Tubifex eggs individually inherit a single maternal centrosome during the first mitosis (mitosis I), and thereby form an asymmetrically organized monastral mitotic apparatus for the unequal cleavage. This maternal centrosome does not duplicate itself at the transition to mitosis I but does so at the transition to the second mitosis (Shimizu T, 1996, Roux's Arch Dev Biol 205: 290-299). To examine whether the maternal centrosome duplication is under cytoplasmic control, we produced syncytial eggs by electric fusion of an early meiosis II egg with an early mitosis I egg in an equator-to-equator fashion. As suggested from the diffusion rate of microinjected Texas Red-dextran (10,000 MW), diffusible ooplasmic components in syncytial eggs appear to be intermingled within 20 min following fusion. At ∼120 min after fusion, the meiosis II spindle and the mitosis I nucleus in each syncytial egg complete respective meiosis and mitosis almost simultaneously. Shortly after completion of the meiosis/mitosis, syncytial eggs individually exhibit three centrosomes, two of which are associated with a nucleus derived from the mitosis I spindle, and the remaining one with a nucleus derived from the meiosis II spindle. These results suggest that maternal (meiosis ll) centrosomes fail to duplicate under cytoplasmic conditions that favor the duplication of mitotic centrosomes and that meiosis II centrosomes are intrinsically distinct from mitotic centrosomes. We propose that inhibitory factors that block centrosome duplication are associated with meiotic centrosomes and that their release during mitosis I gives rise to reproduction-competent maternal centrosomes.
INTRODUCTION
The centrosome in animal cells is the region from which most microtubules assemble. In mitotic cells, it defines the spindle poles, and therefore plays a crucial role in determining the cleavage plane and establishing cell polarity (Balczon, 1996). One of the characteristics of the centrosome is its doubling, which takes place only once in each cell cycle. In cultured mammalian cells, centrosome duplication begins at the G1/S transition and is completed during G2 phase of the cell cycle. Apparently, the centrosome cycle in animal cells is tightly coordinated with the mitotic cycle. Recently, it has been demonstrated that centrosomes have the potential to produce multiple copies during a single cell cycle (Fukasawa et al., 1996). This suggests that in normal cells, centrosome duplication is highly regulated. Although details of this regulation remain to be elucidated (Balczon, 1996), it is obvious that this regulation is vital for cells to be viable.
In embryos of the freshwater oligochaete Tubifex, not only is the centrosome duplication during cleavage stages regulated to proceed in a 1 to 2 fashion, but the number of centrosomes that are inherited during the first mitosis (mitosis I) must also be strictly regulated. Unlike many other animals in which two centrosomes participate in the mitotic apparatus (MA) assembly for the first cleavage (Schatten, 1994), Tubifex eggs individually inherit a single centrosome during mitosis I and thereby form an asymmetrically organized monastral MA for the first cleavage, which is unequal (Fig. 1F; Ishii and Shimizu, 1995; Shimizu, 1996a). We have recently demonstrated that Tubifex eggs that are manipulated to inherit two centrosomes during mitosis I form a symmetrically organized amphiastral MA and divide equally (Ishii and Shimizu, 1997), suggesting that the inheritance of a single centrosome during mitosis I is critical for Tubifex eggs to undergo unequal first cleavage. In view of the fact that inequality of this division is indispensable to generate functional organisms (Shimizu, 1982), it is thought that the regulation of centrosome behavior at the transition to mitosis I is essential for Tubifex embryo-genesis.
Fig. 1
Schematic representations of developmental stages of the Tubifex egg, from metaphase of meiosis I through to the 2-cell stage. (A)-(D), side-views of whole eggs; (E)-(H), equatorial sections. Meiotic (A-D) and mitotic (E-H) figures are presented in the upper part; centrosomes are indicated by dots seen at the center of asters. (A) Metaphase of meiosis I. The metaphase spindle orients itself perpendicular to the animal pole surface. Centrosomes are present at both poles. (B) First PBF is accompanied by the first deformation of the egg surface. The centrosome that has been located at the outer pole of the meiosis I spindle is “discarded” together with the first polar body (1PB); the centrosome inherited by the egg proper does not duplicate itself. (C) Metaphase of meiosis ll. The metaphase-ll spindle is formed at the animal pole. The inner spindle pole possesses a centrosome; however, the outer pole is devoid of centrosomes. (D) Second PBF is accompanied by the second deformation of the egg surface. At the end of this division, the centrosome that had been located at the inner pole of the meiosis II spindle is seen near the pronucleus (stippled circles) of the egg proper; it does not duplicate itself but becomes smaller than before. The second polar body (2PB) is devoid of centrosomes. (E) Early stage of the first mitosis (about 30 min after the second PBF). The interphase nucleus (stippled circle) is associated with a single aster; at this stage, the centrosome at the center of the aster is as large as that seen during meiotic divisions. (F) Metaphase of the first mitosis (about 90 min after the second PBF). The metaphase spindle is monastral and located at the egg's center. The right spindle pole is devoid of either an aster or a centrosome. (G) Telophase of the first mitosis (about 150 min after the second PBF). The centrosome duplicates itself, giving rise to a pair of small dots. (H) Two-cell stage. The duplicated centrosomes are inherited by the larger cell CD but not by the smaller cell AB. The CD cell nucleus is associated with a pair of asters; the AB cell nucleus is devoid of asters.
Fig. 2
Fluorescence micrographs showing the distribution of Texas Red-dextran during the fusion of eggs. An egg at the early stage of mitosis I was injected with Texas Red-dextran and fused electrically with an uninjected meiosis II egg. (A) Shortly after fusion. Texas Red-dextran in the mitosis I egg (left) has begun to move into the meiosis II egg (right). Arrows indicate the boundary between the eggs. The contour of the right egg is not depicted here; arrowheads point to its surface at 3, 6 and 12 o'clock. (B) Fifteen min after fusion. The left and right halves of this syncytial egg correspond to the previous mitosis I and meiosis II eggs, respectively. Texas Red-dextran is distributed throughout the egg. Note the meiotic spindle attached to the surface (arrowhead). Bar, 200 μm.
Given that there is no trace of sperm (paternal) centrosomes in Tubifex eggs during meiosis through to mitosis I (Shimizu, 1996a), it is most likely that the centrosome involved in the first cleavage is maternal in origin (Shimizu, 1996b). In fact, a centrosome that has been located at the inner pole of the meiosis II spindle persists into mitosis I without duplicating itself (Fig. 1C-E) and participates in the MA assembly for the first cleavage (Shimizu, 1996a). Apparently, the inheritance of a single centrosome in the Tubifex egg results simply from the absence of duplication of maternal centrosomes at the transition to mitosis I. However, it is unlikely that the absence of centrosome duplication is merely ascribable to the loss of reproductive capability of maternal centrosomes, since centrosomes involved in the first cleavage do duplicate themselves at the transition to the second mitosis (Fig. 1G; Shimizu, 1996a).
The present study was undertaken to gain an insight into the mechanism that regulates the duplication of maternal centrosomes in Tubifex eggs. Reproductive capability of maternal centrosomes may be regulated by ooplasmic environments, or alternatively, it may be determined by intrinsic factors. To differentiate between these possibilities, we examined the behavior of maternal centrosomes in meiosis II eggs that had been fused electrically with mitosis I eggs. The results of this study show that meiosis II maternal centrosomes do not duplicate themselves in an ooplasmic environment that favors the duplication of mitotic centrosomes. This suggests the importance of intrinsic factors in the regulation of the maternal centrosome duplication.
MATERIALS AND METHODS
Eggs of the freshwater oligochaete Tubifex hattai were obtained according to Shimizu (1982). For the experiments, eggs were taken out of their cocoons. Unless otherwise indicated, all experiments were carried out at room temperature (20-21 °C).
Fusion of eggs
The electric fusion of eggs was performed according to the method of Shimizu (1993). Two eggs that had been removed from the vitelline membranes were placed on 1% agar in the culture medium. While the eggs were held in contact with each other so that their animal-vegetal axes ran parallel to each other, an aqueous solution of 40% polyethylene glycol (PEG 6000; Wako Ltd.) was poured over the eggs, and within one second, they adhered firmly in an equator-to-equator fashion. They were then transferred immediately to a fusion medium (41 mM mannitol in distilled water with 5 mM Tris-HCI, pH 7.0). During this stage, PEG often induced adhesion of eggs to the agar bed as well, so that it was necessary to carry out pipetting or the transfer of adhering eggs with great care. After being rinsed twice in the fusion medium, adhering eggs were placed between electrodes (acryl resin-covered cabon rods with an exposed area of 1 × 5 mm2 in the fusion medium) with the plane of adhesion perpendicular to the electric field and exposed to a single rectangular pulse of 200 V/cm for 100 μsec. The distance between the electrodes was 5 mm, and the resistance between them in the fusion medium was 50 kohms. Treated eggs were transferred to the culture medium and allowed to develop.
To examine the extent to which the cytoplasm is intermingled in fused eggs, an egg that had been microinjected with 100 mg/ml lysine fixable, Texas Red-dextran (dissolved in 0.2 M KCI, 10 mM HEPES, pH 7.2; Molecular Probes, Inc., Eugene, USA) was fused electrically with an uninjected egg as described above. At various times after fusion, they were fixed with 3.5% formaldehyde in 10 mM PIPES (pH 6.9) for 60 min in darkness, mounted in a mixture of one part benzyl alcohol:two parts benzyl benzoate (“Murray's clear”), and viewed with a Zeiss Axioskop epifluorescence microscope.
Immunostaining of microtubules
For examination of the distribution and organization of microtubules, eggs were processed for indirect immunofluorescence according to Ishii and Shimizu (1995). Briefly, the specimens were incubated with a 1:1 mixture of mouse monoclonal antibodies to a- and β-tubulin (1:100 in PBS containing 5% BSA and 0.1% sodium azide; Amersham, Buckinghamshire, England) for 48 hr at 4°C, followed by incubation with goat anti-mouse IgG antibody conjugated to FITC (1:50 in PBS containing 5% BSA and 0.1% sodium azide; Tago Inc., Burlington, USA) for a further 48 hr at 4°C. The specimens were mounted in Murray's clear containing 2.5% n-propyl gallate. Images were collected on a Molecular Dynamics Sarastro-2000 confocal laser scanning microscope.
Immunocytochemistry of centrosomes
To visualize centrosomes, immunocytochemical whole-mounts of fixed eggs were prepared according to Shimizu (1996a). A rabbit polyclonal antibody (kindly donated by Dr. H. C. Joshi) raised against a highly conserved peptide of γ-tubulin (Joshi et al., 1992) was used as a primary antibody. A secondary antibody was goat anti-rabbit IgG antibody conjugated to horseradish peroxidase (HRP; Tago Inc., Burlington, USA). Fixed eggs were incubated with the primary (1:4000 diluted) and secondary (1:2000 diluted) antibodies for 48 hr at 4°C. Color development of HRP activity was carried out with 0.25 mg/ml diaminobenzidine and 0.005% H202 for 45-60 min. The stained specimens were mounted in Murray's clear for observation.
Terminology
In this paper, we use the term centrosome to mean a granular structure that is stained with the anti-γ-tubulin antibody and that is associated with the nucleus in interphase and with spindle poles during mitosis or meiosis (see Bornens, 1992).
RESULTS
Summary of the behavior of centrosomes in the Tubifex egg
Mature Tubifex eggs are oviposited at metaphase of meiosis I and are believed to be fertilized during cocoon deposition (Shimizu, 1982). Fertilized eggs extrude polar bodies twice and enter the first mitosis (mitosis I). From oviposition through to meiosis II, eggs exhibit centrosomes labelled with the anti-γ-tubulin antibody exclusively at poles of meiotic spindles; there is no trace of centrosomes elsewhere in these eggs (Shimizu, 1996a). During the first polar body formation (PBF), the centrosome located at the peripheral pole of the meiosis I spindle is discarded together with the polar body (Fig. 1A and B). At the transition to meiosis II, the centrosome that is retained in the egg proper does not alter in size (Fig. 1B). This centrosome participates in the assembly of the meiosis II spindle, and it is located at its inner pole (Fig. 1C). As meiosis II proceeds from anaphase to telophase, centrosomes become smaller but are still recognizable as tiny dots; eggs individually exhibit no more than one tiny dot (Fig. 1D). This tiny centrosome persists into mitosis I without duplicating itself, resumes its staining with the antibody to a level comparable to that of meiotic centrosomes within 30 min after meiosis II, and participates in the assembly of the first cleavage spindle (Fig. 1E and F). As a result of the involvement of a single centrosome in the spindle assembly, the mitotic apparatus (MA) generated is highly asymmetric in organization (Fig. 1F). The centrosome involved in the first cleavage undergoes duplication during the first half of telophase, i.e., at the beginning of the cleavage furrow fomation (Fig. 1G). The duplicated centrosomes are segregated to a larger cell CD resulting from the first cleavage (Fig. 1H).
Behavior of electrically fused eggs
The lack of duplication of maternal centrosomes at the transition to mitosis I may be ascribable to a deficiency in ooplasmic environments. Alternatively, it could arise due to intrinsic factors. To differentiate between these possibilities, we fused meiosis ll eggs electrically with mitosis I eggs (see Materials and Methods). In this study, an egg shortly after the first PBF was fused with an egg at 30 min of mitosis I. This combination of eggs was chosen for fusion partly because the two nuclei in this syncytial egg are scheduled to enter telophase almost simultaneously (at ∼120 min; see Fig. 4). It is known that during the early cleavage stage, centrosomes duplicate themselves during the first half of telophase when DNA replication sets in (Shimizu, 1996a).
When a mitosis I egg that had been injected with Texas Red-dextran (10,000 MW) was fused with an uninjected meiosis II egg, injected dextran was found to spread throughout the latter egg within 20 min (Fig. 2), suggesting that in fused eggs, diffusible ooplasmic components are intermingled in a comparable time. Thus, we considered that in syncytial eggs, meiosis II spindles and mitosis I nuclei would be exposed to a common ooplasmic environment.
In syncytial eggs, meiotic and mitotic figures were not displaced from their “original” positions either during or after fusion. Meiosis ll spindles remained attached to the cortex, while mitosis I nuclei were located in deeper parts, where MA assembly took place (Fig. 3). As summarized diagrammatically in Fig. 4B, meiosis II spindles and mitosis I nuclei in individual syncytial eggs underwent meiosis and mitosis, respectively, in a normal fashion and at a rate comparable to that in unfused control eggs. At about 120 min, PBF (which corresponds to the second PBF in control eggs) and “first” cleavage furrow formation (corresponding to the first cleavage in control eggs) took place nearly simultaneously in syncytial eggs. Although PBF was successful in all cases examined (36/36), “first” cleavage furrow formation was completed in 19/36 cases (see Fig. 5A), while in the remaining cases (17/36), cleavage furrows were regressed (see Fig. 5B).
Fig. 3
Position of meiotic and mitotic figures in a syncytial egg (A) and unfused control eggs (B, C). Confocal laser scanning images of microtubules. (A) Thirty min after electric fusion of a mitosis I egg (left) with a meiosis II egg (right). The meiosis II spindle with an inner aster is attached to the surface (arrow), while the mitotic figure (arrowhead) consisting of a nucleus and an associated aster is located at the position corresponding to the previous center of the mitosis I egg. (B) An unfused mitosis I egg with an internally located nucleus and an associated aster (arrowhead). (C) An unfused meiosis II egg. The meiotic spindle with an inner aster orients itself perpendicular to the animal pole surface (arrow). Bar, 200 μM.
Fig. 4
Diagrammatic summary of meiotic and mitotic events in syncytial eggs during a 210-min period following fusion. (A) Unfused mitosis I egg (control). (B) Syncytial egg produced by electric fusion of an egg at 30 min of mitosis I (left half) with an egg at the beginning of meiosis II (right half). (C) Unfused meiosis II egg (control). A bundle of three parallel lines represents a meiotic or mitotic spindle; a thick line and a pair of lines crossing this bundle indicate metaphase and anaphase chromosomes, respectively. Stippled circles indicate nuclei with or without asters. (B) During and after fusion of eggs, mitotic and meiotic figures remain at their “original” positions (0 min). Mitotic and meiotic nuclear events such as spindle assembly, anaphase chromosome movements (90 min) and nuclear reformation at telophase (120 min) all proceed at a rate comparable to those in unfused control eggs. At 120 min, formation of cleavage furrows (arrowheads) and a polar body (PB) begins almost simultaneously in individual syncytial eggs. Cleavage furrow formation is completed in some eggs but is abortive in others (see text). Here, a case of the former is shown. The resulting two cells are always unequal in size (210 min). Nuclei derived from the mitosis I spindle are segregated separately into these two cells; the nucleus derived from the meiosis II spindle always belongs to the larger cell. Thus, the resulting larger cell is always found to contain two nuclei.
Fig. 5
(A-D) Distribution of early asters (arrowheads) in syncytial eggs (A, B) and unfused control eggs (C, D) at 210 min after fusion. Confocal laser scanning images of microtubules. (A) This egg has divided into two cells. The smaller cell on the left contains two asters associated with a nucleus derived from the mitosis I spindle, while the larger cell on the right exhibits only one aster associated with a nucleus derived from the meiosis II spindle. (B) This egg has failed to divide. A pair of asters is seen on the left, and the nucleus derived from the mitosis I spindle is located between these. The single aster seen on the right is associated with the nucleus derived from the meiosis II spindle. (C) An unfused control egg shortly after the first cleavage. A pair of asters is present in the larger CD-cell but not in the smaller AB-cell. (D) An unfused control egg shortly after the meiosis. This egg exhibits only one aster. (E) MA assembly in a syncytial egg that has failed to divide. About 270 min after fusion. This egg exhibits an amphiastral MA on the left and a monastral MA on the right. The arrow indicates the anastral spindle pole. The amphiastral MA is tripolar in that an “amphiastral” spindle (which is depicted here) and a “monastral” spindle (which is oriented vertically to this panel) share a common astral pole; the arrowhead indicates the transverse section of this monastral spindle. Bar, 200 μm.
Mitotic spindle assembly in syncytial eggs
Figures 5A and 5B illustrate two representative syncytial eggs at 210 min after fusion, when unfused control eggs were at the early stage of the second and the first mitoses, respectively (Fig. 5C and D). Regardless of whether or not “first” cleavage furrows were regressed, the total number of tubulin-containing foci (or early asters) found in individual syncytial eggs was three in 19 out of 21 cases. A nucleus derived from the mitosis I spindle was always found to be associated with a pair of asters, and a nucleus derived from the meiosis Il spindle with a single aster (Fig. 5A and B). In the remaining two cases, the number of early asters could not be determined precisely, except that former nuclei were associated with a pair of asters.
When syncytial eggs (n = 15) were allowed to develop for a further 60 min (i.e., ∼270 min after fusion), most of these eggs (13/15 cases) individually exhibited two mitotic figures, viz., an amphiastral MA and a monastral MA (Fig. 5E), suggesting that three astral spindle poles are generated in each of these eggs. In the remaining eggs (2/15 cases), however, four astral spindle poles were recognized. As Fig. 6 shows, these spindle poles were distributed in three mitotic figures, viz., two separate monastral MAs and an amphiastral MA.
Fig. 6
(A) A syncytial egg exhibiting four asters (arrowheads). About 270 min after fusion. Two solitary asters seen on the right each belong to a monastral MA, the spindle of which is not depicted in this focal plane. (B and C) Two focal planes of the right half of the egg shown in (A) to illustrate transverse sections (arrows) of the spindles that are oriented vertically to these panels. The spindles seen in (B) and (C) are associated with the asters seen in the upper and lower right of (A), respectively. Bar, 200 μm.
Behavior of meiotic and mitotic centrosomes in a common cytoplasm
The aforementioned results suggest that in most of the syncytial eggs examined, microtubule-organizing centers (MTOCs) derived from meiosis II spindles do not duplicate in a cytoplasmic environment where those of mitosis I spindles do. To test this possibility, we examined the behavior of centrosomes in whole-mount preparations stained with the anti-γ-tubulin antibody.
We found that the nucleus derived from mitosis I spindles was associated with two centrosomes and the nucleus derived from meiosis II spindles with one centrosome (18/20 cases; Fig. 7A and B). We also found two cases where not only the former nucleus but also the latter nucleus were associated with two centrosomes (Fig. 7C-F). It should be mentioned, however, that one of the two centrosomes associated with the latter nucleus was smaller than the other in both the cases examined (compare Fig. 7D with Fig. 7F).
Fig. 7
Distribution of centrosomes in two representative syncytial eggs at 210 min after fusion. Immunoperoxidase staining with the anti-γ-tubulin antibody. Arrowheads indicate centrosomes in focus. (A and B) Two focal planes of an egg that exhibits three centrosomes in total. This egg has divided into two cells of unequal size. The smaller cell on the left contains two centrosomes while the cell on the right contains only one. Judging from the accumulation of yolk-free pole plasm (p), the cell on the left is thought to correspond to the larger CD-cell of the 2-cell embryo (see Fig. 1H). The asterisk in (A) indicates a nucleus derived from the mitosis I spindle (i. e., the sister nucleus of that seen in the left cell). (C-F) Four focal planes of an egg that exhibits four centrosomes in total. (D), (E) and (F) are at higher magnification. Both the smaller cell on the left and the larger cell on the right contain two centrosomes. Judging from the accumulation of pole plasm (p), the nucleus seen in the left cell is thought to have been derived from the mitosis I spindle; asterisks indicate the position of its sister nucleus (which is not depicted in these focal planes). Bar in (C): 100 μm for (A), (B), (D), (E) and (F); 200 μm for (C).
These results suggest that centrosomes in meiosis II spindles are unable to duplicate themselves even under the condition where mitosis I centrosomes are able to do so.
DISCUSSION
The duplication of centrosomes in early Tubifex embryos, which occurs during the first half of telophase of each mitosis, is coupled to the mitotic cycle and is affected by the cytoplasmic environment (Shimizu, 1996a). In contrast, a maternal centrosome located at the inner pole of the meiosis II spindle does not duplicate itself at the transition to mitosis I, even though nuclear events such as DNA replication and chromatin decondensation that are comparable to those at mitotic telophase occur at the end of meiosis II (Shimizu, 1995, 1996a). The present cell-fusion experiments have shown that maternal centrosomes fail to duplicate themselves even when they are exposed, for more than 1 hr, to a cytoplasmic environment that favors the duplication of mitotic centrosomes. This suggests that meiosis II centrosomes are intrinsically different from those during mitoses. It seems unlikely that duplication of maternal centrosomes at meiosis II is determined by the cytoplasmic environment. Rather, reproductive incapability of these centrosomes may be brought about through their intrinsic factors. It is conceivable that inhibitory factors that block centrosome duplication are associated with meiotic centrosomes; maternal centrosomes would become competent to reproduce as a result of the release of such inhibitory factors during mitosis I. In this connection, it is interesting to note that γ-tubulin staining and the size of maternal centrosomes, both of which do not change during maturation divisions, alter significantly at the beginning of mitosis I (Shimizu, 1996a). It seems possible that such structural alterations include the release of putative inhibitory factors and thereby give rise to reproduction-competent maternal centrosomes. It remains to be determined whether these changes in centrosomal structures are triggered by cytoplasmic environments or whether they occur autonomously.
Thus, we suggest that maternal centrosomes in Tubifex eggs do not lose their reproductive ability during meiosis. Rather, they may have potential to duplicate themselves even at the transition to mitosis I. A manifestation of this potential may be the two cases (out of 20 cases) in which meiosis II centrosomes appear to have duplicated themselves (Fig. 7C-F). At present it is not known how these duplicated centrosomes are produced. However, it is not difficult to imagine that accidental dissociation of the putative inhibitory factors would result in the duplication of meiosis II centrosomes in syncytial eggs.
The absence of duplication of maternal centrosomes at the transition to mitosis I is not unique to Tubifex but is found in a variety of organisms such as shrimp, starfish and Urechis (Sluder et al., 1989; Kato et al., 1990; Hertzler and Clark, 1993; Stephano and Gould, 1995). It appears that this behavior of maternal centrosomes in these organisms is due to their intrinsic factors as in Tubifex eggs (Sluder et al., 1989, 1993). However, it is unclear whether the behavior of maternal centrosomes in these animals is controlled by a mechanism similar to that in Tubifex. Unlike those in Tubifex eggs, single maternal (meiosis II) centrosomes that persist into mitosis I in starfish eggs are unable to duplicate themselves at successive mitoses (Sluder et al., 1989), unless the centrosome number is manipulated to increase (Sluder et al., 1993; Washitani-Nemoto et al., 1994), suggesting the complete loss of reproductive capability of maternal centrosomes. In contrast, reproductive capability of maternal centrosomes in Tubifex eggs apparently “resumes” during mitosis I. Thus, it appears that intrinsic factors that give rise to the absence of centrosome duplication are distinct between Tubifex and starfish.
Acknowledgments
The authors wish to thank Dr. Hirsh C. Joshi, Emory University, for the gift of the anti-γ-tubulin antibody. This study was supported in part by Narishige Zoological Science Award and Grant-in-Aid for Scientific Research (04640649) from the Ministry of Education, Science and Culture of Japan to T. S.
