Microtubule nucleating activity of centrosomes in cell-free extracts from Xenopus eggs: involvement of phosphorylation and accumulation of pericentriolar material

Reorganization of the microtubule cytoskeleton during M phase is one of the most drastic changes in intracellular structure during the cell cycle in higher eukaryotes. Cytoplasmic microtubules formed in interphase disappear at the beginning of M phase, and are converted into highly dynamic and organized spindle microtubules.

It has been suggested that microtubule organizing centers (MTOCs), which govern the organization of cytoplasmic microtubules, are modified at the onset of M phase to have a character prerequisite for construction of the mitotic spindle. The centrosome of animal cells consists of a pair of centrioles and pericentriolar material (PCM) (reviewed by McIntosh, 1983; Mazia, 1984, 1987; Brinkley, 1985). PCM has been identified by electron microscopy as amorphous, electron-dense material around the centriole (Robbins et al., 1968), and it is responsible for nucleation of microtubules from the centrosome (Gould and Borisy, 1977). Mitotic centrosomes have a 5-fold higher competence for microtubule nucleation in vitro than interphase centrosomes (Kuriyama and Borisy, 1981). This high competence of the mitotic centrosomes for microtubule nucleation may be essential for the formation of numerous spindle microtubules. In addition, the full construction of spindle microtubules requires the existence of chromatin, which may be necessary for selective stabilization and generation of half-spindle microtubules (Karsenti et al., 1984; Sawin and Mitchison, 1991). In fact, without chromatin, the centrosome can only form a small number of short astral microtubules in mitotic extracts from cytostatic factor (CSF)-arrested Xenopus eggs (Verde et al., 1990).

Little is currently known of the mechanisms regulating centrosomal microtubule nucleating potential during the cell cycle, but two hypotheses have been proposed. The first envisages a cell cycle-dependent variation in the amount of PCM, which is responsible for the microtubule nucleation on the centrosome. PCM has been shown to increase during M phase in mammalian cells (Rieder and Borisy, 1982), and microtubule-organizing granules (PCM in sea urchin eggs) accumulate at the spindle poles in a mitotic cycledependent manner (Endo, 1980). The increase in PCM during M phase may result in an increase in microtubule nucleation sites. However, it is quite unclear how the amount of PCM is regulated.

It is attractive to speculate that the cytoplasm has the ability to supply microtubule nucleating material to the centrioles or centrosomes in a cell cycle-dependent manner. It has been indicated that egg cytoplasm is able to supply microtubule nucleating material to the centriole or the centrosome in vivo. Kuriyama and Kanatani (1981) reported that sperm centrioles could form astral microtubules when they were injected into starfish eggs, while they failed to form astral microtubules in an in vitro system containing purified tubulin. This suggests that centrosomal proteins in the egg cytoplasm gather around the sperm centriole, thereby acquiring microtubule nucleating activity in vivo. Unfortunately, little is known about the kinetics of this process during the cell cycle.

The second idea is that activation of the centrosome is through the action of M phase-activated phosphorylation. Most simply, the microtubule nucleating potential of the centrosome is activated through the phosphorylation of the centrosomal material responsible for microtubule nucleation. This idea is supported by the recognition of centrosomes in mitotic cells of many species by monoclonal antibodies against M phase-specific phosphoproteins (Vandre et al., 1984; Kuriyama, 1989). In addition, Centonze and Borisy (1990) reported that one such antibody can block the microtubule nucleating activity of the centrosomes from mammalian cells.

In the G₂/M transition, protein kinases are activated and protein phosphorylation becomes prominent (Maller et al., 1977; Karsenti et al., 1987). Maturation promoting factor (MPF) (Masui and Markert, 1971; Kishimoto and Kanatani, 1976; Kishimoto, 1988), which is a complex of p34^cdc2 kinase and cyclin B (Gautier et al., 1988; Dunphy et al., 1988; Labbé et al., 1988; Dreatta et al., 1989; Nurse, 1990), plays a central role in G₂/M transition. MAP kinase, originally found as a mitogen-activated serine/threonine kinase (Ray and Sturgill, 1987, 1988; Hoshi et al., 1988; Gotoh et al., 1990), is also activated during M phase of Xenopus oocytes (Gotoh et al., 1991a). MAP kinase is activated downstream of p34^cdc2/cyclin B (Gotoh et al., 1991b).

These M phase-activated kinases have been shown to play a crucial role in the conversion of microtubule dynamics at the onset of the mitosis (Verde et al., 1990; Gotoh et al., 1991a). It has also been suggested that they are involved in the activation of the centrosome during M phase. p34^cdc2 associates with the centrosome in animal cells (Riabowol et al., 1989; Bailly et al., 1989) and with the spindle pole body in yeast which is equivalent to the centrosome (Alfa et al., 1990). In addition, a B-type cyclin homologue of fission yeast p63^cdc13 is also associated with spindle pole body (Alfa et al., 1990), and cyclin B with mitotic spindles in human cells (Pines and Hunter, 1991) and the polar region of mitotic spindles in Drosophila embryos (Maldonado Codina and Glover, 1992).

Using cell-free extracts from Xenopus eggs, we have studied the regulation of the microtubule nucleating potential of the centrosome, in the light of the variation in amount of PCM and the cell cycle-dependent control by protein phosphorylation, since cell-free extracts from Xenopus eggs have advantages for analysis of regulation of microtubule organization (Karsenti et al., 1984; Lohka and Maller, 1985; Gard and Kirschner, 1987a; Verde et al., 1990; Belmont et al., 1990; Gotoh et al., 1991a). We found that extracts from Xenopus eggs are able to cause a time-dependent increase in microtubule nucleation from isolated mammalian centrosomes. PCM around the centrosome increased in egg extracts in vitro with the activation in microtubule nucleation. The ability of the extracts to increase microtubule nucleation from the centrosome is probably controlled by phosphorylation, since the addition of protein kinase inhibitors into the extracts totally blocked the increase in microtubule nucleation from KCl-treated, inactive centrosomes. The activity of the extracts is regulated in a cellcycle dependent manner, mitotic extracts having approximately a 3.5-to 5-fold greater microtubule nucleating potential of KCl-treated centrosomes than interphasic extracts. Furthermore, the ability of interphasic extracts was strikingly enhanced by adding p34^cdc2/cyclin B to the extracts, but not by MAP kinase, a kinase downstream of p34^cdc2/cyclin B. We propose that the microtubule nucleating potential of the centrosome is activated during the cell cycle by two different pathways: (1) accumulation of PCM to the centrosome and (2) phosphorylation of centrosomal materials, such as the microtubule nucleating material or the putative PCM accumulating factor.

Cytoplasmic extracts arrested in a mitotic state were prepared from Xenopus eggs arrested at metaphase II (the second meiotic metaphase) by CSF. Cell-free extracts were prepared according to the method of Verde et al. (1990) with some modifications. The eggs were washed twice with an extraction buffer (100 mM potassium acetate, 2.5 mM magnesium acetate, 60 mM EGTA, 5 μg/ml cytochalasin B, and 1 mM DTT, pH 7.2), and were crushed in a minimum volume of the extraction buffer by centrifugation at 30,000 g and 2°C for 10 min. The upper lipid layer was completely removed by aspiration and the middle, clear supernatant was collected. The supernatant was mixed with 1/100 volume of an ATP regeneration system (100 mM ATP, 1 M creatine phosphate and 8 mg/ml creatine phosphokinase (Sigma)), and again centrifuged in a Beckman TL-100 at 350,000 g and 2°C for 25 min. The lipid layer was aspirated and post-ribosomal highspeed supernatants were collected. For the preparation of extracts used, we centrifuged at higher speeds than the standard protocol to remove membranous and particulate materials. This enables realtime observation under dark field microscopy. “PA20-30 extracts” were prepared from the eggs driven into an interphase state in MMR (0.1 M NaCl, 2 mM KCl, 1 mM MgSO₄, 2 mM CaCl₂, 5 mM HEPES, 0.1 mM EDTA, pH 7.8) by a 4-5 s electric shock (12 V, AC) followed by incubation for 20-30 min at 20°C. The electrodes were separated by 3 cm. This period corresponds to the stage when the sperm aster is growing, histone H₁ kinase activity and MAP kinase activity have decreased and DNA replication has started. These extracts were stored at −80°C until use.

Centrosomes were isolated from CHO cells or L5178Y cells as described previously (Mitchison and Kirschner, 1984; Bornens et al., 1987). The quality of centrosome preparations was analyzed by electron microscopy, aster formation in purified tubulin and double labeling with anti-centrosome antibody and anti-tubulin antibody along with phase contrast microscopy as described by Bornens et al. (1987). From the yield of centrosomes and the final protein content, purification was assumed to be approximately 10³-fold when one cell contains one centrosome. Isolated centrosomes (0.5-1 μl, 10⁷/ml) were added to 15 μl of the extracts. The mixture was mounted onto cleaned coverslips on a Parafilm sheet, and incubated in a moist chamber. Microtubules formed from centrosomes usually stick onto the coverslips during a brief incubation (less than 1-2 min), presumably through the action of microtubuleassociated motor proteins in the extracts.

In the “extracts exchange assay”, the extracts containing 1 μl of KCl-treated L5178Y centrosomes (prepared as described later) were incubated on coverslips for the indicated time at 20°C. The extracts on coverslips were removed by a micropipet. Then we carefully added 100 μl of PEM1 (100 mM PIPES, 1 mM EGTA and 0.5 mM MgCl₂, pH 6.8) containing 0.5 mM GTP at 0°C to the coverslips where the regrown asters were stuck, and very carefully washed by gently pipetting the solution up and down several times. Liquid on the coverslips was carefully aspirated, and the coverslips were washed once more. After the wash, 15 μl of fresh interphase extracts were quickly applied to the coverslips and incubated for 1-2 min at 20°C to form microtubules from the centrosomes. In the assay using purified tubulin, the experiment was done as described above, except that we used native CHO centrosomes and performed a second incubation at 37°C for 20 min in 80 mM PIPES, 1 mM MgSO₄, 1 mM EGTA, 1 mM GTP, pH 6.8, containing DEAE-purified tubulin (3.8 mg/ml) prepared from porcine brains as described (Masuda et al., 1990).

Use of PA20-30 extracts for the second incubation provides several advantages over purified tubulin for detecting quantitative differences in the ability of extracts to increase microtubule nucleation from centrosomes. Interphasic extracts (PA20-30 extracts) contain ‘XMAP’ (Gard and Kirschner, 1987b), which enables microtubules to grow very fast, so that we can detect microtubules long enough even in a very brief incubation (1-2 min). Owing to such a brief incubation, this system worked quite well for detecting the cell cycle dependence and effects of kinase addition to the extracts, in spite of the existence of phosphatase activities in PA20-30 extracts. In fact, using PA20-30 extracts, we obtained similar results to those using purified tubulin (Fig. 7). The brief incubation also prevents detachment of astral microtubules from coverslips, which was often seen in the experiment using purified tubulin. In addition, using PA20-30 extracts, we can easily distinguish individual astral microtubules, well expanded on coverslips against a very low fluorescence background, in contrast to the high fluorescence background in the experiment using purified tubulin, which made individual microtubules obscure. This advantage of PA20-30 extracts may be due to the existence of high concentrations of proteins other than tubulin, and motor proteins which ensure adhesion of the astral microtubules to coverslips.

After the incubation, excess extracts were very carefully removed by micropipet. The coverslips were slowly immersed in PEM1 containing 0.1% glutaraldehyde, and were incubated for 3 min at room temperature. After adequately removing the fixative, the coverslips were then postfixed in −20°C methanol for 5 min. These steps were performed very carefully so as not to form aggregates of proteins in the extracts on coverslips and also not to damage microtubules. Fixed microtubules were then treated with freshly prepared 1 mg/ml NaBO₄ in PBS (20 mM phosphate/KOH, 150 mM NaCl, pH 7.4) for 5 min at room temperature. Immunofluorescence labeling was performed with a monoclonal anti-α tubulin antibody (1/200 dilution, Amersham) and a human anticentrosome antiserum (1/100 dilution, a gift from Dr. Y. Moroi (Moroi et al., 1983) as described in elsewhere (Ohta et al., 1988). Secondary antibodies used were a rhodamine-labeled goat antimouse IgG (1:300 dilution) and a fluorescein-labeled goat antihuman IgG antibody (1:100). Samples were mounted in Mowiol (Hoechst) and observed with a Nikon FX microscope. Pictures were taken on Kodak TriX film. Microtubule length was measured on prints by a digital curvimeter type “D” (Uchida Yohkoh, Tokyo, Japan). Microtubule numbers were counted under the microscope or on negatives enlarged by a low magnification binocular microscope. Growing microtubules at room temperature were directly observed under dark field illumination using a Nikon Optiphoto microscope equipped with an oil immersion condenser lens and a mercury arc lamp.

KCl-treated centrosomes were prepared as described by Klotz et al. (1990). 50 μl of the isolated L5178Y or CHO centrosome preparation (10⁷/ml in 50% sucrose) were diluted with 50 μl of PEM2 (10 mM PIPES, 0.5 mM MgCl₂, 1 mM EGTA, pH 6.8) containing 2 M KCl, and were incubated for 30 min at 0°C. The treated centrosomes were dialyzed against PEM2 for 1 h at 2°C. For the “extracts exchange assay”, we added 1 μl of the KCl-treated centrosomes into 15 μl of extracts, and incubated as described above.

CHO centrosomes (10 μl) were incubated in 90 μl of extracts for 7 and 30 min at 20°C in 1.5 ml microcentrifuge tubes. After the incubation, the centrosomes were fixed by adding 900 μl of PEM1 containing 0.25% glutaraldehyde and incubated for 5 min at room temperature. The fixed centrosomes were sedimentated by a microcentrifuge at 30,000 g for 20 min. The precipitates were washed twice in PEM3 (20 mM PIPES, 0.5 mM MgCl₂ and 1 mM EGTA, pH 6.8), and were postfixed with 1% OsO₄ for 30 min. The precipitates were removed and dehydrated in ethanol and embedded in Rigolac mixture as described elsewhere (Endo, 1980). Thin sections (80 nm-thick) were stained with uranyl acetate and lead citrate, and were observed with a Hitachi H-7000 electron microscope.

2 μl of CHO centrosome preparation (10⁷/ml) were added to 20 μl of extracts, and the mixture was incubated at 20°C. Then 88 μl of the extraction buffer, without cytochalasin B and DTT but containing 20 μM Nocodazole, were added to the reaction mixture. After incubating at 0°C for 30 min, the mixture was combined with 1.4 μl of 8% glutaraldehyde (EM grade, final 0.1%), and was incubated for 10 min at 20°C. Fixed samples were overlaid on 3 ml cushions of PEM3 containing 25% glycerol, and were centrifuged onto round coverslips by a RPS40T-2 rotor (Hitachi) at 12,000 g for 15 min according to the method described by Mitchison and Kirschner (1984). Samples on coverslips were postfixed in methanol at −20°C for 5 min, treated in PBS containing 1 mg/ml NaBO₄ for 3 min, and processed for immunofluorescence labeling by affinity-purified, rabbit anti-Schizosaccharomyces pombe γ-tubulin antibody (1:10 dilution) and 5051 human anticentrosome antibody (1:200) (these antibodies were a gift from Dr. Masuda, The Institute of Physical & Chemical Research (RIKEN)). The anti-S. pombe γ-tubulin antibody used cross-reacts with γ-tubulin of frog and centrosomes from Chinese hamster and human cells. The secondary antibodies used were a fluoresceinlabeled goat anti-human (1:100) and a rhodamine-labeled goat anti-rabbit (1:300). For comparison, we used the same exposure time for negatives and for prints in all cases.

p34^cdc2/cyclin B was purified from starfish maturing oocytes using a suc1-conjugated Sepharose column described elsewhere (Brizuela et al., 1987; Labbé et al., 1989) with some modifications: this preparation contains cyclin B (Labbé et al., 1989; Tachibana et al., 1990), but not cyclin A (E. Okumura and T. Kishimoto, unpublished data). M phase-activated MAP kinase from Xenopus eggs was purified as described (Gotoh et al., 1991a). We added 1.5 μl of purified p34^cdc2/cyclin B (20.1 pmole phosphate transferred to histones/μl min⁻¹) or purified MAP kinase (16 μg/ml, 5.2 pmole phosphate transferred to myelin basic protein/μl min⁻¹) into 15 μl of interphase extracts with the centrosomes. Protein concentration was determined by the method of Bradford (1976) using γ-globulin as a standard. The effect of Staurosporine on kinase activity was examined as follows. Staurosporine (Kyowa Hakko) and/or 2 μl histone (Sigma, type III-S, 12 mg/ml) were added to 6.5 μl of interphase extracts containing 0.33 (−histone) or 1 (+ histone) μCi/μl of [γ-³²P]ATP, and the mixture was incubated for 40 min (−histone) or 30 min (+ histone) at 20°C. The reaction was stopped by adding 7 μl of SDS-PAGE sample buffer. The samples were analyzed by SDS-PAGE followed by autoradiography (without addition of histone) or by counting Cherenkov radiation of ³²P in the histone H₁ bands. Histone kinase activity of the extracts was measured as described (Félix et al., 1989).

We found that the microtubule number per centrosome was increasing dramatically during the incubation of isolated mammalian centrosomes in interphasic extracts prepared 20-30 min after electric activation of CSF-arrested Xeno pus eggs. We call this type of extract “PA20-30 (post-activation 20-30 min) extracts”. Fig. 1 shows the increase in number of astral microtubules during the incubation of isolated CHO centrosomes in PA20-30 extracts. We confirmed that each aster includes the centrosome, detected by a human autoimmune antiserum against the centrosome from a scleroderma patient (Fig. 1D). Microtubule number per centrosome was increased from 53.4 ± 4.3 (s.e.) (6 min, n = 70) up to more than 200 (∼270) (20 min, n = 70) during the incubation in the PA20-30 extracts. We also confirmed a similar increase in number of microtubules from a single centrosome in PA20-30 extracts under a real-time observation by dark field microscopy. Though the increase in microtubule number from the centrosome took place in vitro, it can be said to be reasonable, since in this period eggs are at the stage during which the sperm aster is growing and enlarging from the sperm centrioles which cannot form astral microtubules before their entry into the egg. Fig. 1E indicates time courses of microtubule number per centrosome and mean length of microtubules from centrosomes during the incubation. While microtubule number per centrosome was markedly increasing with time in PA20-30 extracts, mean microtubule length was generally constant after 5 min incubation in the same extracts, ruling out the possibility that the conditions of the extracts were becoming more advantageous for microtubule formation during the incubation.

Using centrosomes stripped of their microtubule nucleating activity by 1 M KCl treatment, we verified that the increase in microtubule number per centrosome is due to the addition of new microtubule nucleation sites on the centrosome. Klotz et al. (1990) reported that treatment of the centrosomes with high concentration of KCl or NaCl destroys the microtubule nucleating activity of PCM. We checked again that no microtubule could be formed from the 1 M KCl-treated CHO centrosomes in purified tubulin at 37°C. However, when we incubated KCl-treated centrosomes in PA20-30 extracts at 22°C, the microtubule nucleating activity recovered and increased rapidly after a 5 min lag (Fig. 2A). These observations suggest that the increase in microtubule number per centrosome in the extracts is due to the addition of new microtubule nucleation sites (activation of microtubule nucleation potential) on centrosomes.

To study whether or not centrosomes show qualitative or quantitative changes during the activation of microtubule nucleation, we examined the changes in signal intensity of the centrosome labeled by two centrosome markers: anti-

S. pombe γ-tubulin antibody which cross-reacts with Xeno pus γ-tubulin and mammalian centrosomes, and 5051 human anti-centrosome autoantibody from a scleroderma patient (Calarco-Gillam et al., 1983). γ-tubulin is a third member of the tubulin superfamily, and is a common component of spindle pole bodies of Aspergillus nidulans (Oakley and Oakley, 1989; Oakley et al., 1990) and S. pombe (Horio et al., 1991), and of centrosomes in Xeno pus, Drosophila and mammalian cells (Zheng et al., 1991; Stearns et al., 1991). γ-tubulin and the antigen of the 5051 antibody are shown to be components of PCM of mammalian centrosomes (Zheng et al., 1991; Stearns et al., 1991; Calarco-Gillam et al., 1983). Based on genetic and immunochemical evidence, γ-tubulin was suggested to be involved in microtubule nucleation on the spindle pole body and centrosomes (Oakley et al., 1990; Joshi et al., 1992).

Anti γ-tubulin (Fig. 3B, E and I) and 5051 anti-centrosome antibody (Fig. 3A, D, G and H) specifically reacted with CHO centrosomes which had been treated for 0 min (A,B) and 25 min (D,E) in PA20-30 extracts. After 25 min incubation in PA20-30 extracts, signals of centrosomes became stronger (compare A, B with D, E). The size of the centrosome signal after the treatment is roughly estimated to be about twice as large (2 μm) as that without incubation (1 μm). Interestingly, the enhancement of signal intensity was also detected even when we added 20 μM of Nocodazole to PA20-30 extracts (Fig. 3G), indicating that the accumulation is not due to a retrograde transport of materials along the astral microtubules. We confirmed that few signals by 5051 anti-centrosome (Fig. 3H) and anti-γ-tubulin antibody (Fig. 3I) can be seen without adding centrosomes to PA20-30 extracts.

To examine the change in the fine structure of centrosomes during the incubation, we observed the centrosomes treated in PA20-30 extracts by electron microscopy. We found that incubation of CHO centrosomes with PA20-30 caused a marked increase in the amount of PCM around the centrioles (Fig. 4). After 30 min incubation (Fig. 4B,D), PCM had an average thickness of 127±29 nm (s.d., average of values on 8 well-defined centrioles in more than 200 random sections), having a value about double that after 7 min (56 ±24 nm, s.d., 4 centrioles) (Fig. 4A,C).

We next studied the relationship between protein phosphorylation and the activation of microtubule nucleating potential of the centrosome, because many lines of evidence have suggested that phosphorylation is involved in the regulation of the centrosomal function. We examined the effect of protein kinase inhibitors on the increase in microtubule nucleating potential of KCl-treated L5178Y centrosomes. PA20-30 extracts had detectable kinase activities including low histone H₁ kinase activity (1-4 pmole/μl min⁻¹). Staurosporine (Tamaoki et al., 1986) and K252a (Kase et al., 1987), first introduced as strong inhibitors for protein kinase C, inhibit various other kinases at higher concentrations. When increasing concentrations of Staurosporine or K252a were added to PA20-30 extracts containing KCl-treated centrosomes, the increase in microtubule nucleating potential of the KCl-treated centrosomes was inhibited in a dosedependent manner (Fig. 5A). Fig. 5B indicates that Staurosporine actually inhibited bulk protein phosphorylation in PA20-30 extracts in roughly the same dose-dependent manner. Phosphate transferred to exogenous histone H₁ in the PA20-30 extracts decreased in the presence of 3 μM Staurosporine to 36% of the control without Staurosporine. Numerous free microtubules were formed in the presence of inhibitors (data not shown), ruling out the possibility that the kinase inhibitors blocked microtubule assembly in PA20-30 extracts. Furthermore, addition of these kinase inhibitors did not block the microtubule nucleation from intact centrosomes in PA20-30 extracts, although it prevented the time-dependent increase in microtubule number per centrosome.

We examined the cell-cycle dependence of the ability of extracts to increase the microtubule nucleating potential of KCl-treated centrosomes. The number of microtubules nucleated from a centrosome depends on both microtubule nucleating potential (i.e. number of microtubule nucleation sites) of the centrosome and the environmental conditions for microtubule formation in solution. In fact, microtubule formation is suppressed in mitotic extracts, but not in interphasic extracts (Karsenti et al., 1984; Verde et al., 1990). This point will result in a misleading underestimate of the ability of mitotic extracts.

In order to compare accurately the intrinsic ability of interphasic and mitotic extracts to increase microtubule nucleating potential, we have developed an “extract exchange assay” (Fig. 6). In this assay, we first incubated KCl-treated centrosomes in PA20-30 interphasic extracts or mitotic extracts, followed by a removal of the extracts and washing with a buffer solution. Then, to form microtubules under the same polymerization conditions, we incubated them in purified tubulin (Fig. 7E,F) or fresh PA20-30 extracts (Figs 6, 7A-D). Both experiments using purified tubulin and PA20-30 extracts gave similar results: mitotic extracts have a greater ability to activate microtubule nucleating potential from centrosomes than interphasic extracts (Fig. 7). Despite its simplicity, the experiment using purified tubulin was not good for quantitative analysis (e.g. counting the number of nucleated microtubules) mainly because of a high fluorescence background due to non-specific adsorption of purified tubulin to coverslips and also the tendency of asters to detach from coverslips during a rather long incubation (20 min instead of 1-2 min in PA20-30 extracts). Since the use of PA20-30 extracts for the second incubation can overcome these difficulties (see Materials and methods), and give similar results to those obtained with purified tubulin, we employed an “extracts exchange assay” using PA20-30 extracts in the following experiment.

Fig. 7 indicates the results of the extract exchange assay for cell-cycle dependence in the ability of the extracts to activate KCl-treated L5178Y centrosomes. Centrosomes preincubated in mitotic extracts (Fig. 7C,D) nucleated many more microtubules after the extracts exchange than centrosomes preincubated in PA20-30 extracts (Fig. 7A,B). Mean microtubule number per centrosome in mitotic extracts was roughly estimated to be 3.5-fold (118.4±15.6: 33.8±3.9, n = 25) after 5 min and about 5-fold (127±11.9: ∼600, n = 25) after 25 min, higher than that in PA20-30 extracts. The histone H₁ kinase activity of mitotic extracts (14.9 pmole/μl min⁻¹) was about 5-fold higher than that of PA20-30 extracts (2.9 pmole/μl min⁻¹). The quantitative difference in the ability of both types of extract to activate microtubule nucleation is generally in good agreement with the difference in microtubule nucleating activity in vitro between interphasic and mitotic centrosomes from CHO cells (Kuriyama and Borisy, 1981).

We also examined the accumulation of PCM in mitotic extracts by electron microscopy and immunofluorescence staining by anti-γ-tubulin and 5051 anti-centrosome antibodies. Accumulation of PCM in mitotic extracts is supported by the results of both the immunofluorescence (Fig. 3C,F) and the electron microscopy (Fig. 4E,F). Interestingly, the increase in signal intensity of centrosomes labeled by anti-γ-tubulin and 5051 anti-centrosome antibodies seemed to occur equally in both PA20-30 interphasic extracts and CSF-arrested mitotic extracts (compare Fig. 3E,F). In addition, from the results of electron microscopy, it seems that there is little difference in the amount of PCM in mitotic extracts and in PA20-30 extracts after 30 min incubation. After 25 min in mitotic extracts, PCM had an average thickness of 131 ±40 nm (s.d., average of values on 50 well-defined centrioles in more than 500 random sections), which was similar to the results obtained by interphasic extracts, whereas we found that PCM of the centrosomes incubated in mitotic extracts nucleates many more microtubules than those in PA20-30 extracts (Fig. 4B,F).

The results in Figs 5 and 7 suggest that the ability of the extracts to increase microtubule nucleating potential of the centrosome is under the control of phosphorylation mediated by M phase-activated kinases. We then added two major M phase-activated kinases, p34^cdc2/cyclin B and MAP kinase, to PA20-30 extracts, and examined the ability of the extracts to activate KCl-treated L5178Y centrosomes, by using the extracts exchange assay described above. MAP kinase is an M phase-activated kinase in the downstream of p34^cdc2/cyclin B (Gotoh et al., 1991b), and was suggested to be involved in the conversion of microtubule dynamics at the onset of M phase (Gotoh et al., 1991a) as well as p34^cdc2/cyclin B (Verde et al., 1990). We found that addition of purified p34^cdc2/cyclin B to PA20-30 extracts caused marked promotion of the ability of the extracts to increase microtubule nucleating potential; the microtubule number per centrosome was more than 200 (239±18.4, n = 25), in contrast to a control with the buffer solution alone (45.6±2.7, n = 25) (Fig. 8A,B), while MAP kinase had little effect (80.8±7.9, n = 30, Fig. 8D), compared with buffer solution alone (102.9±10, n = 30). The ability of p34^cdc2/cyclin B and the basal activity of PA2030 extracts to increase microtubule nucleating potential of KCl-treated centrosomes was almost totally inhibited by 5 μM Staurosporine (Fig. 8C). When we added p34^cdc2/cyclin B or MAP kinase to a simple reconstitution system using purified tubulin and isolated centrosomes in the presence of ATP, little effect could be detected (data not shown). These results suggest that phosphorylation, presumably mediated by an indirect action of p34^cdc2/cyclin B but not by MAP kinase, is involved in the increase of microtubule nucleating potential of centrosomes in Xenopus mitotic extracts.

We found that the microtubule nucleating activity of isolated centrosomes is activated with time of incubation in Xenopus egg extracts. Most likely, the increase in microtubule nucleating activity is due to an increase in microtubule nucleating sites. Mean length of microtubules from the centrosomes in PA20-30 extracts was generally constant during the incubation (Fig. 1E). This rules out the possibility that improvement of the conditions for microtubule growth in the extracts caused an apparent increase in microtubule number per centrosome. The isolated centrosomes used here can nucleate 33.7±1.5 (s.e., n = 80) microtubules in a simple system with purified brain tubulin even in the presence of brain MAP2 promoting microtubule assembly (Hoshi et al., 1992). Therefore, it is implausible that more than 200 microtubules could be formed from a single centrosome used here without increasing microtubule nucleation sites. In addition, the extracts were able to restore microtubule nucleating activity of centrosomes stripped of nucleating activity by KCl treatment. Increase in microtubule nucleation sites is also reinforced by the increase of PCM revealed by immunostaining with anti-γ-tubulin and 5051 anti-centrosome antibodies, and by electron microscopy.

From in vitro results, we can speculate that the microtubule nucleating potential of the centrosome is activated during the cell cycle through at least two different pathways: (1) accumulation of PCM to centrosomes and (2) phosphorylation of centrosomal materials. Correlation between the time-dependent increase in microtubule nucleation and accumulation of PCM in vitro suggests that the assembly of PCM around the centrioles from the cytoplasm plays an important role in the regulation of microtubule nucleating potential of the centrosome. Presumably, accumulation of PCM around the centrioles plays a role in the activation of the centrosome during M phase in vivo. Rieder and Borisy (1982) have shown that PCM of PtK2 cells increases in accordance with the increase in microtubule nucleating activity of the centrosome in the G₂/M transition. Endo (1980) has shown that the microtubule-organizing granules (PCM in sea urchin eggs) accumulate at the poles in a mitotic cycle-dependent manner. However, from our results, the kinetics of accumulation of PCM seem similar in both type of extracts (Figs 3E,F and 4B,F). Possibly, the accumulation of PCM in egg extracts in vitro may not necessarily be controlled in a cell cycle-dependent manner. However, this issue should be examined further by other more quantitative assay systems.

The results also imply that centrosomes are activated during M phase through phosphorylation of centrosomal components. Addition of protein kinase inhibitors into interphasic extracts and extracts driven into a mitotic state by the addition of p34^cdc2/cyclin B totally blocked the recovery and increase in microtubule nucleating potential of KCltreated centrosomes. On the other hand, addition of active p34^cdc2/cyclin B kinase into interphasic extracts promoted the increase in microtubule nucleation from KCl-treated centrosomes as described below. These are consistent with previous reports. Protein phosphorylation is activated during M phase in Xenopus eggs (Maller et al., 1977; Karsenti et al., 1987). Monoclonal antibodies to M phasespecific phosphoproteins recognize the mitotic centrosome (Vandré et al., 1984; Kuriyama, 1989), having maximal activity during the cell cycle (Kuriyama and Borisy, 1981). There should be at least two possibilities for the process regulated by M phase-activated phosphorylation: (1) tubulin interaction on centrosomes and (2) accumulation of PCM to the centrosome. If the former possibility is true, phosphorylated microtubule nucleating material on centrosomes is activated to bind tubulin stronger. The latter requires activation of putative PCM-accumulating factor, and is not necessarily supported by our results, since many more microtubules seem to emanate from the PCM in mitotic extracts than in the interphasic extracts while there seems to be little difference in the amount of PCM in both type of extracts (see Figs 3 and 4). However, this possibility must be examined more carefully in future work.

The proposed activation mechanism of the centrosomal activity by M phase-activated phosphorylation seems rather self-contradictory, because phosphorylation has been shown to suppress microtubule assembly during M phase (Verde et al., 1990; Gotoh et al., 1991a). However, these apparently opposite effects of phosphorylation reactions can explain well the fact that in vivo the microtubule organization in mitotic phase is highly dependent on the centrosomes, which are preferential sites for microtubule nucleation (see Fig. 9). Activated centrosomes during M phase are supposed to be competent for the construction of numerous spindle microtubules if the chromatins are in the vicinity of the centrosome, since the existence of chromatin is necessary for the selective stabilization and the generation of half-spindle microtubules (Karsenti et al., 1984; Sawin and Mitchison, 1991).

We demonstrated that the increase in microtubule nucleating potential is strikingly enhanced by the addition of p34^cdc2/cyclin B to interphasic extracts. This idea is consistent with previous data that the microtubule nucleating activity in vitro reaches a maximum at M phase (Kuriyama and Borisy, 1981), where the kinase activity of MPF (p34^cdc2/cyclin B) is also maximal and p34^cdc2 is localized in the mitotic centrosome and spindle pole body (Vandre et al., 1984; Kuriyama, 1989; Riabowol et al., 1989; Bailly et al., 1989; Alfa et al., 1990). In addition, cyclin B associates with spindle pole body (Alfa et al., 1990) and with the polar region of mitotic spindles in Drosophila embryos, whereas cyclin A mainly associates with chromatin during M phase (Maldonado-Codina and Glover, 1992). In addition, most recent work by Masuda et al. (1992) revealed that interphase extracts from Xenopus eggs gained an ability to convert inactive, interphase spindle pole body of fission yeast to a competent state for microtubule nucleation when the interphase extracts were incubated with Δ90 sea urchin cyclin B1. However, purified p34^cdc2/cyclin B did not directly activate microtubule nucleation from centrosomes in a purified in vitro system (data not shown). Therefore, centrosomes should be activated through a downstream pathway regulated by p34^cdc2/cyclin B, as claimed by Masuda et al. (1992).

In contrast to our results, Belmont et al. (1990) reported that there was no increase in microtubule number from centrosomes incubated in extracts driven into M phase by the addition of Δ90 sea urchin cyclin B1 into interphase extracts prepared by a complete destruction of cyclins. In addition, Buendia et al. (1992) have claimed that addition of bacterially produced cyclin A into cyclin-depleted interphase extracts caused approximately a 1.8-fold increase in the microtubule nucleating activity of the centrosomes complemented in the extracts, whereas the addition of Δ90 sea urchin cyclin B1 or purified starfish p34^cdc2/cyclin B complex did not, although they had mentioned previously that the addition of p34^cdc2/cyclin B in interphasic extracts from electrically activated Xenopus eggs brought about a promotion of apparent microtublule nucleating activity of the centrosome (Verde et al. 1990). They concluded that a cyclin B-dependent phosphorylation of centrosomal antigens may not be essential in the activation processes of the centrosome during M phase.

The reason for the discrepancy is not clear at present. It is possible that the preparation of interphasic extracts is of prime importance in the activation assay of the centrosome by p34^cdc2/cyclin B. Judging from the time when we prepared the PA20-30 extracts (20-30 min after the electrical activation), our PA20-30 extracts could to have trace amounts of cyclins. However, the interphase extracts prepared by Belmont et al. (1990) and Buendia et al. (1992) lack cyclin A and B completely after the destruction of cyclins and inhibition of protein synthesis by cycloheximide. So it is possible that cyclin-depleted interphase extracts are not competent for the full activation of microtubule nucleating activity of the centrosome by p34^cdc2/cyclin B. And it is also possible that p34^cdc2/cyclin B can cause full activation of the centrosome only with the action of cyclin A-associated kinase in cell-free extracts. This point should be examined by adding cyclin A and B together in the cyclin-depleted interphase extracts.

In contrast to the p34^cdc2/cyclin B, the addition of MAP kinase, an M phase-activated kinase downstream of p34^cdc2/cyclin B (Gotoh et al., 1991b), had little effect on the ability of interphase extracts to recover and increase microtubule nucleating potential of the centrosome. Therefore, the pathway by MAP kinase only acts in converting microtubules to a dynamic state in the microtubule reorganization at G₂/M transition.

The mechanism regulating microtubule nucleation of the centrosome during the cell cycle must be highly complicated. Our data imply at least two complex pathways: accumulation of PCM to centrosomes and p34^cdc2/cyclin Bmediated phosphorylation. The molecular basis for these pathways should be examined extensively in future work.

We are grateful to Dr H. Masuda (Riken Institute) for his careful discussion on this manuscript, and for his gifts of affinity-purified anti-S. pombe γ-tubulin, 5051 anti-centrosome antibody, MPM2 antibody, and DEAE-purified tubulin. We also thank G. Fujii, R. Miyoshi and Prof. K. Shiokawa (Univ. of Tokyo) for kind instructions on handling Xenopus eggs and animals; Y. Miyata and Dr I. Yahara (Tokyo Metropolitan Institute of Medical Science) for a kind gift of L5178Y cells; Dr Y. Moroi (Itoh National Hospital) for providing us with human anti-centrosome antiserum; Prof. D. Mazia (Stanford University) for a helpful discussion and Dr T. Shibata (RIKEN Institute) and Prof. J. Hyams (University College London) for help in the preparation of this article.