Proc. Nati. Acad. Sci. USA Vol. 88, pp. 9929-9933, November 1991 Cell Biology Centrosomes competent for parthenogenesis in Xenopus eggs support procentriole budding in cell-free extracts (cell cycle/protein synthesis) FRE9DERIC TOURNIER*, MAREK CYRKLAFFt, ERIC KARSENTIt, AND MICHEL BORNENS* *Centre de Gdn6tique Mol6culaire, Centre National de la Recherche Scientifique, 2 Avenue de la terrasse, 91198 Gif/Yvette, France; and tEuropean Molecular Biology Laboratory, Postfach 102209, Meyerhofstrasse 1, D-6900 Heidelberg, Federal Republic of Germany Communicated by Andre Lwoff, August 1, 1991 (received for review April 22, 1991) cell cycle from which they have been isolated: centrosomes isolated from G1 or G2 human lymphoid cells (KE37 cell line) or from quiescent cells (peripheral human lymphocytes in Go) were shown to possess a similar parthenogenetic activity (8). We have isolated (10) centrosomes from calf thymocytes (CTs) and shown that the two centrioles were linearly associated by their proximal ends through a mass of dense material. Furthermore, they were unable to induce egg cleavage, although centrosomes isolated from other bovine cells and from thymocytes of other species could induce egg cleavage (6). These results suggest that the centrosome cycle can be blocked when the centrioles are prevented from separating into a nonlinear configuration, a step that might be critical for the initiation of procentriole budding. Therefore, we have prepared cell-free extracts from Xenopus eggs in the mitotic or interphasic stage, in which we could directly compare the duplicative capacity of centrosomes that are competent and incompetent (such as CT centrosomes) to induce parthenogenesis. We show that competent centrosomes from human lymphoblasts (KE37 cell line), synchronized in G1 phase, initiate procentriole budding in interphasic extracts from Xenopus eggs in the absence of protein synthesis, whereas incompetent (CT) centrosomes do not. Since CT centrosomes do not support parthenogenesis, the present results strongly suggest that duplication of the foreign centrosome is required for centrosome-induced parthenogenesis. Heterologous centrosomes from diversed speABSTRACT cies including humans promote egg cleavage when i iected into metaphase-arrested Xenopus eggs. We have recently isolated centrosomes from calf thymocytes and shown that they were unable to induce egg cleavage, an inability that was apparently correlated with the peculiar structure of these centrosomes rather than with a lack of microtubule-nucleating activity: the two centrioles were associated in a colinear orientation by their proximal ends. To promote cleavage, a heterologous centrosome probably is required to duplicate, although this has not yet been demonstrated. Therefore, we designed an in vitro assay that would enable us to directly observe the duplication process. We show that competent centrosomes from KE37 cells synchronized in G1 phase initiate procentriole budding in interphasic extracts from Xenopus eggs in the absence of protein synthesis, whereas calf thymocyte centrosomes do not. Since calf thymocyte centrosomes do not support parthenogenesis, the present results suggest that duplication of the foreign centrosome is required for centrosome-induced parthenogenesis. Furthermore, procentriole budding takes place in the absence of protein synthesis in egg extracts arrested in S phase. This in vitro assay should contribute to the identification of molecular mechanisms involved in the initiation of centrosome duplication. Downloaded by guest on June 3, 2020 In somatic cells, centrosome duplication encompasses the whole cell cycle and progression in the centrosome cycle appears to be a marker of progression in the cell cycle. The importance of centrosomes in establishing the spatial organization of the mitotic apparatus mandates that the cell tightly controls the reproduction of its centrosome, which must occur only once in each cell cycle, prior to mitosis. The morphological events of the centrosome duplication cycle in somatic cells have been described and involve budding of daughter centrioles off the wall of the parent structures in late G1 phase, elongation of the daughter centrioles during S and G2 phases, redistribution of pericentriolar material, and separation of duplicated centrosomes in prophase (1-3). However, the molecular mechanisms that govern centrosome duplication are not known. As amphibian eggs apparently lack a functional centrosome, which can be replaced by the injection of foreign centrosomes (parthenogenetic development), they represent a favorable system to study the duplication of centrosomes. In this system, no species specificity is apparently required for the centrosomes: heterologous centrosomes isolated from sea urchin (4), rodent (5, 6), or human (7-9) cells can induce cleavage of Xenopus eggs. In this parthenogenetic test, the heterologous centrosomes are believed to duplicate, but direct experimental evidence is lacking. The parthenogenetic activity of the centrosome is independent of the stage of the MATERIALS AND METHODS Animals. Xenopus laevis females were obtained from the Service d'Elevage d'Amphibiens du Centre National de la Recherche Scientifique (France). Eggs were obtained from females injected 3-8 days before use with 100 international units of pregnant-mare-serum gonadotropin (Intervet, Angers, France) and 1 day before with 1000 international units of human chorionic gonadotropin (Sigma). The eggs were collected in 100 mM NaCl to prevent activation. Preparation of Extracts. Egg extracts were prepared according to Felix et al. (11). Egg jelly was removed in 2% (wt/vol) cysteine hydrochloride (pH 7.8), and eggs were washed four times in 0.25 x MMR (lx MMR = 0.1 M NaCl/2 mM KCl/1 mM MgSO4/2 mM CaCl2/5 mM Hepes/0.1 mM EDTA, pH 7.4) containing cycloheximide (150 pug/ml; Sigma), and then activated by an electric shock (11) in the same buffer for 90 min at 20'C. Activated eggs were transferred to Beckman SW 50.1 rotor tubes filled with ice-cold acetate buffer [AB = 100 mM potassium acetate/2.5 mM magnesium acetate/cytochalasin D (10 pug/ml; Sigma)/1 mM dithiothreitol/250 mM sucrose/leupeptin (10 pug/ml)/pepstatin (10 kkg/ ml)/aprotinin (10 ,g/ml), pH 7.2]. Excess acetate buffer was removed from the tubes containing packed eggs prior to centrifugation. The eggs were crushed by centrifugation at The publication costs of this article were defrayed in part by page charge payment. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. §1734 solely to indicate this fact. Abbreviation: CT, calf thymocyte. 9929 9930 Cell Biology: Toumier et al. Downloaded by guest on June 3, 2020 10,000 x g (with maximum acceleration rate) for 10 min at 40C in a L5-65 Beckman centrifuge. The cytoplasmic material between the upper lipid layer and the yolk pellet was collected (10,000 x g supernatant) and an ATP regenerating system (10 mM creatine phosphate, creatine phosphokinase (80 jig/ml), and 1 mM ATP (Boehringer Mannheim)] and protease inhibitors as above were added. A sample was kept in ice during further steps. The 100,000 x g extracts were obtained by further centrifugation of the 10,000 x g supernatant at 100,000 X g for 60 min at 40C in the SW 50.1 rotor with adaptors for 0.6-ml tubes. The yellow cytoplasmic layer (100,000 x g supernatant) was collected and kept at 4TC, because freezing the extracts gave inconsistent results in the centrosome duplication assay. Cell Synchronization and Isolation of Centrosomes. KE37 is a human T-lymphoblastic cell line. KE37 cells, cultured in RPMI 1640 medium (Eurobio, Les Ulis, France) containing 7% (vol/vol) calf serum, were first synchronized at the G1/S-phase border with 2.5 mM thymidine (Sigma) for 20 hr (12) and then separated in a cell-cycle-dependent manner using centrifugal elutriation according to Tournier et al. (8). Centrifugal elutriation led to an enriched fraction containing >80% G1 phase cells. Centrosomes were prepared from KE37 cells in G1 phase according to Tournier et al. (8) and from CTs according to Komesli et al. (10). Duplication Assay. Typically, 10 ,ul of KE37 G1 centrosomes (1 x 104 centrosomes per ul) was mixed with 100,ul of extract (either 10,000 x g or 100,000 X g supernatant) at 4°C. The mixture was equally distributed into four tubes and incubated at Proc. Natl. Acad. Sci. USA 88 (1991) room temperature for 0-3 hr. After incubation, nocodazole was added to 30 ,&g/ml and the tubes were kept at 4°C for 30 min. Then 0.8 ml of lysis buffer [1 mM Tris'HCl, pH 7.5/0.1% 2-mercaptoethanol/0.5% Nonidet P-40/0.5 mM MgClJ1 mM phenylmethylsulfonyl fluoride/aprotinin (100 pg/ml)/leupeptin (1 ,ug/ml)/pepstatin (1 ,ug/ml)] at 4°C was added to each tube and the material in each tube was layered onto 25% (vol/vol) glycerol in RG2 buffer (80 mM Pipes.KOH/1 mM MgCl2/1 mM EGTA, pH 6.8) in a 15-ml modified Corex tube with a special adaptor containing a 12-mm round coverslip. The material was centrifuged at 12,000 rpm (20,000 x g) in a JS-13.1 rotor (Beckman) at 4°C for 15 min to apply the centrosomes to the coverslips. The coverslips were removed, fixed in methanol at -20°C for 5 min, and processed for immunofluorescence staining. Immunofluorescence Study. Centrosome duplication was monitored by measuring the numbers of centrioles and centrosomes by double immunofluorescence (7). A monoclonal antibody against a-tubulin (Amersham) stained centrioles and a human autoimmune serum stained the pericentriolar material of human centrosomes (J. E. Dominguez, E.K., and J. Avila, unpublished data). Negative Staining. For negative staining of electron microscopy preparations, 15 ,ul of G0 centrosomes (1 x 104 centrosomes per ,l) was mixed at 4°C with 300 ,l of extract. Further steps were identical to the immunofluorescence protocol until the lysis step. For each point (at 0, 2, and 3 hr of incubation), 100 ,l of extract was mixed with 200 Al of lysis buffer and incubated 2 min at 4°C. The material was loaded FIG. 1. Double immunofluorescence staining of G1 centrosomes (KE37) in Xenopus extracts arrested in S phase at 0 time (a and a') and after 3 hr of incubation (b and b'). Centrosomes stained with the human autoimmune serum (a) appear to be doublets of tubulin-containing dots corresponding to centrioles (a'). In this experiment, the two centrioles in a centrosome do not split (see Table 1) and after 3 hr of incubation in the extract, the centrosomes (b) contain three or four tubulin-containing dots as judged by the anti-tubulin staining (b'). In the lower parts of each panel, other examples at 0 time (a and a') and after 3 hr of incubation (b and b') are shown. The variable intensities of the centrioles probably depend on their orientation on the coverslip. (x 16,900.) Cell ~ ~1 Biology: Toumier et al. Proc. Natl. Acad. Sci. USA 88 (1991) onto 200 Al of 25% glycerol in RG2 buffer in a 0.6-ml tube containing a Teflon adaptator supporting a carbon grid. The material was centrifuged at 14,000 rpm (25,000 x g) in a SW 50.1 rotor (Beckman) for 20 min (40C) to apply centrosomes to the grid. Grids were then removed, rinsed in RG2 buffer, stained with 2% (wt/vol) phosphotungstate (pH 7 with KOH), and dried. RESULTS Duplication of G, Centrosomes in Cell-Free Extracts: Immunofluorescence Study. Competent centrosomes were isolated from KE37 cells synchronized in G1 phase. We have shown (8) that these centrosomes were capable of inducing parthenogenesis when injected into unfertilized Xenopus eggs. The use of cell-free extracts and immunofluorescence to study microtubule polymerization nucleated by isolated -J 2 °0 FE 0 0l _2 ~~~~~~0 'U zC~ .)_1 ~ ~ ~ ~ ~ ~L Q 010 LM. M0 z 1 00 ( TIME (min) 0 z 2cn 60 0 A. o 40 0U 200 D >20 3 IL 0 0 .10 TIME (min) C,, 'U -j c 0 I- z 2- Lou 0 o z 0 1 Downloaded by guest on June 3, 2020 TIME 00 200 (min) FIG. 2. Duplication of KE37 centrosomes in Xenopus extracts arrested in S phase. (a) Numbers of centrosomes (u) and centrioles (m) were recorded at 0 time and after 1, 2, and 3 hr of incubation. After 3 hr of incubation, the number of centrosomes is constant (as the two centrioles in a given centrosome did not split in the extract) whereas the number of centrioles doubles. Approximately 1 x 105 centrosomes were added at 0 time in 100 1.l of extract. Ten fields (>100 centrosomes) were recorded for each point after incubation. The number of centrioles at 0 time was standardized (1 on the Y axis). (b) The number of duplicated forms (three or four centrioles per centrosome) increases from 1% to 48%. (c) The number of KE37 centrioles remains constant with the 10,000 x g supernatant (e) incubated for 0-3 hr. In the corresponding 100,000 x g supernatant (o), the number of centrioles increases after 2 hr of incubation. This suggests that the high-speed (100,000 x g) pellet contains an inhibitory activity for centrosome duplication. 9931 centrosomes has been described (13-15). In the present study, visualization of the centrosomes by immunofluorescent staining was done after nocodazole-induced disassembly of the microtubules, which had been nucleated in the extract by the added centrosomes. Double staining was used to identify both the pericentriolar domain and the centrioles. This allowed us to identify the centrosomes unambiguously by distinguishing them from tubulin aggregates. More reproducible results were obtained from cytoplasmic extracts of Xenopus eggs prepared from activated oocytes in the presence of cycloheximide (interphasic extracts). G1 centrosomes from KE37 cells were incubated in 100,000 x g Xenopus egg extracts for 0-3 hr. At 0 time, centrosomes were observed as pairs of centrioles (Fig. 1 a and a'), and after 3 hr of incubation, centrosomes contained three or four tubulin dots (Fig. 1 b and b'). We interpreted these dots as duplicative forms of centrosomes in which the centrioles did not separate. The numbers of centrosomes and centrioles were recorded at various incubation times by double immunofluorescence. After 3 hr of incubation, the number of centrosomes remained constant but the number of centrioles increased from 1 (reference value at incubation time) to 1.95 (Fig. 2a) and the number of centrosomes containing 3 or 4 identifiable centrioles increased from 1% to 48% (Fig. 2b). Typically, the number of KE37 centrioles doubled after 2-3 hr of incubation. We noted, however, that the efficiency of centrosome duplication depended largely on the batch of extract used. In other experiments, the two centrioles in a given centrosome often separated from each other over large distances (Table 1) during the first hour of incubation as judged by the increase in the proportion of single centrioles counted at this time. Duplicated forms observed after 3 hr of incubation were identified as a pair of tubulin-containing dots associated with centrosome labeling. When the same kinetic experiments were done using low-speed extracts (10,000 x g), no increase in centriole number was detected after 2 or 3 hr of incubation (Fig. 2c) or up to 6 hr of incubation (data not shown), suggesting that the 10,000 x g supernatant contained an activity that inhibited centrosome duplication. Duplicative Forms of Centrosomes Observed by Negative Staining. To validate our scoring method by an independent approach (electron microscopy), we examined negatively stained centrosomes that had been incubated in the extract. Table 1. Centriole splitting in interphasic extracts during the first hour of incubation Centriole form, no. Double Single n Time, hr Cells Exp. 488 (66) I 5 0 388 (44) KE37 189 (16) 1 952 (84) 148 (75) II 49 (25) 2 0 180 (73) 67 (27) 1 183 (81) 42 (19) 2 0 CT 45 (23) 152 (77) 1 n represents the number of experiments. Centrosomes were recorded as pairs of tubulin-containing dots at 0 time and after 1 hr of incubation. For KE37 centrioles (experiments I), a splitting of the two centrioles in a given centrosome was observed (44% of single centrioles at 0 time and 84% after 1 hr of incubation). In five experiments (I), duplicated forms observed after 3 hr of incubation in the extract were identified as pairs of tubulin-containing dots associated with centrosome labeling. In two experiments (II), a splitting of the two centrioles was not observed (25% of single centrioles at 0 time and 27% after 1 hr of incubation). The duplicated forms observed after 3 hr in the latter case (II) were recorded as three or four tubulin-containing dots. For CT centrosomes, in all cases, we did not observe any significant splitting of the centrioles after 1 hr of incubation. Numbers in parentheses are percent of total centrioles. 9932 Cell Biology: Tournier et al. Proc. Natl. Acad Sci. USA 88 (1991) a CD t, .r, q..P:.. 2 LU ....I . I.z LU 't U LIC.) 0 LU m z z t 0 1 00 200 TIME (min) FIG. 4. Centrosomes isolated from CTs do not duplicate in 100,000 x g supernatant, whereas KE37 centrosomes initiate procentriole budding, as shown by the increase of the number of centrioles. o, CT centrioles; *, KE37 centrioles. c ir .3. :, .8 '1 . FIG. 3. Duplicating centrosomes observed by negative staining. Centrioles were in pairs (a and c) or were single (b and d), without (a and b) or with procentriole buds (arrows) (c and d). (Bar = 0.1 AM.) Centrioles were scored either as single centrioles or as pairs and with or without an associated orthogonal procentriole (Fig. 3). A representative experiment is shown in Table 2. At 0 time, 46% of the centrioles were pairs and 52% were single. After 2 hr of incubation, 12% of centrioles were pairs and 82% of centrioles were single. These results were in agreement with immunofluorescence observations and indicated extensive splitting of the centrioles during incubation in the extracts. Duplicated forms were only 4% of the total centrioles at 0 time (Table 2), which is consistent with the fact that centrosomes were prepared from G1 cells. After 2 hr and 3 hr Table 2. Centriole splitting and duplicated forms by electron microscopy after negative staining Centriole form, no. Duplicated Double Triple forms, no. Single 1 (2) 97 4 (4) 0 30(46) 34(52) 2 21 14 (82) 2 (12) 1 (6) 13 (62) 5 (19) 2 (7) 3 36 20 (74) 21 (58) n represents the number of centrioles. Note that, at 0 time, 52% of centrioles were observed as single centrioles and 46% were observed as centriole pairs. After 2 hr of incubation, single centrioles represented 82% of the total centrioles observed. Centrioles were sometimes observed (2-7%) as triplets. Duplicated forms correspond to centrioles observed with a centriole bud. For duplicated forms, 4% were observed at 0 time, 62% were observed after 2 hr of incubation, Downloaded by guest on June 3, 2020 Time, hr n and 58% were observed after 3 hr of incubation. Numbers in parentheses are percent of total centrioles. of incubation, an orthogonal bud (procentriole) was detected on 60% of the scored centrioles. CT Centrosomes Do Not Support Procentriole Budding. CT centrosomes, which display linearly associated centrioles (10), do not support cleavage when injected into unfertilized Xenopus eggs (6). This suggested that these centrosomes were incompetent to initiate bud formation in an adequate cytoplasmic environment. As shown in Fig. 4, we did not observe an increase in centriole number in an extract incubated with CT centrosomes. Since, in the same extract, KE37 centrosomes do form buds, this result shows that CT centrosomes are incompetent for duplication and interestingly, CT centrioles do not separate (Table 1). DISCUSSION We have compared the capacity of centrosomes to initiate a centrosome duplication cycle in cell-free extracts by using an immunofluorescence-based assay. Centrosomes competent to induce parthenogenesis (G1 centrosomes from KE37 cells) initiate procentriole budding whereas incompetent centrosomes (CT centrosomes) do not. The electron microscopy study supported the conclusion that the increase in tubulincontaining dots observed by immunofluorescence during incubation of the centrosomes in extracts corresponds to an increase in centriole or procentriole number. But it also indicated that only procentrioles were formed. Indeed, we never observed pairs of centrioles tightly associated in an orthogonal configuration in which the two centrioles had the same length. Further incubation (up to 6 hr) did not increase the number of centrioles (data not shown). In particular, we never obtained evidence for successive rounds of centriolar duplication. This suggests that the duplication process is not completed under our in vitro conditions. This may have several causes. First, the high-speed supernatant (100,000 x g) may lack components necessary for bud elongation. Such components may or may not be in the high-speed pellet, because the pellet apparently also contains an activity from the 10,000 x g supernatant that was shown to inhibit the initiation of centriole duplication (Fig. 2c). The most likely explanation is that the components necessary for assembling full-length centrioles are limiting in comparison to the number of centrosomes incubated in the extract. We added between 500 and 1500 centrosomes per ,ul of extract, which is the minimum number required to make the immunofluorescence study feasible. This figure is close to the number of centrosomes produced per embryo at the midblastula transition (16) (one embryo has a volume of about 1 "I). From such a figure, we may conclude that a single centrosome duplication cycle at most could take place under our Cell Biology: Toumier et al. Downloaded by guest on June 3, 2020 conditions. If this were the case, adding centrosomes at a lower concentration should allow multiple rounds of duplication in the extract, but we have not succeeded in designing an experiment to test this hypothesis. Another implication of these results is that the first steps of centrosome duplication are independent of the cytoplasmic oscillator that drives the cell cycle, since procentriole budding takes place in cycloheximide-treated interphasic extracts. However, the incompleteness of the duplication cycle could suggest that a signal from the "mitotic clock" is necessary for the late steps of the process. This is unlikely since multiple cycles of centrosome duplication occurred in cycloheximide-treated Xenopus blastomeres (16) and a similar result was also reported for sea urchin eggs (17). Centrosome duplication can be experimentally uncoupled from the synthesis of mitotic cyclins (refs. 16 and 17 and this report). Our experimental system should, allow investigation ofthe coupling between the centrosome duplication cycle and the cytoplasmic oscillator. The asynchrony of centrosome reproduction in the absence of protein synthesis (16, 17) suggests indeed that phasing of the cytoplasmic oscillator and the centrosome cycle occurs in normal development. p34cdc2, a suci homologue (18), and cyclin B (E. Bailly, J. Pines, T. Hunter, and M.B., unpublished data), the major components of the cytoplasmic oscillator, are associated with the centrosome in G2 and M phases. It will be of interest to study centrosome duplication in cycling extracts and the role of protein phosphorylation controlled by various cyclins in synchronization of centrosome duplication with the mitotic cycle. Preliminary experiments in mitotic extracts suggest that procentriole budding cannot occur during mitosis (data not shown). The inability of CT centrosomes to initiate a centriole bud in vitro strongly suggests that, in parthenogenetic Xenopus eggs, cleavage requires the duplication of the injected centrosome. This also indicates that the structural configuration of the centrosome is essential to initiate budding in a permissive cytoplasmic environment. Specific egg proteins are able to assemble on the two centrioles of the heterologous centrosome to form the two poles of the first mitotic spindle, in which each pole should contain a chimeric centrosome. The incompetence of CT centrosomes is not simply due to the species or the tissue origin, because the linear arrangement of the CT centrosomes apparently represents a locked configuration. Proc. Natl. Acad. Sci. USA 88 (1991) 9933 In conclusion, the present data demonstrate the potential of our in vitro assay to describe in molecular terms the orthogonal budding of procentrioles from the proximal end of parental centrioles. We thank Dr. Spencer Brown, Dr. Pedro Santamaria, and Eric Bailly for their helpful comments and English corrections on the manuscript. We also thank J. E. Dominguez for the gift of the human autoimmune serum and Daniel Meur for the art work. This work has been supported by Centre National de la Recherche Scientifique and by Grant 87.C.0555 from Ministere de la Recherche et de la Technologie to M.B. and by the Association pour la Recherche sur le Cancer. F.T. is a recipient of a fellowship from the Ministere de la Recherche et de la Technologie. 1. Kuriyama, R. & Borisy, G. G. (1981) J. Cell Biol. 91, 822-826. 2. Rieder, C. R. & Borisy, G. G. (1982) Biol. Cell. 44, 117-132. 3. Kochanski, R. S. & Borisy, G. G. (1990) J. Cell Biol. 110, 1599-1605. 4. Maller, J., Poccia, D., Nishioka, D., Kido, P., Gerhart, J. & Hartman, H. (1976) Exp. Cell Res. 99, 285-294. 5. Karsenti, E., Newport, J., Hubble, R. & Kirschner, M. W. (1984) J. Cell Biol. 98, 1730-1745. 6. Tournier, F., Komesli, S., Paintrand, M., Job, D. & Bornens, M. (1991) J. Cell Biol. 113, 1361-1369. 7. Bornens, M., Paintrand, M., Berges, J., Marty, M. & Karsenti, E. (1987) Cell Motil. Cytoskeleton 8, 238-249. 8. Tournier, F., Karsenti, E. & Bornens, M. (1989) Dev. Biol. 136, 321-329. 9. Klotz, C., Dabauvalle, M. C., Paintrand, M., Weber, H., Bornens, M. & Karsenti, E. (1990) J. Cell Biol. 110, 405-415. 10. Komesli, S., Tournier, F., Paintrand, M., Margolis, R., Job, D. & Bornens, M. (1989) J. Cell Biol. 109, 2869-2878. 11. Felix, M. A., Pines, J., Hunt, T. & Karsenti, E. (1989) EMBO J. 8, 3059-3069. 12. Bostock, C. J., Prescott, D. M. & Kirkpatrick, J. B. (1971) Exp. Cell Res. 68, 163-168. 13. Verde, F., Labbe, J. C., Dorde, M. & Karsenti, E. (1990) Nature (London) 343, 233-238. 14. Belmont, L. D., Hyman, A. A., Sawin, K. E. & Mitchison, T. J. (1990) Cell 62, 579-589. 15. Sawin, K. E. & Mitchison, T. J. (1991) J. Cell Biol. 112, 925-940. 16. Gard, D. L., Hafezi, S., Zhang, T. & Doxsey, S. J. (1990) J. Cell Biol. 110, 2033-2042. 17. Sluder, G., Miller, F. J., Cole, R. & Rieder, C. L. (1990) J. Cell Biol. 110, 2025-2032. 18. Bailly, E., Doree, M., Nurse, P. & Bornens, M. (1989) EMBO J. 8, 3985-3995.