeeding. Some cells also lacked CVPs, whereas the other MTOCs appeared to be normal (Fig. 4 A, a). Because progeny cell samples also contain ∼5% unmated parental cells which can grow normally after refeeding, we restricted subsequent analyses to cells without OAs or with abnormal OP development. We classified cells into four categories: cells with normal longitudinal MT structure (LG-MT) (Fig. 4 A, b); cells beginning to lose their shape and round up (Fig. 4 A, c, round); cells with misplaced division furrows (Fig. 4 A, d and e), and cells with abnormal nuclear morphology, including two dividing Mics, elongated dividing Mics, abnormal Macs, or small nuclear fragments (Fig. 4 A, f–i). These cells showed progressive changes in cytoplasmic and nuclear morphology. From 7.5 to 9.5 h, most contained longitudinal MTs, whereas >50% of them had started to round up. Some cells had two or more normal dividing Mics and a few cells showed abnormalities in nuclear divisions and cytokinesis. By 12 h, >80% of the cells were round. By 14.5 h, almost all the cells (∼90%) were round and the longitudinal MTs in some cells (∼8%) were disorganized. Anti–α-tubulin staining in these cells also showed disturbed cytoplasmic MTs, some of which formed bundles. In addition, more cells (∼10%) show abnormalities in cytokinesis and nuclear morphology. The cTTMG-HA strain, containing an MTT1-driven GTU1 coding sequence, showed similar progression of phenotypes when shifted to CdCl2-free SPP medium (unpublished data).
We conclude from these studies that Tetrahymena γ-tubulin is essential for the normal function of both BBs and nuclei, and that BBs are more sensitive to the depletion of γ-tubulin than nuclei. After refeeding, unmated parental cells that contaminate the progeny of a knockout heterokaryon mating, soon outgrow progeny cells. In addition, it is not possible to restore γ-tubulin to these cells to study the recovery of MTOCs and MTs. Therefore, for subsequent studies, we used strain cTTMG-HA to analyze the effect of inactivation of GTU1 expression after transfer to CdCl2-free medium and the effects of re-addition of CdCl2 after depletion.
γ-Tubulin is required for both duplication and maintenance of normal basal bodies
To study the localization of γ-tubulin in BBs, we stained depleted cTTMG-HA cells with anti-HA antibody (Fig. 5)
. By 17 h, when the culture had stopped growing, HA staining could not be detected on somatic BBs, OAs, or CVPs, but still remained or even increased on both nuclear envelopes. In WT cells, the Mac envelope does not stain strongly before the Mic starts dividing. However, in the depleted cells, we observed many cells with interphase Mics that had small, strongly staining patches on the Mac envelope. Thus, the more rapid effects of γ-tubulin depletion on MT systems organized by BBs than on nuclei can be explained by the fact that γ-tubulin is depleted more rapidly from BBs. Interestingly, Mac staining also disappeared eventually, and by 30 h after depletion, the only detectable MTOCs in most cells were three to five spots on Mic envelopes (Fig. 5). These studies indicate that γ-tubulin is a dynamic component of Tetrahymena MTOCs, as it is of centrosomes in mammalian cells (Khodjakov and Rieder, 1999).
Next, we examined the effects of γ-tubulin depletion on BB components and BB associated structures. Antibodies to centrin, an essential component of BBs that is also found in yeast spindle pole bodies (Biggins and Rose, 1994), stained BBs strongly. Surprisingly, centrin staining could not be detected in the other γ-tubulin containing MTOCs (Fig. 5) and could not be detected in isolated nuclei on Western blots (unpublished data).
Surprisingly, anti-centrin staining of somatic BBs and OAs remained almost normal for many hours, even after γ-tubulin could no longer be detected, suggesting that γ-tubulin could be lost without immediate disruption of BBs. We classified centrin staining patterns into four categories (Fig. 6, A and C)
. 9 h after initiating depletion, most cells are type I, with normal OAs, some of which were dividing cells lacking normal oral primordia. Stage I cells can be subdivided into those with normal centrin staining (Fig. 6 A, stage Ia) and cells whose centrin staining on somatic BBs at the posterior end was partially disrupted (Fig. 6 A, stage Ib). After 11 to 12 h of depletion, the number of ΔOA cells, most with near normal organization of somatic BBs detected by centrin staining (Fig. 6 A, stage II), started to increase. Still later, more cells showed disrupted or near absence of centrin staining at their posterior ends (Fig. 6 A, stage III a), whereas retaining punctate rows of BBs at their anterior ends (Fig. 6 A, stage IIIb). Eventually, most punctate BBs disappeared and centrin staining accumulated at the anterior end in large aggregates (Fig. 6 A, stage IV).
The changes in centrin distribution (Fig. 6 C) can be analyzed relative to the growth of cells after cadmium depletion (Fig. 6 B). The culture grew actively for ∼9 h, slowed down at ∼11 h, and completely stopped by 24 h. As in Fig. 6 B and in three additional, independent experiments, the decrease in the number of cells containing an OA that occurred when culture growth had slowed or stopped always exceeded the increase in cell number. Cells without an oral apparatus cannot grow in SPP medium and we have never observed a dividing cell completely lacking an anterior OA, suggesting that the old OA is never completely resorbed in a dividing cell. Therefore, it is likely that loss of oral BBs is not due simply to their failure to replicate in dividing cells. Loss of punctate, somatic BBs also continues after cells have stopped dividing. From 24 to 31h after the culture stopped dividing, the number of type IV cells increased dramatically, from 25 to 46%.
To confirm that the appearance of type III cells results from the disassembly of BBs rather than dilution by cell growth, we shifted cells containing a γ-tubulin cold-sensitive (cs) mutation (which showed no phenotype in Aspergillus; Jung et al., 2001), to 15°C in SPP/CdCl2 (0.5 μg/ml) and studied the kinetics of the changes in centrin distribution. These cells stopped dividing completely by 48 h (Fig. 6 D), after which the number of type III cells increased ∼40% from days 2–6, whereas the cell number remained unchanged (Fig. 6 E).
The progressive changes that occurred at the later stages of depletion are not due to cell death, because ∼75% of the cells that had been in CdCl2-free SPP medium for 61 h could regrow when plated as single cells in SPP containing 0.5 μg/ml CdCl2. Moreover, >90% of the cs mutant cells which had been maintained at 15°C for 6 d recovered after they were transferred back to permissive temperature (30°C). In the depletion study, ∼80% of the cells started making new BBs after we added CdCl2 (see below). These analyses indicate that γ-tubulin is not only required for BB replication in Tetrahymena, but also for maintenance of BBs in the highly organized OA and the cortical rows.
We also examined the localization of glutamylated tubulin, another marker for BB composition. In Tetrahymena, tubulin is glutamylated in a number of MT systems (Gaertig, 2000), but BBs stain particularly strongly with an antibody (R-polyE) raised against polyglutamic acid which reacts mainly with tubulins on Western blots of Tetrahymena whole cell protein (unpublished data). In the early stages of γ-tubulin depletion, glutamylated tubulin in BBs colocalizes with centrin staining (Fig. 7
A) in punctate BBs. By 17 h, some cells showed disperse centrin staining at their posterior ends which does not colocalize with the more punctate polyE staining. At later stages (30 h after depletion), staining of both centrin and glutamylated tubulin is no longer punctate and does not colocalize in most areas of the cell. The complete loss of organized oral and somatic BBs, coupled with the distinct patterns of disproportionation of the three BB components, indicate that BBs are dynamic structures whose components turn over at different rates.
A possible explanation for the apparent BB disassembly observed with centrin and polyglutamylation staining is that many BBs were not mature when γ-tubulin was either depleted or inactivated and that immature BBs are unstable. To test this, we analyzed the effects of depletion on transverse microtubules (TMs) a marker for the MTOC activity of BBs. TMs appear late during BB formation, so are markers for relatively mature BBs (Allen, 1969). We visualized TMs with anti–α-tubulin antibodies (Fig. 7 B). By 17 h, as the number of BBs dropped, many cells started losing their TMs. By 30 h, many cells lacked detectable TMs. Interestingly, TMs disappear after γ-tubulin, at about the time that the punctate distribution of centrin and glutamylated tubulin are lost, suggesting that components of BBs other than γ-tubulin are sufficient to maintain TMs after they are formed.
To determine whether loss of γ-tubulin specifically affected the MTOC-related function of BBs or whether non–MT-associated functions were also affected, we analyzed kinetodesmal fibers (KFs, also known as striated rootlets). KFs are nonmicrotubular components associated with BBs in mature ciliary units that can be detected by antibody 5D8 (Jerka-Dziadosz et al., 1995). KFs also are disrupted by γ-tubulin depletion. In undepleted cells, the 5D8 antibody stained regions associated with the somatic BBs, but not the OA. In depleted cells, 5D8 staining showed progressive changes similar to those detected by anti-centrin antibody. Staining adjacent to somatic BBs started disappearing by 17 h and accumulated at the anterior end of cells by 30 h (Fig. 7 C). However, the distribution of centrin and the KF fiber antigen did not overlap completely (Fig. 7 C). Thus, the disappearance of both transverse MTs and KFs is more likely a secondary effect due to loss of BB structure rather than a direct effect of γ-tubulin depletion.
To determine whether γ-tubulin depletion disrupted cortical structures not directly associated with BBs, we examined the distribution of TFKBP12 (12 kD Tetrahymena FK506-binding protein), which normally surrounds the somatic BBs (Dr. Osamu Numata, personal communication). Anti-TFKBP12 staining (Fig. 7 D) was affected by the depletion of γ-tubulin only after punctate BBs and other cortical MT structures had largely been disrupted or had disappeared. In undepleted cells, TFKBP12 localized to rings around the somatic BBs. After depletion, this ring structure was retained but the rings became randomly distributed near the cell surface. These observations suggest that the changes in BB components and BB associated structures reflect specific effects of γ-tubulin depletion on the BBs, and not a more general affect on all cortical structures.
The effects of γ-tubulin depletion are reversible
To reinduce GTU1 expression, we shifted cTTMG-HA cells that had been in cadmium-free SPP medium for either 28 or 42 h to SPP medium containing 1.0 μg/ml CdCl2. At various times after reinduction, cells were fixed and stained with anti-centrin, anti–R-polyE, and anti-KF (5D8) antibodies (Fig. 8)
. The BB components rapidly relocalized to newly formed somatic BBs. By 1 h after induction, we observed punctate areas throughout the cell, including the posterior end of the cell where there were almost no such structures after depletion, which could be stained by all three antibodies (Fig. 8, a and b). Initially, newly formed BBs were arranged randomly. However, by 5 h, ∼79% of the cells had short, randomly oriented rows of BB of different lengths. In some of these rows, most of the BBs were not ciliated (Fig. 8, c). After 12 h in CdCl2, the BBs of most cells were organized into the correct pattern, cells regained WT shape, became motile, and started to divide (unpublished data).
The Tetrahymena thermophila GTU1 gene was shown to be a single-copy gene encoding a conventional γ-tubulin that localized to four major sites that agree well with γ-tubulin localization in other ciliates (Liang et al., 1996; Ruiz et al., 2000). The first is BBs, including those associated with somatic and oral cilia and with the OP before it becomes ciliated. Interestingly, the BBs are stained more strongly in dividing than in interphase cells, suggesting either that the amounts of γ-tubulin associated with BBs increases, or that a structural change occurs making it more accessible to antibody. The second is small patches associated with Mac envelopes in dividing cells. The third is patches associated with interphase Mic envelopes and the mitotic spindle, with preferential accumulation at the poles of dividing Mics. In Tetrahymena, γ-tubulin staining associated with the nuclear envelopes also increased in dividing cells. In vertebrate somatic cell lines, γ-tubulin staining associated with the centrosome remains relatively constant throughout interphase and increases markedly at the onset of mitosis (Khodjakov and Rieder, 1999). Thus, MTOCs associated with nuclear divisions in diverse organisms show similar dynamic changes during the cell cycle, likely reflecting the recruitment of γ-tubulin during MTOC duplication and/or activation. The fourth is the CVPs.
We analyzed γ-tubulin function by knocking out the GTU1 gene or expressing it under control of the inducible-repressible MTT1 promoter and repressing its expression. As in other organisms, the gene encoding γ-tubulin is essential in Tetrahymena. In the absence of γ-tubulin, cells stop growing and show progressive changes in cell and nuclear morphology (Fig. 9)
. The first visible phenotypes are the loss of the OA and CVPs. The cells then began to lose their shape and round up. The cortical MT structure is also affected. We also observed bundles of cytoplasmic MTs in some cells, consistent with previous studies in fungi showing that depletion of γ-tubulin led to the bundling of cytoplasmic MTs (Sobel and Snyder, 1995; Jung et al., 2001). Thus, in Tetrahymena as in fungi, one function of γ-tubulin may be to regulate the dynamics of cytoplasmic MTs. A novel finding of this study is that different MTOCs have different sensitivities to γ-tubulin depletion: CVPs and BBs are more sensitive than the nucleus-associated MTOCs. Abnormally dividing Macs and Mics were not observed until after many cells had lost their OAs and much of their somatic ciliature, had lost normal shape, and had badly mis-localized division furrows. Consistent with the resistance of nuclei to γ-tubulin depletion, immunofluorescence staining of HA-tagged γ-tubulin was retained, and may even have increased, on nuclear envelopes, especially those of Mics, after staining of BBs and CVPs was lost, suggesting that γ-tubulin in nuclear MTOCs turns over more slowly than γ-tubulin in BBs and CVPs. Thus, in addition to producing MT arrays that have different sensitivities to microtubule depolymerizing drugs (Morrissette and Sibley, 2002), MTOCs themselves have different stabilities.
Two pathways for BB duplication have been identified (Marshall et al., 2001). In the more common, templated pathway, new BBs form in association with preexisting BBs. In the de novo pathway, BBs form in the absence of preexisting BBs. In Tetrahymena, most somatic BBs form once per cell cycle by a templated mechanism, (Allen, 1969) doubling the total number of somatic BBs each cell cycle. In contrast, to generate a new OA, a large number of BBs in the OP region is produced during a small portion of the cell cycle, either by rapid, repeated templating or, more likely, by a de novo pathway. A requirement for γ-tubulin in this rapid synthesis of new BBs likely explains why dividing cells without normal OP development appeared before other phenotypes were observed. We conclude that, as in Paramecium (Ruiz et al., 1999), γ-tubulin is required for BB duplication. Two lines of evidence support another novel finding of our study, that γ-tubulin is required for maintenance of mature BBs. First, the increase in cells lacking OAs was greater than the increase in cell number as growth in depleted cultures slowed and eventually stopped. Thus, some cells without OAs cannot be derived from cells that divided after failing to form a new OA. Second, cells continued to lose somatic BBs after they stopped growing. In both the depletion study and cs mutant, the number of type IV or type III cells increased dramatically after the culture had stopped dividing. Loss of function of mature somatic BBs was also indicated by the loss of transverse MTs, which are nucleated from nearly fully formed BBs (Allen, 1969), after the culture stopped dividing. Similarly, kinetodesmal fibers, a nonmicrotubular element associated with mature ciliary units, was disrupted after the culture stopped dividing.
Surprisingly, the components of the BBs did not disappear in concerted fashion. First, γ-tubulin disappeared from almost all the BBs and from the cytoplasm, followed by the delocalization of centrin and glutamylated tubulin, key components of BBs. At late stages of depletion, few punctate structures remained in the posterior ends of the cells that could be stained by both antibodies. Most of the centrin and glutamylated tubulin staining localized to the anterior of cells, but, for the most part, did not colocalize. This gradual loss of different BB components after γ-tubulin depletion differs from the scattering of BB components that occurred after microinjection of anti-polyE antibody into tissue culture cells (Bobinnec et al., 1998). Asynchronous loss of components suggests that the functions of γ-tubulin in organizing and maintaining BBs are distinct. Consistent with this, the BBs of Xenopus sperm lack γ-tubulin, but acquire it in the egg cytoplasm before functioning in centrosomes (Stearns and Kirschner, 1994). In mammalian cells, γ-tubulin is a component of the core of the centriole (Fuller et al., 1995) whose dimensions are similar to those reported for the relatively stable macrotubules (γ-tubules) produced by overexpression of γ-tubulin (Shu and Joshi, 1995). Thus, it is likely that γ-tubulin self-assembles into tubules, which organize the centriole and fill the centriolar lumen. In Tetrahymena, the loss of γ-tubulin from BBs during depletion indicates that, whatever its structure, the γ-tubulin of BBs is dynamic and is required for long-term retention of other BB components. In addition, the increase in nuclear staining during depletion suggests BBs and other MTOCs likely compete for γ-tubulin pool in the cell.
During recovery, γ-tubulin synthesis precedes the appearance of both centrin and glutamylated tubulin in newly formed BBs, consistent with a nucleating function for γ-tubulin. We first observed punctate, BB-like structures labeled by centrin, R-polyE, or KF antibodies, suggesting that the individualized new BB structures are relatively mature. Later, short rows of BBs appear. Compared with WT cells, there are two major differences in the morphology of these newly formed BB rows. First, the way they are arranged is different. Instead of being organized in continuous rows of relatively uniform length that are positioned roughly parallel to the long axis of the cell, the newly formed BB rows are randomly oriented and vary in length. This random pattern of the BB rows suggests that the first newly formed BBs were probably formed de novo at multiple sites, instead of by a templated pathway from some undetectable BB precursors that had retained their position in the cell. However, the fact that the precise cortical pattern was eventually reestablished suggests the existence of unknown mechanisms that serve to position already formed rows of BBs. The random distribution of TFKBP12 staining after depletion argues against the existence of a persistent, precise cortical pattern as an explanation for this reorganization.
The ability of ciliates to form BBs de novo was demonstrated by Hammersmith and Grimes (1981) who showed that encystment of Oxytricha fallax cells led to a loss of all the visible ciliary structures and BBs, which could be reformed upon excystment. However, in Oxytricha, all the cortical structures reformed in a correct pattern, indicating that the hypothetical determinative factors/structures for the cortical patterns were retained precisely and dictated the positions in which new BBs formed. The basis for these differences is not known. However, it is possible that, in our study, the depletion of γ-tubulin might have led to the disruption of the hypothetical determinative factors/structures. It is not known whether γ-tubulin is retained and remains localized in Oxytricha cysts.
Another observation consistent with a de novo pathway of BB formation in recovering cells is that the newly formed BB rows in these cells are not always ciliated as they are in WT cells. In Tetrahymena, new somatic BBs are normally generated immediately anterior and perpendicular to an old one, followed by a 90° rotation and migration to the surface; cilia form when the new BB contacts the cell surface (Allen, 1969). Therefore, in WT cells, only the newly formed, immature BBs are not ciliated. Thus, most somatic BBs duplicate once per cell cycle by a templated pathway, and adjacent unciliated BBs are rare. However, in the recovering cells, we observed BB rows with four or more adjacent unciliated BBs, suggesting either that BBs in these rows are forming more rapidly or that ciliogenesis is slower than in normal cells.
In conclusion, γ-tubulin is found in four MTOCs in Tetrahymena that show different sensitivities to γ-tubulin depletion, suggesting that the concentration of γ-tubulin in the cell could regulate the function of different MTOCs. γ-Tubulin is not only required for the formation of BBs but also for the retention of both centrin and glutamylated tubulin in mature BBs, and is, therefore, essential for maintaining the normal structure of BBs after they are formed. Finally, Tetrahymena cells can reform BBs, probably by a de novo pathway, and can reestablish the normal cortical structure from randomly arranged rows of BBs in the absence of obvious, preexisting organization of BBs.
Materials And Methods
Strains, culture growth, and conjugation
Strains are described in Table I
. Tetrahymena cells were grown in SPP (1% proteose peptone, 0.2% glucose, 0.1% yeast extract, 0.003% EDTA × ferric sodium salt) at 30°C (Gorovsky et al., 1975). For conjugation, mid-logarithmic phase cultures of two different mating types were washed two times and resuspended in starvation medium (10 mM Tris-Cl, pH 7.5). After 16–20 h, equal numbers of each mating type were mixed at 2 × 105 cells/ml and incubated at 30°C without shaking.
Gene cloning and sequence analysis
Two degenerate primers were designed from regions conserved among available γ-tubulin sequences. First, strand cDNA was prepared by reverse transcription of RNA isolated from growing cells (White et al., 1988) and used as a template for RT-PCR. PCR products were cloned into Bluescript vectors (Stratagene) and sequenced.
For inverse PCR, Tetrahymena genomic DNA isolated (Gaertig et al., 1994b) from WT cells was digested with HindIII or BglII, self ligated, and amplified by inverse PCR with internal primers (GTUB01/02 and 04/05; Fig. 1) based on the partial genomic sequence of GTU1. Two primers (GTUB06/07) designed from the known end sequences of the cloned GTU1 genomic fragment were then used for genomic PCR to amplify the complete genomic GTU1 gene, which was cloned into pCRII (Invitrogen).
GTU1 coding regions were sequenced in independent clones three times in coding regions and twice in flanking regions (EMBL/GenBank/DDBJ under accession no. U96076; Fig. 1). γ-Tubulin sequences were aligned using CLUSTAL software W (v. 1.6) (Thompson et al., 1994) on the DDBJ web site. Trees were produced by neighbor-joining (Saitou and Nei, 1987). Bootstrap values were calculated (Felsenstein, 1985) by resampling 1,000 data sets.
Germline gene knockouts
For the GTU1 knockout construct, pΔGN, the neo2 gene cassette which confers paromomycin (pm) resistance when expressed in Tetrahymena was inserted between a 1.9-kb fragment of the GTU1 5′ and a 1.0-kb fragment of the GTU1 3′ flanking sequence (Gaertig et al., 1994a). The Mic GTU1 gene was disrupted using biolistic particle bombardment (Cassidy-Hanley et al., 1997) to transform conjugating CU428 and B2086 cells 2.5 h after mixing with a 4.3-kb KpnI-SacI fragment released from pΔGN. Knockout heterokaryon strains with disrupted GTU1 genes in their Mics and WT GTU1 genes in their Macs (GTUKO5 and GTUKO6) were created as described (Hai and Gorovsky, 1997; Hai et al., 2000). When these strains conjugate, the old, pm-sensitive Macs are replaced by new ones produced from the cells' Mics. Consequently, the neo2 gene disrupting the GTU1 gene is expressed, allowing drug selection for successful mating. However, if GTU1 is essential, the mating progeny will die unless they are transformed with a GTU1 gene that supports growth.
Tagged and rescuing plasmids
To construct plasmid pG2N-HA for the GTU1-HA fusion gene under control of the GTU1 promoter, two neo2 gene cassettes were inserted into the EcoRV site of the WT GTU1 3′ flanking sequence in pBL-GTU4. A HA epitope sequence (CYPYDVPDYASL) was inserted before the GTU1 stop codon using the MORPHTM mutagenesis Kit (5 Prime - 3 Prime, Inc.). The insert of pG2N-HA was cut out with NsiI and NotI and transformed into starved Tetrahymena strain CU428 cells. Transformants were initially selected in 60–80 μg/ml pm and then transferred every 3 to 4 d into media containing increasing drug concentrations until cells failed to grow (3 mg/ml). To construct a rescuing plasmid containing either the WT GTU1 coding region (pTTMG) or the GTU1-HA fusion gene (pTTMG-HA) under control of the MTT1 promoter, the GTU1 or GTU1-HA coding region was inserted between the MTT1 5′ and 3′ flanking regions. The mutated cs gtu1 gene (pK358AK359A-HA) was produced by PCR mutagenesis. Knockout heterokaryon strains GTUKO5 and GTUKO6 were mated and transformed with the KpnI and SacI fragments 24 h after mixing. Rescued progeny were selected with 60 μg/ml pm in the presence of 1.0 μg/ml CdCl2.
Southern blot analysis and genomic PCR
To determine if the GTU1 gene is a single copy gene in Tetrahymena, genomic DNA was digested with EcoRV, HindIII, or BglII, electrophoresed, blotted (Magnagraph membrane; Osmonics), and hybridized at 65°C with an [α-32P] dATP-labeled PCR product from coding region (Ausubel et al., 1988). To confirm the genotype of the GTU1-HA transformants, the GTU1-HA fusion region of strain tG2N-HA, cTTMG-HA, or cK358AK359A-HA was PCR amplified and sequenced.
Western blot analysis
To demonstrate expression of HA-tagged γ-tubulin, CU428, tG2N-HA, and cTTMG-HA strains were grown to 2 × 105 cells/ml in SPP/CdCl2 (1.0 μg/ml) at 30°C, harvested and washed in 10 mM Tris-Cl, pH 7.5, containing protease inhibitors (Complete Mini; Roche). The cells were pelleted and dissolved in SDS sample buffer containing 5% β-mercaptoethanol (BME), 10% glycerol, 2% SDS, 60 mM Tris-Cl, pH 6.8. Proteins from 0.5–1 × 105 cells per lane were separated on 10% SDS-PAGE and transferred onto an Immobilon-P membrane (Millipore) by a semidry electrophoretic transfer unit (Bio-Rad Laboratories). The membrane was blocked in 5% dry milk in TBST (100 mM Tris-HCl, pH 7.5, 0.9% NaCl, 0.1% Tween 20) for 1 h at room temperature and incubated overnight at 4°C with monoclonal anti-HA antibody (16B12; Covance) or anti–β-tubulin (DM1B; Sigma-Aldrich) at 1:20,000 dilution. The membrane was washed in TBST at room temperature, incubated in rabbit anti–mouse IgG HRP-conjugated secondary antibody (Zymed Laboratories, Inc.) at 1:20,000 dilution for 1 h at room temperature, and washed five times for 5 min each in TBST, and then once in TBS. Proteins were detected using Western Blot Chemiluminescence Reagent (NEN).
Depletion/Inactivation of γ-tubulin
To bring the knockout phenotype to expression, heterokaryons GTUKO5 and GTUKO6 were mated. At 4 to 5 h after mixing, paired cells, which swim poorly and settle to the bottom of the dishes, were collected into fresh starvation medium. This was repeated three times, enriching pairs to ∼95% of the population. Mating cells were refed with SPP medium 24 h after mixing. In the depletion studies, log phase cTTMG or cTTMG-HA cells, whose GTU1 gene is under the control of the MTT1 promoter requiring CdCl2 for expression, were shifted from SPP medium with to medium without CdCl2 (0.5 μg/ml). To study the phenotype of the cs mutant, log phase cK358AK359A-HA cells were shifted from 30 to 15°C. Growth rates were measured by counting cells using a ZB1 Coulter counter and plotted using Cricket Graph III (Computer Associates).
To localize γ-tubulin in vivo, log phase cells from strain tG2N-HA or cTTMG-HA were processed for immunofluorescent labeling as described (Gaertig et al., 1995). HA antibody (16B12) was used at a 1:1,000 dilution. Secondary antibodies were goat anti–mouse FITC (Sigma-Aldrich) at a 1:200 dilution or Alexa fluor 568 goat anti–mouse IgG (Molecular Probes) at a 1:500 dilution. DAPI (4′, 6′-diamidino-2-phenylindole) was used for DNA staining at 100 ng/ml. To study the effects of γ-tubulin depletion on BBs, GTU1 knockout heterokaryon progeny or cTTMG-HA cells in CdCl2-free SPP medium were stained with anti-centrin monoclonal or polyclonal antibodies, provided by Dr. Jeffrey L. Salisbury (Mayo Clinic Foundation, Rochester, NY). Anti-glutamic acid polyclonal antibody (R-polyE), a rabbit antibody raised commercially (Alpha Diagnostic International) against a Cys (Glu)9 peptide coupled to keyhole limpet hemocyanin, was used to stain BBs and cilia. MT structures were stained with anti–α-tubulin antibodies (DM1A; Sigma-Aldrich). All antibodies were used at a 1:1,000 dilution. Anti-TFKBP12 polyclonal antibody, provided by Dr. Osamu Numata (University of Tsukuba, Ibaraki, Japan) was used to stain the TFKBP12 (Tetrahymena thermophila FK506 binding protein of 12 kD) at a 1:250 dilution. Secondary antibodies were goat anti–mouse FITC (Zymed Laboratories, Inc.), goat anti–rabbit rhodamine (Zymed), and goat anti–rabbit FITC (Sigma) at 1:200 dilutions. To stain kinetodesmal fibers (KF, also known as striated rootlets), 2–5 × 105 depleted cells were placed on ice for 5 min, fixed in 35% EtOH/0.12% Triton X-100 on ice for 30 min, washed with cold TBS twice (Nelsen et al., 1994), and stained with mAb FI-5D8 (Jerka-Dziadosz et al., 1995) provided by Dr. Joseph Frankel (University of Iowa, Iowa City, IA). Secondary antibodies, Alexa fluor 568 goat anti–mouse IgG (Molecular Probes) and goat anti–rabbit FITC (Sigma-Aldrich), were used at a 1:500 or 1:250 dilution, respectively. Four to eight stacked images of cell were obtained using a Leica TCS SP confocal microscope (PL Apo 100× lens, N.A. 1.4, and 1.8× zoom).
We thank Dr. Joseph Frankel for the anti-KF antibody (5D8) and for his enlightening discussions. We thank Dr. Jeffrey L. Salisbury for the two anti-centrin antibodies and Dr. Osamu Numata for the anti-TFBKP12 antibody. We are grateful to Hiram Lyon for help with confocal microscopy. We thank Josephine Bowen for technical assistance and for critical reading of the manuscript.
This work was supported by National Institutes of Health grant GM-26973.