Microtubule arrays of the zebrafish yolk cell: organization and function during epiboly

During gastrulation of a vertebrate embryo, various cell movements including involution and convergence lead to formation of a gastrula with the three classically defined germ layers and a blueprint of the body plan. In amphibians and fishes, gastrulation is preceded and accompanied by epiboly -a process of vegetal expansion of the blastoderm (Keller, 1980; Trinkaus, 1951). The mechanisms that determine directions of cell migrations and forces that drive epiboly and gastrulation movements are still poorly understood. Several features make teleost embryos attractive for studies of these processes (Trinkaus et al., 1992; Kimmel, 1989). First, due to optical clarity of certain teleost embryos, cell movements and behavior can be observed in intact embryos (Trinkaus et al., 1991; Warga and Kimmel, 1990). Second, the relatively large size of Fundulus heteroclitus or zebrafish (Danio rerio) embryos allows for easy experimental manipulations (Ho, 1992). Finally, zebrafish is amenable to genetic analysis (Streisinger et al., 1981).

At the onset of epiboly in teleosts with meroblastic early cleavages, a syncytial yolk cell is capped by the blastoderm (Fig. 1A). A superficial, single-cell-thick enveloping layer (EVL) is attached by its vegetal rim to the yolk cell, and covers a mass of deep cells. Three compartments of continuous cortical cytoplasm can be distinguished in the yolk cell (Trinkaus, 1992, 1993; Long, 1984). A thin, anuclear yolk cytoplasmic layer (YCL) surrounds the bulk of the yolk mass with the vegetal pole. The external yolk syncytial layer (YSL), located between the YCL and the blastoderm rim, is a relatively thick belt of cytoplasm populated by the yolk syncytial nuclei (YSN). Beneath the blastoderm is the internal YSL comprising a thinner cytoplasm also populated by the YSN. Epiboly in Fundulus begins with a contraction of the external YSL that becomes covered by the expanding blastoderm. Subsequently, the YSL and blastoderm expand vegetally, while the YCL progressively disappears (Trinkaus, 1993; Warga and Kimmel, 1990). Studies in Fundulus indicated that the epiboly of the YSL is autonomous, i.e. it can occur in the absence of the blastoderm (Betchaku and Trinkaus, 1978; Trinkaus, 1951). Moreover, vegetal expansion of the EVL during epiboly may be driven by the epibolic movements of the YSL, to which the EVL is tightly linked at its margin (Betchaku and Trinkaus, 1978). The molecular basis for the expansion of the YSL is not understood. It has been suggested that microfilaments may play a role in this process (Trinkaus, 1984). Recently, it has been proposed that epiboly is also driven by a pulling force dependent on microtubules present in the yolk cell (Strähle and Jesuthasan, 1993). However, developmental changes of the yolk cell microtubule organization and their involvement in distinct aspects of epiboly were not studied.

Here, we demonstrate that two distinct microtubule arrays exist in the cortical cytoplasm of the zebrafish yolk cell. The YCL contains an array of microtubules aligned in the direction of epiboly, extending toward the vegetal pole. Once formed, the YSL exhibits its own network of intercrossing interphase or mitotic microtubules. Our studies demonstrate that the changes in the configuration of the yolk cell microtubules are strictly correlated with epibolic movements. Analysis of epiboly in embryos treated with nocodazole and taxol indicates that microtubules are necessary and actively involved in epibolic movements of the syncytial nuclei, and only partially required for the remaining aspects of epiboly.

Reagents were obtained from Sigma Chemical Co. (St Louis, MO), with the following exceptions. Formaldehyde (EM grade) and Ladd-o-lac from LADD (Burlington, VT), glutaraldehyde from Ted Pella (Redding, CA), taxol from Drug Synthesis and Chemistry Branch, Developmental Therapeutics Program, Division of Cancer Treatment, of the National Cancer Institute (Bethesda, MD), and Hoechst 33258 from Molecular Probes (Eugene, OR). Antibodies: KMX-1, a mouse IgG monoclonal antibody from Boehringer Mannheim (Germany); DM1A a mouse IgG antibody (ICN Immunobiologicals, Costa Mesa, CA); alkaline phosphatase-conjugated anti-mouse IgG serum and Texas Red-conjugated anti-mouse IgG antibody Jackson ImmunoRe-search Laboratories, Inc. (West Grove, PA). Vectastain ABC Elite Mouse IgG kit was from Vector Laboratories, Inc. (Burlington, CA). The sample of a purified mouse tubulin was a generous gift of Dr B. Weinstein (MGH, Boston, MA).

Wild-type AB strain zebrafish embryos were obtained by single pair crosses. Embryos were collected within the first hour after fertilization and maintained at 28.5-29.5°C in ‘egg water’ (0.03% Instant Ocean salt mix in Millipore MilliQplus deionized water) (Westerfield, 1993). From each egg lay 8-cell embryos were selected, ensuring that they differed less than 15 minutes with respect to the time of fertilization. Staging of embryos was performed as described by Kimmel et al. (1993). Developmental times are given in hours after fertilization.

Nocodazole (Methyl-5[2-thienylcarbonyl]1H-benzimidazol-2-yl)-carbamate was prepared as a stock solution of 10 mg/ml in 50% DMSO and diluted to concentrations of 0.5-20 μg/ml. Taxol was prepared as a 10 mM stock solution in DMSO and diluted to concentrations of 10-100 μM. In the case of the 50 and 100 μM dilutions, precipitation was observed. Therefore the actual concentrations of taxol were probably lower. Culture of control embryos in corresponding concentrations of DMSO in egg water did not have any adverse effect on development.

Total proteins from 4- and 20-hour-old embryos were resolved in 10% SDS-acrylamide gel and transferred to Immobilon™PVDF membrane (Millipore, Bedford, MA) (Towbin et al., 1979). Immunoblotting was performed as described in Birkett et al. (1985). Two monoclonal antibodies, α-tubulin-specific DM1A (Gard, 1991) and β-tubulin-specific KMX-1 (Birkett et al., 1985), detected a single band corresponding to protein(s) of apparent molecular mass about 50×10³ in the samples from zebrafish embryos. The detected zebrafish protein(s) comigrated with protein(s) recognized by the two anti-bodies in a fraction of purified mouse tubulins (data not shown). This indicates that the KMX-1 and DM1A antibodies most likely specifically recognize tubulin in zebrafish.

The best preservation of microtubules and structure of the embryo were achieved with the formaldehyde-glutaraldehyde-taxol fix (FGT fix) developed for Xenopus oocytes (Gard, 1991). Embryos were fixed at room temperature or at 28.5°C for 2-4 hours. As in Xenopus (Gard, 1991), microtubules were also preserved when taxol was omitted from the FGT-fix (data not shown). Staining with the KMX-1 and DM1A antibodies was performed as described by Gard (1991) with the following modifications. Incubations with antibodies were carried out overnight in a cold room. Washes were performed at room temperature: 1× 5 minutes and 3× 30 minutes. Embryos stained with the secondary antibody conjugated to Texas Red were dehydrated and mounted in benzyl benzoate-benzyl alcohol (Gard, 1991), in bridge slides described by Warga and Kimmel (1990), and sealed with Ladd-o-lac. Alternatively, detection of the antigen was carried out using the Vectastain Avidin/Biotin/Horseradish peroxidase ABC Elite System according to standard procedures (Roth et al., 1989). Embryos stained with the Vectastain ABC detection system were kept in 100% glycerol. To visualize DNA, embryos stained with antibodies were washed 4-5 times for 5 to 10 minutes with deionized water, and then stained for 10 minutes with 0.2 μg/ml solutions of DAPI (4′,6-diamidino-2-phenylindole) or Hoechst 33258. Embryos stained with DNA dyes were washed once with PBS and transferred to 1% w/v p-phenylenediamine, 90% v/v glycerol, 0.1 M Tris-HCl, pH 7.5.

Observation of living embryos was performed either with a dissecting microscope Wild, or with a Zeiss Axiophot microscope using Nomarski optics. Embryos stained with fluorescent antibodies or DNA dyes were observed under epifluorescent illumination using Neofluar objectives on the Axiophot microscope, or using confocal imaging (Bio-Rad MRC 600 attached to a Zeiss Axioscop).

Two anti-tubulin monoclonal antibodies, DM1A and KMX-1, were examined for their ability to detect zebrafish tubulins by western blotting (see Materials and Methods) and by whole-mount immunocytochemistry. In early cleavage-stage embryos the marginal blastomeres contacting the yolk cell maintain the cytoplasmic bridges with the YCL (Kimmel and Law, 1985a). At these stages, we observed an array of microtubules that appeared to emerge from the marginal blastomeres into the yolk cell, and extended along the animal-vegetal axis towards the vegetal pole. The YCL microtubules persisted even during synchronous mitotic divisions of the blastoderm cells (Fig. 2A,B, and data not shown).

At the 500-1000 cell stage (3h) the marginal blastomeres fuse with the yolk cell (Kimmel and Law, 1985b) forming the YSL. At this stage the syncytial nuclei were organized in a single row in the narrow external YSL (Fig. 3A) (Trinkaus, 1993; Wilson, 1889). In the late blastula embryos (3.7-4h) the enlarged external YSL usually took shape of a syncytial corona (Long, 1980; Ballard, 1973), with single nuclei in each of the vegetal protrusions of the corona (Fig. 4A). Within individual blastomeres of the late blastula embryos, microtubule arrays typical of animal cells were observed: mitotic spindles, midbodies and interphase arrays emanating from centers adjacent to cell nuclei (Fig. 3B). The yolk cell exhibited two distinct types of microtubule organization. In the external YSL microtubules radiated from centers associated with the syncytial nuclei both during mitosis (prometaphase in Fig. 3B,C) and interphase (see below) forming a network. During metasynchronous mitotic divisions of the YSN (Kane et al., 1992), astral microtubules of adjacent mitotic arrays formed in the YSL appeared to interdigitate (Fig. 3C). The YSL network of intercrossing or mitotic microtubules usually filled the entire syncytial corona (Fig. 3B,D). The YCL exhibited a distinct set of microtubules that extended vegetally from the centers serving also as mitotic poles and associated primarily with the most vegetal YSN (Fig. 3B,D,E). In the YCL these micro-tubules formed a dense array aligned along the animal-vegetal axis (Fig. 3F). Some of these microtubules appeared to end at various latitudes of the yolk sphere while the remaining micro-tubules of the array extended to the vegetal pole where they

met in a concentric fashion (Fig. 3G). In some experiments microtubules were observed to cover only 75-90% of the yolk sphere. The YCL microtubules persisted during divisions of the YSN when only mitotic spindle microtubules were visible in the YSL.

Sections of embryos previously stained with anti-tubulin antibodies revealed that the yolk cell microtubules were present only in the thin cortical cytoplasm of the YCL and external YSL. In the blastoderm, microtubules were stained deeper than in the yolk cell, in two to four layers of blastomeres; however, the internal layers were not stained. Staining of embryos bisected after fixation additionally revealed micro-tubules in the deeper blastoderm layers and in the internal YSL, while no microtubules were detected in the deeper, yolk-containing center of the yolk cell (data not shown). We cannot, however, exclude the possibility that the apparent lack of microtubules in the center of the yolk cell is due to incomplete fixation.

Epiboly in zebrafish and Fundulus starts in the late blastula, after cessation of mitotic divisions of the YSN (Kane et al., 1992; Trinkaus, 1993). At the onset of epiboly, in Fundulus, the external YSL contracts and the YSN become densely packed (Trinkaus, 1984; 1993). Similarly in zebrafish, starting at the sphere/dome stage, the wide belt of the external YSL narrowed in the animal-vegetal direction and the syncytial nuclei became increasingly crowded. When the blastoderm expanding vegetally, reached 30% of yolk sphere latitude (30% epiboly), the nuclei of the external YSL concentrated near and were partially covered by the rim of the blastoderm (Fig. 4A-B′). Simultaneously, the microtubule network of the external YSL became denser (Figs 4C, 1B). By 40% epiboly usually most of the YSN were covered by the blastoderm and only a narrow belt of the external YSL with a dense microtubule network was visible below the blastoderm rim (Fig. 4D). At 50% epiboly, the blastoderm margin arrives at the equator of the embryo and deep cells engage in the gastrulation movements of involution and convergence (Kimmel and Warga, 1990). At this stage blastoderm cells almost completely covered the YSN and the associated microtubule network. The YCL microtubules emerged from below the rim of EVL cells (Figs 5E,F, 1C).

The initial stages of gastrulation are correlated with a transient cessation of epiboly by deep cells (Warga and Kimmel, 1990). Consequently, after formation of the embryonic shield, when the EVL covered approximately 65% of the yolk cell, the deep cells lagged in epibolic movements (Fig. 5A,C). At this stage, the YSN, surrounded by their microtubule network, reemerged from below the EVL rim around the circumference of the embryo and populated up to 70% of the yolk cytoplasm (Figs 5A, 1D). Some of the nuclei of the external YSL exhibited variable, elongated shapes (Fig. 5A,B). The YSL microtubules surrounded the nuclei forming basket-like arrangements (Fig. 5B,C). The expansion of the YSN and their associated microtubule network, and the apparent shortening of the YCL microtubules continued until the YSL and the blastoderm covered the entire yolk cell (data not shown).

The analysis described above revealed changes in the organization of microtubule arrays in the external YSL and in the YCL. To determine the microtubule configuration in the internal YSL, the blastoderm cap was removed manually from fixed embryos. Uncapped yolk cells were stained with the anti-tubulin antibody. Fig. 6 shows the microtubule configuration of the yolk cell from an embryo at 60% epiboly. As in the external YSL, microtubules of the internal YSL form a dense network surrounding the YSN. Such an organization of the internal YSL microtubules was observed from the dome stage, until the completion of epiboly.

To investigate a potential role of microtubules in epibolic movements, we studied the effects of the microtubule depolymerizing agent nocodazole (Lee et al., 1980), and the microtubule stabilizing agent taxol (Schiff et al., 1979), on epiboly of the YSN, YSL and blastoderm. Embryos were cultured in the presence of drugs starting at the late blastula (3.7-4 h, Fig. 3D), or 40% epiboly stages. The progress of development was monitored in treated and control embryos for the next several hours until the completion of epiboly in control embryos. First, at the time when control embryos reached 30 or 40% epiboly, treated embryos were inspected with Nomarski optics to determine whether the thick external YSL became narrow and the YSN densely packed (Trinkaus, 1993). Second, the epibolic movements during the later stages of epiboly were monitored by determining the vegetal extent of the YSN, the rim of EVL and of deep cells in live and fixed embryos. The distribution of YSN and the organization of microtubules were also assessed in embryos fixed and stained with anti-tubulin antibodies and a DNA dye.

Treatment of the late blastula stage embryos with 5-20 μg/ml nocodazole dramatically affected the organization of micro-tubules, cell divisions, epiboly and gastrulation. In embryos fixed 30 minutes after addition of 10 μg/ml nocodazole, micro-tubule arrays of blastoderm and yolk cell were completely dis- organized (Fig. 7A). After one hour of incubation with nocodazole, microtubules were no longer detected in the blastoderm cells, while only few short microtubules still persisted in the yolk cell (data not shown). Blastoderm cells remained large throughout the experiment, indicating that cell divisions were inhibited (Fig. 8). Neither germ ring nor embryonic shield were detected, indicating that both involution and convergence towards the dorsal side (Warga and Kimmel, 1990) were either greatly repressed or completely stopped.

Distinct aspects of epiboly were variably impaired in the absence of microtubules. The external YSL remained a wide belt below the blastoderm ring and no crowding of the YSN was observed (Fig. 7C). Embryos treated at the sphere stage (4h) for 1h with 0.5 μg/ml nocodazole exhibited numerous long microtubules of the yolk cell (data not shown). In these embryos, in contrast to the embryos treated with higher concentrations of nocodazole, the YSN concentrated close to the blastoderm rim (Fig. 7B versus C). These observations are consistent with a function of the yolk cell microtubules in the com-paction of YSN and contraction of the YSL at the beginning of epiboly.

In the later stages of epiboly, the movements of the YSN towards the vegetal pole appeared blocked in the nocodazole treated embryos, while the vegetal expansion of EVL and deep blastoderm cells were only partially inhibited. At 75% epiboly in control embryos, YSN covered up to 80% of the yolk sphere (Fig. 8A,D). At the same time in treated embryos, the YSN reached only 40% of the yolk sphere latitude and did not exhibit elongated shapes, characteristic to this stage of epiboly (Fig. 8F). The border between the YSL and the YCL was not easily visible in the nocodazole-treated embryos. Therefore, we could not determine whether movements of other components of the YSL were also inhibited. The margin of deep cells was uneven, and reached only to 40-50% of the yolk sphere latitude, almost completely covering the YSN (Fig. 8B,F). The EVL rim arrived, however, at the 55-60% of the yolk sphere latitude, and was vegetal to both the deep cells and the YSN (Figs 8B,E, 1E). The EVL and deep cells of treated embryos moved very slowly toward the vegetal pole. When control embryos were at the tail bud stage (9.5h), the EVL rim of the treated embryos was about 70% of the yolk sphere latitude, and treated embryos started to degenerate (data not shown).

Thus, in the nocodazole-treated embryos all epibolic movements of the YSN were either severely inhibited or completely blocked. In contrast, epiboly of EVL was only partially repressed. It has been proposed that epiboly of the EVL is passive and is driven by epiboly of the surface of the YSL to which the EVL is tightly linked (Betchaku and Trinkaus, 1978). Epiboly of the surface of the YSL has been explained in part by the process of endocytosis localized during epiboly to the narrow ring of the external YSL, vegetally to the EVL rim (Betchaku and Trinkaus, 1986). To test whether the process of endocytosis is still occurring in embryos in which microtubules have been disrupted, embryos treated for 1 or 2 hours with 5 or 10 μg/ml nocodazole and control embryos were incubated in a solution of Lucifer Yellow (Betchaku and Trinkaus, 1986). Fluorescent microscopy revealed a ring of endocytic vesicles localized vegetally to the blastoderm margin in control embryos (Fig. 8G-I), as well as in nocodazole-treated embryos (Fig. 8J-L). This observation suggests that the disruption of microtubules does not equally affect all the aspects of epiboly in zebrafish embryos, but rather primarily impairs movements of the YSN.

30 minutes after incubation of sphere-stage embryos (4h) in 100 μM taxol, the YSN were covered by the blastoderm, and only a belt of a dense network of the YSL microtubules was visible vegetal to the blastoderm rim (Fig. 7D). Thus the contraction of the YSL was not inhibited. In taxol-treated embryos both the YSL, and the YCL microtubule arrays had a denser appearance than in control embryos (Fig. 7D versus Fig. 4D), covered the vegetal pole completely and were more resistant to nocodazole (data not shown). The movements of the YSN, EVL and deep cells towards the vegetal pole in the later stages of epiboly were delayed. In contrast to nocodazole treatment, in embryos treated with taxol, epibolic movements of YSN, EVL and deep cells were affected to a similar extent. When control embryos reached 60% epiboly (s.d.±6%, 48 embryos examined), treated embryos exhibited only 48% epiboly (s.d.±6%, 46 embryos examined). At the completion of epiboly in control embryos (Fig. 9A, 9h; 96±5%, 23 embryos examined), the blastoderm of treated embryos arrived at 82±7% of the yolk sphere latitude (35 embryos examined; Fig. 9B,C). However, when control embryos were at the 2-somite stage (10.5h), the embryos treated with taxol also exhibited two somites and a forming notochord (Fig. 9D,E). This indicated that taxol treatment affected only epiboly, but not gastrulation or morphogenesis. However, most of the taxol-treated embryos have not completed epiboly at this stage. Instead, they formed abnormally shaped gastrulae, with a portion of the yolk cell protruding from contracted blastopore lips (Fig. 9D). Embryos continued development in the taxol solution for the next several days, although up to 30% of them exhibited tail defects (data not shown).

Examination of the tubulin staining pattern of taxol-treated embryos delayed in epiboly revealed bundles of microtubules and large gaps in the microtubule arrays of the YCL and YSL (Fig. 7E). Examination of live taxol-treated embryos with Nomarski optics indicated that the yolk cortical cytoplasm exhibited apparent discontinuities or thinner regions between thicker cytoplasmic regions (Fig. 9C,F). The apparent dis-continuities of the YCL observed in live embryos most likely corresponded to gaps in the microtubule array, with thick cytoplasmic regions corresponding to bundles of microtubules.

This work demonstrates that the yolk cell of the zebrafish embryo is equipped with two distinct microtubule arrays. The YSL exhibits a network of mitotic or interphase microtubules, while an array of microtubules aligned along the animal-vegetal axis exists in the YCL (Fig. 1A). The developmental changes in the organization of these microtubule arrays correlate with epibolic movements (Fig. 1A-E). We provide evidence that these microtubule arrays are actively involved in the epibolic movements of the YSN, and contribute to other aspects of epiboly.

During early cleavage stages the YCL microtubules appear to emerge from the marginal blastomeres. Most likely micro-tubules extend through large cytoplasmic bridges connecting these blastomeres with the YCL (Kimmel and Law, 1986a). After formation of the YSL, microtubules of the YCL appear to emanate from the centers associated with the syncytial nuclei. Since these centers also radiate mitotic spindle micro-tubules during the mitotic divisions of the YSN, they probably correspond to true microtubule organizing centers. We presume that when the marginal blastomeres completely fuse with the yolk cell, generating the YSL (Kimmel and Law, 1985b), both nuclei and microtubule organizing centers are introduced into the yolk cell. Additionally, the animal-vegetal YCL microtubules are likely to exhibit uniform polarity: the minus ends in the blastomeres or in the YSL with plus ends pointing toward the vegetal pole.

During the period of mitotic divisions of the YSN, the external YSL expands and the YSN spread towards the vegetal pole while maintaining a very regular distribution (Trinkaus, 1993 and this work). We found that the astral microtubules of neighboring spindles in the YSL overlap and interdigitate. A similar configuration of interdigitating astral microtubules has been reported for syncytial nuclei undergoing cortical migration during the 8th and 9th division cycles in the Drosophila embryo (Baker et al., 1993). These authors propose that microtubule-dependent forces generated by plus-end directed microtubule motors act between anti-parallel astral microtubules of adjacent spindles to push nuclei apart. These putative repulsive forces drive nuclei toward the surface (Baker et al., 1993). A similar array of forces could explain both the spreading of the YSN and their regular distribution during mitotic cycles.

Our analysis revealed that the changes in the organization of the yolk cell microtubules correlate with both the process of crowding of the YSN at the beginning of epiboly, and with their subsequent vegetal movements (Fig. 1). These correlations could result from the action of a yet different cellular component, such as a microfilament network (Betchaku and Trinkaus, 1978), acting on both the YSN and the microtubules, or on the entire YSL. Alternatively, the correlations could indicate that microtubule activity is required for epibolic movements. This alternative is supported by analysis of embryos treated with nocodazole. In the absence of micro-tubules, the YSL did not contract and the YSN did not become densely packed. Further, YSN movement toward the vegetal pole in the later stages of epiboly was greatly inhibited. Additionally, the YSN of the nocodazole-treated embryos did not exhibit elongated shapes observed during epiboly in control embryos (Fig. 1D,E). This indicated that the tension that stretches the nuclei along the animal-vegetal axis during epiboly is reduced or absent from the nocodazole-treated embryos.

Several observations argue that the inhibition of nuclear movements in the nocodazole-treated embryos results specifically from the loss of microtubule function. First, some treatments with nocodazole that led to inhibition of the nuclear movements were initiated at the sphere or at later stages of epiboly, after the cessation of mitotic divisions of the syncytial nuclei (Kane et al., 1992). Thus, observed effects are rather unlikely to be an indirect consequence of interference with proliferation of the YSN. Second, when sphere-stage embryos were treated with a low concentration of nocodazole such that disruption of microtubules was delayed, then the YSN became densely packed near the blastoderm rim. Finally, epiboly of blastoderm proceeded farther than epiboly of the YSN. We cannot exclude, however, that some aspects of inhibition of epiboly in nocodazole-treated embryos were secondary to the disruption of microtubules.

Microtubule depolymerizing drugs are known to inhibit directional movements of treated cells (Vasiliev, 1991). Loss of directional movements of deep cells could both explain the impairment of gastrulation movements (Winklbauer and Nagel, 1991), and account for the observed slow vegetal expansion of deep cells in the embryos treated with nocodazole. Epiboly of the EVL was the least affected in the absence of microtubules. It has been proposed that epiboly of the EVL is accompanied and driven in part by a process of endocytosis localized vegetally to the blastoderm rim (Betchaku and Trinkaus, 1986). We have demonstrated that localized endo-cytosis does occur in embryos treated with nocodazole and, thus, it could account for the observed expansion of EVL. This observation underscores the requirement for microtubule function specifically in YSN movements.

Taxol treatment experiments also support the involvement of microtubules in epiboly. In embryos treated with taxol, the vegetal movements of both the YSN and blastoderm were delayed to the same degree. The inhibition of epiboly was specific, since gastrulation and morphogenesis seemed to progress normally. In treated embryos, microtubules were longer than those of untreated embryos, more resistant to nocodazole, and often microtubule bundles and gaps in microtubule arrays were observed. Thus, inhibition of epiboly is most likely mediated by the effect of taxol on microtubules of the yolk cell. Interphase PtK2 cells treated with taxol form long microtubule bundles. This is accompanied by the loss of centrosome-nucleated microtubules (DeBrabander et al., 1981). Therefore, the changes in the microtubule organization in the yolk cell of taxol-treated embryos are consistent with effects of taxol on interphase cells.

How might taxol delay epiboly? Stabilization of micro-tubules by taxol has been proposed to explain the inhibition of pronuclear movement in sea urchin embryos (Schatten et al., 1982). Since YCL microtubules become progressively shorter during epiboly, taxol stabilization could interfere with this process. Alternatively, impairment of epiboly could be only a secondary consequence of the increased microtubule stability.

The YSL provides the major force in the vegetal spreading of the overlaying blastoderm (Trinkaus, 1951; Betchaku and Trinkaus, 1978). The following mechanisms were proposed to be involved in the epiboly of the YSL by Trinkaus and his col-leagues (Trinkaus, 1984). Microfilaments forming a concentric ring in the external YSL were proposed to generate a con-strictive force that would drive expansion of the YSL. Epiboly of the surface of the yolk cell involves expansion of the surface of the YSL with simultaneous diminishing of the surface of the YCL. The expanding surface of the YSL would tag along the EVL attached to it. The diminishing of the surface of the YCL is explained by a process of programmed endocytosis taking place in the area of the external YSL in front of the EVL. Con-current expansion of the YSL surface would be achieved by the release of the excess of membrane stored in the internal YSL in a form of membranous microvilli.

Recently, it has been proposed that an array of YCL microtubules would aid epiboly by pulling the blastoderm toward the vegetal pole (Strähle and Jesuthasan, 1993). The taxol-induced general delay in epiboly that we observed is reminiscent of the phenotype described for embryos treated with UV or short pulses of nocodazole, when the YCL microtubules are present but somewhat disorganized (Strähle and Jesuthasan, 1993). The above treatments indicate that microtubules contribute to general epibolic movements. However, the observation of a block of YSN movements, with only partial inhibition of epiboly of the blastoderm and an unaffected localized endocytosis in the absence of microtubules, points to a more complex role of microtubules in epiboly. Therefore, we propose that the epibolic movements of the yolk syncytial nuclei are not coupled to the process of endocytosis and to epiboly of the blastoderm. Further, epibolic movements of the YSN but not of the blastoderm are absolutely dependent on microtubules.

We hypothesize that both the YSL network and the YCL microtubules contribute differently to various aspects of epiboly, and that they are primarily involved in movements of the YSN. Microtubules have been implicated in nuclear movements in a variety of organisms, based on either morphological criteria (Picket-Heaps, 1991; Lloyd et al., 1987), correlation between the organization of microtubules and nuclear movements (Baker et al., 1993), sensitivity of nuclear movements to anti-microtubule drugs (Picket-Heaps, 1991; Lutz and Inouye, 1982; Oakley and Morris, 1981), or on a genetic evidence (Huffaker et al., 1988; Morris, 1976). One possible mechanism for microtubule-mediated nuclear migration involves molecules that interact between the micro-tubules and a nucleus generating force to move it along micro-tubules (e.g. Osmani et al., 1990). Such a mechanism, involving the YSL and YCL microtubules and a plus end-directed microtubule motor, could explain epibolic movements of the YSN. Alternatively, motor molecules working between interdigitating microtubules in the YSL network would generate a pushing force similar to that used in anaphase B of mitosis, effectively expanding the YSL and spreading YSN (e.g. Nislow et al., 1992; reviewed in Saunders, 1993).

The YCL animal-vegetal microtubules would act in the expansion of the margin of the YSL and possibly in transport of cellular components localized in the external YSL. Since the cellular components of the external YSL, like microfilaments, may also be involved in epiboly; interference with the YCL microtubules could affect all epibolic movements. Motor molecules, similar to those involved in pulling the spindle pole by astral microtubules (Saunders, 1993), would pull their organizing centers located in the YSL toward the vegetal pole. Coordinate action of the YSL and YCL arrays would result in the expansion of the whole YSL, whereas action of the YSL network alone at the beginning of epiboly could account for crowding of the YSN.

Distinguishing between these models will require identification of the involved motor molecules and their functional analysis during epiboly. Additionally, it is important to under-stand the relationship between the microtubule and microfila-ment systems in this process. Since the large syncytial yolk cell proves very accessible to experimental analysis, further studies of microtubule arrays of the yolk cell in zebrafish can con-tribute to our understanding of the role of the cytoskeleton in vertebrate development.

We are thankful to Elisabeth Vogelsang for sectioning embryos stained with anti-tubulin antibodies, and to Dr Rocco Rotello for help with western blotting. Special thanks to Dr Robert M. Ezzell for giving us access to the confocal microscope, and Dr Yimin Ge for instructions on collection and processing of confocal images. We would like to thank our colleagues Derek Stemple, Timothy Burland, Alexander Schier, Eliza Shah, Brant Weinstein, Fried Zwartkruis and Dr John Trinkaus for the constructive comments on the manuscript. Part of this work has been presented in an abstract form at the Northeast Regional Developmental Biology Conference in Woods Hole, March 1993. This work has been supported by National Institute of Health grant HD29761 and by Bristol-Myers Squibb.