The fission yeast γ-tubulin is essential for mitosis and is localized at microtubule organizing centers

Microtubules play a fundamental role in cell growth, division, movement, and intracellular transport. Heterodimers of distinct polypeptides, α- and /3-tubulin, are the functional subunits of microtubules and capable of assembly in vitro into microtubules. Recently, however, a new class of tubulin (designated γ-tubulin) was reported to be present in a filamentous fungus Aspergillus nidulans (Oakley and Oakley, 1989). The amino acid sequence of A. nidulans γ-tubulin predicted from the nucleotide sequence of the cloned gene showed approximately 30% identity with α- or β-tubulins. Disruption of the γ-tubulin gene is lethal and nuclear division is blocked in the disruptant (Oakley et al. 1990). Immunofluorescence microscopy revealed that A. nidulans γ-tubulin is localized at the spindle pole body (SPB, a fungal equivalent to the centrosome).

We addressed the question of whether γ-tubulin exists in a distantly related organism, namely, the fission yeast Schizosaccharomyces pombe and, if it exists, whether it plays a role similar to that in A. nidulans. S. pombe is an excellent organism in which to study the dynamic properties of microtubular structures during the cell division cycle (Toda et al. 1984; Hiraoka et al. 1984; Tanaka and Kanbe, 1986; Hagan and Hyams, 1988; Kanbe et al. 1989,1990; Masuda et al. 1990; Hagan and Yanagida, 1990; Hagan et al. 1990; reviewed by Hirano and Yanagida, 1989). The mitotic spindle appears only during mitosis, concomitant with the abrupt disappearance of cytoplasmic microtubules, which are abundant in interphase cells. The structure of the S. pombe mitotic spindle appears to be analogous to that of higher eukaryotes.

In this paper, we report the isolation of γ-tubulin gene from S. pombe, the identification of the gene product and its essential role in mitosis. Immunofluorescence microscopy using antibodies against γ-tubulin revealed the behavior of SPBs (spindle pole bodies) during the cell division cycle. Anti-γ-tubulin antibodies and electron microscopy also identified other putative MTOCs (microtubule organizing centers) that exist at the cell equator and are apparently the centers for cytoplasmic microtubules (Hagan and Hyams, 1988), suggesting that tubulin may be a universal component of MTOCs.

Schizosaccharomyces pombe haploid strains, derivatives of h⁻leu1, were used. A Ura⁻ homozygous diploid 5A/1D (h⁺/h⁻his2/+ leul/leul ura4/ura4 ade6-210/ade6-216) was employed for disruption of the γ-tubulin gene. Culture media for S. pombe were YPD (complete rich medium; 1 % yeast extract, 2 % polypeptone, 2% glucose; 1.6% agar was added for plates), SD (minimal medium; 0.67 % yeast nitrogen base without amino acid, 2% glucose; 1.7% agar was added for plates), and EMM2 (minimal medium; Mitchison, 1970). Escherichia coli was grown in LB (0.5% yeast extract, 1% polypeptone, 1% NaCl, pH 7.5; 1.5% agar was added for plates).

An S. pombe genomic library using the vector pDB248′ (Beach et al. 1982), and containing inserts partially digested with Sau3AI was employed. Hybridization was done at 50 °C using an A. nidulans γ-tubulin cDNA probe (1.5 kb EcoRI fragment of pLO13; Oakley and Oakley, 1989) as a probe and washed in 2×SSC (SSC is 0.15 M NaCl, 0.015 M sodium citrate, pH 7.0) at 50°C. A strongly hybridizing clone (pCT100) was obtained. The nucleotide sequence of pCT100 was determined and it was shown that pCT100 contained only the N-terminal coding region of the S. pombe γ-tubulin gene.

Because of the failure to obtain more DNA clones from the Sau3AI partial library, two other S. pombe genomic DNA libraries were constructed and screened with the probe pCT100. One was a genomic library using the vector pDB248′ and contained Hindlll-digested fragments. The other was a library using sCosl (Evans et al. 1989) as the vector, and containing Sau3AI partially digested genomic DNA. One clone was obtained from each library. Plasmid pCT110, derived from the pDB248’ library, contained a 7.7 kb HindlH fragment that contained all of the coding region except the COOH-terminal nine residues. The other clone (cos29-c2) from the cosmid library contained the fulllength coding region that was subcloned into the PuuII/SacI fragment at the Sad site of vector pSK248, a derivative of pDB248’.

The standard procedures described by Maniatis et al. (1982) were followed for Southern hybridization. Nucleotide sequences were determined by the dideoxy method (Yanisch-Perron et al. 1985) using the plasmid ‘Bluescript’ (Stratagene).

The γ-tubulin gene was mapped on the chromosome by the 8-base recognizing enzyme Notl and a pulse field gel electrophoresis (Fan et al. 1989). Genomic DNA of an S. pombe wild-type h⁻ haploid strain was digested with Not1 enzyme, and the fragments were electrophoretically separated and probed with the ³²P-labelled γ-tubulin gene (pCT110 PvuII/HindIII 3.5 kb fragment).

The gtb1⁺ was disrupted by one-step gene replacement (Rothstein, 1983). Plasmids that contained the ura4⁺ gene inserted between two EcoRV sites of the gtb1⁺ gene were constructed (pCT136 and 137), and were integrated onto the chromosome by transformation of a diploid strain 5A/1D. Ura⁺ diploid transformants were obtained. Genomic DNAs of the transformants were isolated, restricted with EcdRN, run in electrophoresis, and probed with pCT100. Bands of hybridization of the sizes expected for the disruption of the gtb1⁺ gene were obtained for the diploids transformants.

The procedure described by Hagan and Hyams (1988) was followed for preparing and fixing cells. For microtubule staining, the monoclonal antibody raised against Trypanosoma brucei α-tubulin (TAT-1; Woods et al. 1989) and a Texas Red-conjugated goat anti-mouse IgG and IgM antibodies (EY Lab.) were used as primary and secondary antibodies, respectively. Staining of γ-tubulin was enhanced by employing the avidin-biotin-conjugated (ABC) method. Permeabilized cells were incubated with rabbit antibodies against A. nidulans γ-tubuhn and washed three times with PEMBAL (100 HIM Pipes, 1 mM EGTA, 0.1 mM MgSO₄, 1% bovine serum albumin, 0.1% NaN₃, and 100 mM lysine; Hagan and Hyams, 1988) buffer. Cells were further incubated with goat biotinylated antibodies against rabbit IgG (Cappel, 1/200 diluted) and washed three times with PEMBAL buffer once more. Then cells were incubated with FITC-coryugated avidin DN (Vector, 1/200 diluted) for the formation of the biotin-avidin complex. Finally, antibodγ-treated cells were suspended in one drop of phosphate-buffered saline containing 0.5μgml^-1 DAPI, Imgml⁻¹ of bovine serum albumin, and 5% gelatin. For fluorescence microscopy, cells were air-dried onto poly-lysine (1 mgml^-1)-coated coverslips and inverted onto Elvanol mounting media.

Rabbit antiserum against A. nidulans γ-tubulin (Oakley et al. 1990) was used. The fusion protein, which contained all but three N-terminal amino acids of A. nidulans γ-tubulin, was overexpressed in E. coli and purified for raising anti-serum.

Rabbit antiserum was raised against S. pombe γ-tubulin fusion protein using the promoter of bacteriophage T7 genelO (Rosenberg et al. 1987). At first, an S. pombe cDNA library in Agtll (kind gift from Drs V. Simanis and P. Nurse) was screened using the S. pombe γ-tubulin sequence (3.5 kb PcuII-Windlll fragment in pCT110) as the probe. Two cDNA clones were obtained and subcloned into the EcoRI site of Bluescript. By nucleotide sequence determination, one of them, pCT220, was shown to contain the continuous fifth and sixth exons of the γ-tubuhn gene. It was used for constructing the fusion protein expression plasmid. The Spel/EcoRl fragment (containing amino acids 383-421) of pCT220 and the EcoEN/Spel fragment (amino acids 246-383) of pCT110 were ligated at the Spel site and then inserted into, the BamHI site of expression vector pAR3038 (Rosenberg et al. 1987). The resulting plasmid (pCT221) contained the coding frame for 12 N- and eight C-terminal amino acids of the bacteriophage T7 gene 10, plus two amino acids formed by linker nucleotides, and 176 amino acids of the fifth and sixth exons of the S. pombe γ-tubulin gene. The fused protein was produced in E. coli strain BL2HDE3), and purified according to the procedure described by Marston (1987). The procedures to make antisera were described by Hirano et al. (1988).

We isolated the 8. pombe DNAs (pCT100, 110 and 134; Fig. 1A) from three genomic DNA libraries by hybridization using the A. nidulans γ-tubulin cDNA (pLO13; Oakley and Oakley, 1989) as a probe (Materials and methods). Nucleotide sequence determination indicated that all three DNA clones were derived from the same genomic region, and the sequence of pCT134 was found to contain the full coding region. The other two clones lacked the C-terminal region.

Southern hybridization was done using pCT110 as the probe under conditions of low stringency. No band with significant intensity was found except self-hybridizing bands (data not shown). Thus S. pombe γ-tubulin appears to be encoded by a unique gene.

Chromosomal mapping performed by Not1 digestion and PFG electrophoresis (Fan et al. 1989) indicated that the γ-tubulin gene was hybridized to the 630 kb Notl fragment from the right arm of chromosome II (data not shown). We designated the gene for S. pombe γ-tubulin gtb1⁺.

The nucleotide sequence of a 1826 bp portion of pCT134 was determined (Fig. 1B). A coding region interrupted by six short putative introns (consensus sequences indicated by the underlines) predicts an amino acid sequence highly similar to Aspergillus γ-tubulin. It consists of 446 amino acids (M_r 49912). The overall identity between the two γ-tubulins is 77.3 % (Fig. 1C). The amino acid sequence is somewhat less conserved in the C terminus. The identity between γ- and αl-, α2- and β-tubulins of S. pombe was around 30 %.

The gtb1⁺gene was disrupted by one-step gene replacement (Rothstein, 1983) in order to determine whether it is essential.

The Ura⁺ heterozygous diploid transformants (Materials and methods) were sporulated, and 50 tetrads were dissected. No more than two spores were viable, and all the viable spores were Ura⁻ (data not shown). Spores containing the disrupted gtb1⁺ germinated and produced elongated cells, which did not divide. These findings indicate that the γ-tubulin gene is essential for division of S. pombe cells.

A heterozygous diploid that contained the disrupted gtb1⁺ gene was transformed with a multicopy plasmid pCT134 carrying the γ-tubulin gene and a marker gene LEU2. Leu⁺ transformants were isolated and sporulated. Leu⁺ Ura⁺ haploid segregants were frequently (approximately 50 %) obtained, indicating that the above plasmid could rescue the lethality of the haploid gene disruptant. Thus the gene disruptant is rescued by high dosage γ-tubulin gene.

The Ura⁺ Leu⁺ haploid segregants were first grown in a selective medium and then transferred to a rich medium. Plasmid loss (Booher and Beach, 1987) would take place in a fraction of the growing cell population, thus causing the defective phenotype in cells not harboring the plasmid.

Ura⁺ Leu⁺ cells were incubated in rich medium for 8 or 20 h (equivalent of 3 or 8 generation times) at 33 °C, fixed, stained with DAPI (4′,6′-diamidino-2-phenylindole) and treated for immunofluorescence microscopy using the monoclonal antibody TAT1 against Trypanosoma brucei α-tubulin (a gift from Dr Keith Gull; Woods et al. 1989). As shown in Fig. 2A, a fraction of cells (∼ 10%) showed strongly condensed chromosomes by DAPI staining (the middle panel). In these cells, microtubule organization as revealed by the TATI antibody was abnormal (the left panel). The defects could be classified into two types. The very faint, dot-like staining in the center of the condensed chromosomes (above), or an intensely stained, elongated spindle running through the condensed chromosomes (below) that have not undergone chromosome disjunction. These mitotically abnormal cells were seen neither in transformant cells under selective conditions nor in wildtype cells transformed with plasmid containing the γ-tubulin gene, suggesting that the aberrant phenotype was perhaps not’due to the high-dosage expression in a subpopulation of cells containing the plasmid.

Immunobloting was done using affinity-purified antibodies against A. nidulans γ-tubulin (Oakley et al. 1990) and the truncated protein of gtb1⁺ prepared in the present study (Materials and methods). A polypeptide band of M_r 51000 was detected in the S. pombe extracts (data not shown), consistent with the calculated M_r from the determined sequence.

γ-Tubulin localization in wild-type S. pombe cells was determined by immunofluorescence microscopy using antibodies against A. nidulans γ-tubulin. We always found a single weakly fluorescent dot at the periphery of the nucleus in interphase cells (Fig. 2B, left). Superimposition of the immunofluorescence micrographs with DAPI-stained images (right) indicated that the dot in the interphase cells was located at the periphery of hemispherical chromosomal DNA region (Toda et al. 1981; Hirano and Yanagida, 1989).

To prove that the dots represent SPBs, we stained temperature-sensitive cutl-cdcll mutant cells that produced multiple SPBs at restrictive temperature (Uzawa et al. 1990). As shown in Fig. 2C, multiple dots were seen by anti-γ-tubuhn antibodies in the cutl-cdcll mutant cell.

Fig. 2. Immunofluorescence microscopy of γ-tubulin gene disruptant, wild-type and mulitple SPB mutant cells. (A) S. pombe haploid γ-tubulin gene disruptant cells complemented by a multicopy plasmid carrying γ-tubulin gene were grown in rich medium. A fraction of the cells lose the plasmid and display the defective phenotype. Cells showing the mitotic defects were observed by immunofluorescence microscopy using a monoclonal antibody against T. brucei α-tubulin (left) and by DAPI staining of chromosomal DNA (middle). Superimposed images of anti-α-tubulin and DAPI-stained cells are illustrated (right). The arrows indicate the same cell. (B) Immunofluorescence micrographs of wild-type cells stained with anti-γ-tubulin antibodies. Wild-type haploid interphase cells were stained with anti-γ-tubulin antibodies (left) and DAPI (right). Small dots on the periphery of the nucleus, representing SPBs, were revealed by anti-γ-tubulin antibodies. (C) Cells of cutl-cdcll that produced multiple SPBs at restrictive temperature were stained by anti-γ-tubulin antibodies (above) and DAPI (below). Intensely fluorescent dots were seen with anti-γ-tubulin antibodies. Two immunofluorescence micrographs taken at different focal planes are shown. (D) Fluorescence micrographs of wild-type mitotic cells. Anti-α-(left panel) and anti-γ-tubulin antibodies were used to double-stain the wild-type mitotic cells (DAPI was used for counter staining, shown in the right panel). The cells are in different mitotic stages. The dot-like structures are the SPBs, which are located at the nuclear surface (indicated by the arrows in the right panel) and the ends of the spindle, and the cytoplasmic microtubule foci were transiently observed at the cell equator during telophase and cytokinesis (lower two panels). The SPBs and putative MTOCs were rarely in the same focal plane so that the fluorescence intensities of two mitotic SPBs were not equal, due to the difference in focal planes. The cytoplasmic microtubules stained by anti-α-tubulin antibodies (left) appeared to be reorganized from these putative MTOCs. Bara, 10 μ m.

The number of dots, estimated by through focusing, is consistent with that of SPBs determined by serial-section electron microscopy of such mutant cells incubated at restrictive temperature for a similar duration. The immunofluorescence of the dots is highly intense, suggesting that more γ-tubulin is present in each of the mutant SPBs than in wild type, consistent with the result of thin-section electron microscopy showing that SPBs in the mutant cells are larger than those in wild-type mitotic cells (Uzawa et al. 1990).

In mitotic cells containing a spindle in each nucleus but no cytoplasmic microtubule array, the duplicated dots were separated by a distance corresponding to the length of the spindle (left and middle panels in Fig. 2D). These dots are relatively more intense than those seen in the interphase cells and precisely located at the spindle ends. The locations of the dots in the chromosomal DNA regions are indicated by arrows (right panel).

With anti-γ-tubulin antibodies we found other types of dots in the cytoplasm at the time of telophase and cytokinesis (Fig. 2D, below). The positions of the stained dots corresponded to the putative MTOCs previously reported (Hagan and Hyams, 1988). These dots appear at the equatorial plane of the cell after anaphase and prior to cell separation, and are always located at the center of converging microtubules, which are thought to function in re-establishing the cytoplasmic microtubule arrays in the daughter cells.

Consistent with observations by immunofluorescence microscopy, microtubules were observed to converge at the center near the cell equator of dividing cells by freezesubstitution electron microscopy (Fig. 3A-B). In serial sections (Fig. 3C-E), approximately ten microtubules were apparently focused in the central area. A vesicle-like structure (indicated by the arrow in Fig. 3A) was seen at the center. It remains to be seen whether such a structure is a ubiquitous component at the center.

The combined results presented in this study are summarized in Fig. 4. The structures revealed by anti-γ-tubulin antibodies (indicated by the small circle) may be the key cellular components responsible for dynamic changes in microtubular structures during the cell cycle of S. pombe.

We have isolated an S. pombe gene, designated gtb1⁺, the product of which has an amino acid sequence very similar (77.3%) to that of A. nidulans γ-tubulin (Oakley and Oakley, 1989; Oakley et al. 1990). The gtb1⁺ gene is unique in the genome, and its product cross-reacts with antibodies against A. nidulans γ-tubulin. Therefore, we conclude that the gtb1⁺ gene product is the S. pombe γ-tubulin. Considering that these two organisms are evolutionarily quite distant, the functional and structural conservation of γ-tubulin is high, like the levels of α- and β-tubulins (the amino acid sequence similarity between S. pombe and A. nidulans α- and β-tubulins is ∼ 75 %).

The results of the gene disruption indicate that the S. pombe γ-tubulin gene is essential, consistent with the notion that α-, and γ-tubulin genes are functionally distinct (Oakley et al. 1990). The mitotic defects observed in the γ-tubulin gene disruptant are not identical to those of α- and β-tubulin mutants (Toda et al. 1983; Umesono et al. 1983; Hiraoka et al. 1984; Adachi et al. 1986; Kanbe et al. 1990).

In the β-tubulin mutant, spindles are absent, SPBs do not duplicate and the chromosomes condense so that the cells are arrested at a stage similar to prophase. In the tri-tubulin mutant, which contains a small amount of functional tubulin, chromosomes were segregated by a short distance in roughly half of the cells. Anaphase A took place in these cells but anaphase B did not.

Two types of aberrant cells were generated by plasmid loss from the haploid γ-tubulin gene-disruptant strain carrying the plasmid. By anti-u-tubulin staining, one group of cells showed no normal spindle, but a very faint, tiny dot in the center of condensed chromosomes. The location of the dots was reminiscent of that of kinetochores (Hirano et al. 1988) rather than SPBs. The other type of cells displayed intensely fluorescent, long spindles running through the undivided, condensed chromosomes. Plasmid loss takes place asynchronously so that different amounts of tubulin may be present in cells after plasmid loss. The cells that have no spindle may lack γ-tubulin and the cells with a spindle may have a subnormal amount of tubulin. Thus, in the former cells, spindle formation could not be initiated in the absence of γ-tubulin. In the latter cells, on the other hand, the amount of γ-tubulin present in the SPB might be just sufficient to permit spindle formation but not kinetochore microtubule formation, which occurs at a later mitotic stage. This interpretation implies that γ-tubulin is required for SPB duplication, spindle and kinetochore microtubule formation of SPB.

S. pombe SPBs were stained by anti-γ-tubulin antibodies throughout the cell cycle, showing that stagespecific activation of SPB is not needed for γ-tubulin association. This finding also implies that γ-tubulin is bound to SPB without microtubules: the interphase SPB in S. pombe does not appear to be an active MTOC, as judged from electron microscopy as well as immunofluorescence microscopy (Tanaka and Kanbe, 1986; Hagan and Hyams, 1988; Kanbe et al. 1990). The ability to initiate microtubule formation might be suppressed for interphase γ-tubulin, and activated upon entry into mitosis. SPB duplication in S. pombe might require the mitotic activation of γ-tubulin. Recently, Alfa et al. (1991) demonstrated the presence of the p34^cdc2/p63^cdc13 protein kinase at the mitotic spindle of S. pombe. Thus it is possible that mitotic activation of γ-tubulin requires its phosphorylation. The fluoresence intensity of γ-tubulin staining was higher in mitotic cells than in interphase cells, suggesting that the accumulation of γ-tubulin at the SPB may be cell cycle-regulated.

Immunofluorescence microscopy using anti-γ-tubulin antibodies demonstrated, in addition to SPBs, the novel localization of γ-tubulin at the cytoplasmic putative MTOCs. This is particularly significant because it suggests that γ-tubulin could be a component of all MTOCs, not just SPBs. The cytoplasmic microtubule foci appear at the cell equator in post anaphase, initially as two dots in telophase changing to one at the time of cytokinesis (Hagan and Hyams, 1988). The cytoplasmic microtubule array may play a role in the re-establishment of growth polarity and the spatial organization of organelles in S. pombe.

The disappearance of cytoplasmic microtubules before mitosis would necessitate the de novo formation of cytoplasmic MTOCs for the next cell cycle in S. pombe. This situation is strikingly different from the budding yeast Saccharomyces cerevisiae in which cytoplasmic microtubules (and the mitotic spindle) remain throughout the cell cycle. Fission yeast γ-tubulin may be required for the formation of these cytoplasmic MTOCs. The mechanism of formation and detailed structure are of great interest in regard to its spatial and temporal control. If the polarity of the cytoplasmic microtubule array is predetermined in the daughter cells (Fig. 4), most, if not all, microtubule-dependent transport would be directed toward the old cell end from the new one. This may be consistent with the preferential cell growth that occurs at the old end immediately after cell division (Mitchison and Nurse, 1985).

We thank Dr Keith Gull for the monoclonal antibody TATI, Dr Toshio Kanbe for electron microscopy, and Drs Tim Stearns and Marc Kirschner for communicating information prior to publication. This work was supported by the grants from the Ministry of Education, Science and Culture of Japan and the Biodesign Project of the Riken Institute to M. Y. and NIH to B. R. O.