The fundamental body plan of zebrafish embryos begins to emerge during the gastrula period when a series of extensive cell movements and rearrangements (epiboly, involution/ingression, extension, and convergence), result in the development of the three germ layers, i.e., the endoderm, mesoderm, and ectoderm, as well as the dorsoventral and anterioposterior body axes.
Epiboly is the first of these morphogenetic movements to occur, starting at the end of the blastula period (Kimmel et al., 1995). At the onset of epiboly, the embryo, which has just completed a series of rapid cleavages, consists of a high-piled mound of cells, a blastoderm, sitting on top of a large syncytial yolk cell (Kane and Adams, 2002). The developing blastoderm is composed of a cohesive layer of surface peridermal cells called the enveloping layer (EVL), which encloses a more loosely packed population of deep cells (DCs). The blastoderm is in contact with the yolk cell by means of the yolk syncytial layer (YSL). This structure forms during the blastula period when the blastomeres that lie against the yolk cell at the rim of the blastoderm collapse to release their cytoplasm and nuclei into the adjoining cytoplasm of the yolk cell. When it first forms, the YSL is a narrow ring around the edge of the blastoderm (called the external YSL, or E-YSL), but within a few cell division cycles, it spreads underneath the blastoderm to additionally form a complete internal syncytium (called the internal YSL, or I-YSL; Kimmel et al., 1995). Once formed, the I-YSL is a thin layer of relatively sparsely nucleated cytoplasm that underlies the blastoderm, while the E-YSL is a broader belt of cortical cytoplasm that exhibits more densely packed nuclei. This belt lies between the I-YSL and the yolk cytoplasmic layer (YCL), a thin, non-nucleated, superficial layer of cytoplasm that encompasses the rest of the yolk cell (Sakaguchi et al., 2002; Keller et al., 2003). During the earliest stages of epiboly, the yolk cell changes shape as the surface of the I-YSL (which was up until then flat), domes upward toward the animal pole, and the blastoderm changes from a high-piled mound of cells to a cup-shaped multilayer of cells around the doming yolk (Kimmel et al., 1995). The blastoderm and YSL then progressively spread over and around the remaining yolk cell toward the vegetal pole until, by the end of the gastrula period, they completely cover the yolk cell. During this process, the YCL and yolk cell membrane are sequentially assimilated and removed, respectively.
Even though some of the earliest descriptions of epiboly in fish were published more than a century ago (Morgan, 1895), the mechanisms involved in and forces that drive this process are still not well understood, although some of the components have now been identified. The E-YSL has long been suggested to be a major driving force for epiboly after Trinkaus (1951) demonstrated in the killifish, Fundulus heteroclitus, that epiboly of the E-YSL can occur autonomously, i.e., in the absence of the blastoderm. More recently, components of the cytoskeleton, namely microfilaments and microtubules, have been proposed to play a role in driving epiboly. By using electron microscopy, Betchaku and Trinkaus (1978) reported that, in Fundulus, networks of thin (4–6 nm) microfilaments are found in the cortex of EVL cells (particularly at the leading edge of the marginal cells), in the cortex of the YCL and in the cortical cytoplasm of the E-YSL. They also reported the presence of thicker (10 nm) microfilaments that were arranged in bundles in the marginal cytoplasm of each marginal EVL cell and in the E-YSL beneath the EVL marginal contacts (Betchaku and Trinkaus, 1978). They suggested that the thin microfilaments in the E-YSL might play a key role in epiboly by exerting tension on the attached margin of the EVL cells and discussed how the other populations of microfilaments might provide the additional contractile forces required to accomplish this process.
Since then, Strähle and Jesuthasan (1993) demonstrated that, in zebrafish, epiboly is driven to a certain extent by microtubules that extend from the YSL into the YCL. They showed that the irradiation of embryos with ultraviolet light or treatment with the microtubule depolymerizing agent, nocodazole, resulted in impaired epiboly. Subsequently, Solnica-Krezel and Driever (1994) demonstrated, again in zebrafish, that there are in fact two well-defined arrays of microtubules: one is an intercrossing network in the YSL, while the other emanates from this YSL network and extends along the animal–vegetal axis in the YCL. They demonstrated that, when embryos were treated with the microtubule stabilizing agent Taxol, the whole of epiboly was delayed. When embryos were treated with nocodazole, however, the epibolic movements of the YSL nuclei were blocked but epiboly of the EVL and DCs was only partially inhibited by the drug (Solnica-Krezel and Driever, 1994). Furthermore, the endocytosis previously suggested to play a key role in the epiboly of the YSL (Betchaku and Trinkaus, 1986) was still observed to occur in nocodazole-treated embryos. Solnica-Krezel and Driever (1994) subsequently concluded that, while microtubules are required for epibolic movements of the YSL nuclei in zebrafish, they are only partially needed for the other processes involved in epiboly.
Thus, in an attempt to identify the components that drive epiboly of the EVL and DCs, we have begun to reinvestigate the role of microfilaments, first proposed by Betchaku and Trinkaus (1978), in this process in zebrafish. Zalik et al. (1999) have described the localization of F-actin in zebrafish embryos from the earliest cleavage stages to 50% epiboly. They demonstrated that, during the first half of epiboly, the tight cohesion of the cells in the EVL is maintained by actin in the cell cortex that colocalizes with cell surface cadherins, which are known to maintain the integrity of cell groups through the formation of adherens junctions (Lilien et al., 2002). At the margins approaching the YSL, however, the EVL cells extend actin-rich but cadherin-free lamellipodial extensions. It was suggested, thus, that these marginal EVL cells might actively participate in early epiboly (Zalik et al., 1999).
Here, we reinvestigate the localization of microfilaments during epiboly in zebrafish embryos, and we present new data to show that several novel actin-based structures appear after 50% epiboly. In addition, we have demonstrated previously that a rhythmic series of intercellular Ca2+ waves traverse the blastoderm margin of zebrafish embryos. These waves first appear at ∼65% epiboly and subsequently arise every 5–10 min throughout the latter half of epiboly (Gilland et al., 1999; Webb and Miller, 2003). To investigate the significance of these Ca2+ signals during zebrafish epiboly, the Ca2+ chelator 5,5′-dibromo-BAPTA (DBB) was introduced into the blastoderm margin at mid-epiboly and its effect on epiboly and the organization of microfilaments was determined. The possible role of these new F-actin structures in driving the later stages of epiboly and the role of Ca2+ in their formation and function is discussed.
Organization of the Actin Cytoskeleton During Epiboly
Zebrafish embryos were fixed at 30% epiboly, shield stage, 85% epiboly, and 95% epiboly and then incubated with fluorescein isothiocyanate (FITC) or rhodamine-tagged phalloidin to label the F-actin and examined by confocal microscopy (fixation series n = 4; see Fig. 1). F-actin is localized at the periphery of the enveloping layer (EVL) cells and in the cortex of the exposed yolk cell in the vegetal region at all stages examined (Fig. 1A–E). The vegetal actin structure is most extensive at 30% epiboly and shield stages (Fig. 1A,B) but is diminished considerably by 85% epiboly (Fig. 1C) and has all but disappeared by 95% epiboly (Fig. 1D). At 85% epiboly, two new F-actin–based structures are apparent when embryos are viewed at low magnification (see Fig. 1C). These structures appear as two adjoining rings, one wider than the other, which are localized at the EVL margin (Fig. 1C). The wider of these two rings is also apparent at 95% epiboly (Fig. 1D).
The F-actin structures observed at 85% epiboly can be seen in more detail at higher magnification (see Fig. 1E). Figure 1Ei shows a stack of 72 optical slices (z step of 1 μm) that have been projected as a single image. As well as the F-actin at the periphery of the EVL cells and in the YCL at the vegetal pole that was observed at low magnification (Fig. 1A–D), it is now possible to additionally see F-actin structures: (1) in the DC population particularly at the leading edge margin of the DCs, (2) a ring at the leading edge of the EVL, and (3) a punctate band (of ∼16 μm in width) that lies vegetal to the EVL leading edge ring located in the E-YSL periphery. Selecting single optical sections from the stack in Figure 1Ei, it is possible to see more clearly F-actin that is located at the periphery of individual DCs and also, more obviously, at the leading margin of the DC population (Fig. 1Eii) as well as the ring of F-actin at the leading edge of the EVL and the punctate band at the E-YSL periphery, at the vegetal rim of the EVL leading edge (Fig. 1Eiii). The mean ± SEM width of the punctate actin band is ∼16 ± 1.5 μm (n = 15).
The presence of the actin structures identified by labeling with FITC–phalloidin was confirmed by injecting live embryos with rhodamine-tagged actin at the single-cell stage and then observing the fluorescence from “total” actin (i.e., both G- and F-actin) during epiboly (see Fig. 2). Figure 2A is a projected view of 10 optical sections (z step is 1 μm) through the leading edge of the blastoderm in a representative embryo at ∼80% epiboly (n = 10). These images confirm that the compact punctate actin band is closely associated with the leading edge of the EVL, and at its vegetal margin, it becomes more diffuse as it extends into the peripheral cortex of the yolk (Fig. 2A). The compact actin band in this representative example is ∼10 μm wide and ∼6 μm deep, whereas the diffuse actin clusters in the cortex of the YCL extend less than ∼2 μm deep into the embryo. The mean ± SEM width of the punctate actin band, measured in live embryos labeled with rhodamine-actin, is ∼11 ± 0.7 μm (n = 20 embryos from 10 experiments). Conspicuous levels of actin are also shown in the EVL cell peripheries and in the DCs (see Fig. 2A) and also in the YCL of the unoccluded yolk in the vegetal region, as illustrated in the projected image through the vegetal pole of a representative embryo at ∼80% epiboly (see Fig. 2B).
The formation of the punctate actin band is shown in the representative embryo (n = 10) in Figure 2C. At 40% epiboly, it initially appears as a wide, diffuse band of actin clusters (Fig. 2Ci), then between 50% epiboly and shield stage, the actin clusters become more concentrated (Fig. 2Cii–iii), forming a compact, punctate band at the leading edge of the EVL cells (Fig. 2Civ). The ring of actin at the DC margin can also be observed in these images (Fig. 2Ciii–iv).
Effect of Cytochalasin B on Late Epiboly
Embryos were treated with the microfilament-disrupting agent cytochalasin B at 50% epiboly to determine its effect on the latter stages of epiboly (see Fig. 3). Embryos incubated with 30% Danieau's solution containing dimethylsulfoxide (DMSO) (at 1:500–1:250) completed epiboly at the same rate and were morphologically comparable to embryos incubated in Danieau's solution alone (compare Fig. 3B with Fig. 1C). Embryos treated with cytochalasin B (prepared in DMSO and diluted in Danieau's solution) exhibited a slowing of epiboly, a failure of yolk cell occlusion, and the eventual lysis of the embryo through this unoccluded portion of the yolk cell. In addition, the effect of cytochalasin B on both embryo viability and rate of epiboly in the surviving embryos was dose-dependent. By the time that all (i.e., 100%; n = 8/8) of the DMSO-treated control embryos had reached 95% epiboly, only 60% (n = 15/25) of the embryos treated with 10 μg/ml cytochalasin B survived the treatment and these had only reached 75% epiboly. Furthermore, only 20% (n = 5/25) of the embryos treated with 20 μg/ml cytochalasin B survived the treatment, and these embryos were still at shield stage (Fig. 3A).
Embryos treated with 10 μg/ml cytochalasin B at 50% epiboly were fixed at 8.5 hours postfertilization (hpf) (i.e., when the DMSO-treated control embryos had reached 85% epiboly), and F-actin was labeled with rhodamine-phalloidin (n = 7; see Fig. 3B–G). This made it possible to see that cytochalasin B treatment led to an uncoupling of the DCs and EVL, and blocked epiboly of the DCs to a greater degree than the EVL. Whereas in the control embryos, both the EVL and DCs moved across the yolk more-or-less together such that they both covered the yolk by ∼85% when the embryos were fixed (Fig. 3B), in the cytochalasin B–treated embryos fixed at the same time, the EVL covered the yolk by just 60% and the DCs covered it by just 50% (Fig. 3C). In addition, the F-actin–based structures were disorganized in the cytochalasin B–treated embryos. At low magnification, it is possible to see that, while the actin is still visible at the periphery of the EVL cells, the vegetal actin structure in the yolk cell is completely disrupted and the punctate actin band has a disorganized, almost patchy appearance compared with the compact, well-organized, evenly distributed band in the control embryo (compare Fig. 3C with 3B). The uncoupling of the EVL and DC leading margins in the cytochalasin B–treated embryos compared with the DMSO-treated control embryos, can be seen more clearly in the short stacks of 10 optical sections each, shown in Figure 3D,E. In these representative embryos, the distance between the EVL and DC leading margins in the control and cytochalasin B–treated embryos is ∼28 μm and ∼118 μm, respectively. The organization of F-actin in both cytochalasin B–treated and DMSO-control embryos can be observed in greater detail at higher magnification (see Fig. 3F,G). Here, it is possible to see that the F-actin at the EVL cell peripheries is also disorganized and diffuse compared with the F-actin at the peripheries of the EVL cells in the DMSO controls. In addition, there is a separation of the punctate actin band with the EVL leading margin in the cytochalasin B–treated embryo (see Fig. 3G) where a tight connection can be observed in the control (see Fig. 3F).
Effect of Dibromo-BAPTA and Dimethyl-BAPTA on Epiboly and F-Actin Organization
Embryos were injected into the yolk at the base of the blastoderm at early shield stage with mixtures of rhodamine-labeled dextran and either DBB or 5,5′-dimethyl-BAPTA (DMB). The rhodamine-labeled dextran was used to confirm that the BAPTA buffers were distributed evenly around the rim of the blastoderm margin (data not shown). The embryos were then allowed to develop for a further 2 hr, after which they were fixed and labeled with FITC–phalloidin. Representative examples (n = 5) of DBB- and DMB-injected embryos are illustrated in Figure 4. Whereas DMB-injected embryos undergo normal epiboly (i.e., at 8 hpf, these embryos are at ∼75% epiboly) and display a normal pattern of F-actin labeling (compare Fig. 4A with Fig. 3B and Fig. 4C with Fig. 3F), the DBB-injected embryos undergo developmental arrest at shield stage (Fig. 4B), i.e., shortly after microinjection. Furthermore, the pattern of F-actin at the periphery of the EVL cells seems to be unaffected by DBB treatment. However, the F-actin in the yolk cortex is completely disrupted, the leading edges of the EVL and DCs exhibit an uneven, blebby appearance, and the punctate band at the EVL margin is absent (compare Fig. 4B,D with Fig. 4A,C). DBB-treated embryos eventually lyse through the unoccluded yolk.
A comparison was made between the width of the punctate actin band in embryos treated with DMSO alone and in embryos treated with cytochalasin B, DMB, and DBB (see Fig. 4E). Whereas the mean ± SEM width of the bands in the DMSO-control and DMB-treated embryos was similar, the mean width of the band in the cytochalasin B–treated embryo was ∼3 times wider than the control (i.e., ∼53 ± 3 μm compared with ∼17 ± 0.6 μm) and in the DBB-treated embryo there was no obvious band.
Punctate Actin Band Colocalizes With a Zone of Surface Membrane Foldings and Active Endocytosis
The region of the punctate actin band can also be seen as a distinct structure by Nomarski microscopy (see Fig. 5A). With image enhancement (see Fig. 5B) it is possible to see that this is a region where the membrane appears to be highly folded. The presence of a zone of endocytosis at the EVL margin was demonstrated by the internalization of FITC-labeled dextran (10 kDa; Cooper and D' Amico, 1996). When embryos were loaded with rhodamine-tagged actin at the single-cell stage and then incubated with FITC-labeled dextran at 30% epiboly, there was neither a punctate band of actin nor a zone of endocytosis at the blastoderm margin (data not shown). However, when rhodamine-actin–loaded embryos were incubated with FITC-labeled dextran at 60% epiboly, the punctate band of actin at the EVL margin (Fig. 5C) and a zone of active endocytosis (Fig. 5D) were found to be colocalized (Fig. 5E) exhibiting an overlap of ∼84%.
Amount of Membrane Internalization During Epiboly
The amount of yolk cell membrane internalized between sphere and shield stages, and from shield stage to the end of epiboly was estimated (n = 6). Thus, the percentage mean ± SEM of membrane removed between sphere and shield stages was calculated to be 13.2 ± 1.3% of the total surface area of the yolk cell membrane (measured at sphere stage) with the remaining 86.8 ± 1.3% being removed from shield stage to the end of epiboly.
F-Actin-Based Structures Observed Throughout Epiboly
Embryos fixed and stained throughout epiboly clearly show that F-actin is localized in the periphery of the large angular enveloping layer (EVL) cells (Figs. 1–4). This observation has already been reported in embryos fixed up to 50% epiboly by Zalik et al. (1999). They suggested that, as the EVL cells are tightly attached both to each other and to the underlying peripheral E-YSL and move across the yolk cell as a coherent layer, the F-actin may be involved in cell-to-cell adhesion, possibly by interaction with cell surface cadherins. Zalik et al. (1999) reported that at the margins adjoining the E-YSL, the EVL cells extend actin-rich filopodial protrusions from their apicolateral domains and are capable of lamellipodial activity. They subsequently suggested that the EVL cells may actively participate in epiboly. At the magnification that we used to visualize the embryos during early epiboly, it was not possible to see these filopodial protrusions. However, the uneven appearance of the EVL leading edge (see Fig. 1A), caused by individual cells that appear to be moving ahead of the others, suggests that these cells might be actively migrating at this stage. Once the blastoderm has advanced to the equator, however, the leading edge of the EVL appears as a more uniform, continuous margin (see Fig. 1B) and it remains like this for the latter half of epiboly (see Fig. 1C,D).
Zalik et al. (1999) demonstrated that treatment of embryos at 30–50% epiboly with the actin-disrupting drugs cytochalasin D or dihydrocytochalasin B induced the dissociation of the EVL and yolk cell herniation and frequently led to a complete dissociation of the blastomeres and disintegration of the yolk cell. When we treated embryos with cytochalasin B during early epiboly, we also observed dissociation of the EVL and separation of the other (deep cell) blastomeres (data not shown). However, when we treated embryos at 50% epiboly, the EVL remained intact, although, like Zalik et al. (1999), we did see disruption of the actin at the periphery of the EVL cells (see Fig. 3G). Zalik et al. (1999) also treated embryos with ethyleneglycoltetraacetic acid (EGTA) to determine the effect of removing Ca2+ on embryo integrity. They showed that, although 20 mM EGTA induced cell dissociation, the actin at the EVL cell peripheries was not visibly affected. We treated embryos with another Ca2+ chelator, DBB (Pethig et al., 1989), at ∼5 mM and similarly showed that there was no visible change in the pattern of F-actin labeling at the periphery of the EVL cells compared with embryos treated with DMB (which is a good negative control as its Kd is 10-fold lower than DBB, see Fig. 4C) or untreated controls (see Figs. 1E, 3F).
A dense region of actin was also observed in the vegetal cortex of the unoccluded yolk cell throughout epiboly as demonstrated both in fixed (see Figs. 1, 3B, 4A) and live (see Fig. 2B) embryos. In addition, this vegetal “mat” of actin was disrupted by treatment with either cytochalasin B or DBB (Figs. 3C and 4B, respectively), and embryos always eventually lysed through the unoccluded region of the yolk cell. We therefore suggest that this dense vegetal mat of F-actin contributes to the mechanical integrity of the yolk cell and, thus, helps to maintain the shape of the vegetal portion of the embryo. Such a structure would be particularly important during this stage of development when the major cell rearrangements and morphogenetic movements of involution/ingression, convergence, and extension, as well as epiboly, subject the vegetal region of the yolk cell to a variety of different forces (Keller et al., 2003). In this respect, the F-actin vegetal mat may interact with the population of microtubules that are known to run in the YCL from the E-YSL to the vegetal pole (Solnica-Krezel and Driever, 1994).
F-Actin–Based Structures Observed After 50% Epiboly
As we mentioned above, the EVL leading edge has an uneven appearance during early epiboly but becomes more uniformly continuous at around shield stage (see Fig. 1). At this time, an F-actin ring becomes visible at the leading edge of the EVL where it appears (at the resolution examined) to run from cell to cell around the circumference of the margin. We suggest that this might be a contractile actomyosin ring analogous to those formed during dorsal closure in Drosophila (Martin and Wood, 2002), during ventral enclosure in the nematode Caenorhabditis elegans (Williams-Masson et al., 1997), and in embryonic skin during wound closure (Nodder and Martin, 1997), which operate like a purse-string in these systems to draw together and fuse epithelial edges.
Zalik et al. (1999) reported that at 30% epiboly, the DCs do not exhibit actin localization in the cell cortex. We did not examine the localization of F-actin in the DCs in embryos during early epiboly. However, during the latter stages of epiboly, we visualized clear labeling of F-actin in the periphery of the DCs (see Fig. 1Eii). The pattern of labeling in the DCs is different from that observed in the EVL cell peripheries as it is more diffuse. Interestingly, so-called “meshworks” of cortical microfilaments have also been demonstrated in the deep cells of Fundulus gastrula (Hogan and Trinkaus, 1977). Furthermore, it has been reported previously that a mesh-like network of actin plays a major role in most crawling cells (Verkhovsky et al., 2003). Thus, we suggest that the pattern of actin labeling in the DCs might reflect their motile nature as they are known to undergo active directed migration during internalization, convergence, and extension and display the pronounced blebbing and filopodia that are characteristic of such cells (Warga and Kimmel, 1990; Ulrich et al., 2003).
A continuous ring of actin, similar but again more diffuse than that observed at the EVL margin, also appears to form at the DC leading edge (see Figs. 1Ei, 1Eii, 2Ciii). In a recent review, Kane and Adams (2002) proposed that the movement of cells from superficial to deeper locations occurs within one to two cell diameters of the blastoderm margin. They suggested that the most superficial DCs first move toward the margin of the DC blastoderm, after which they are internalized into deeper locations (Kane and Adams, 2002). Thus, the apparently continuous ring of actin at the DC margin might be a less stable structure, with a higher turnover rate, than the EVL marginal ring as the cells at the DC margin are continually being replaced (Kane and Adams, 2002). There is indirect evidence, from the zebrafish half-baked (hab) mutant (Kane et al., 1996), for the existence of a purse-string mechanism occurring at the DC margin during the latter half of epiboly. Whereas epiboly of the EVL in the hab mutant is normal, epiboly of the DCs is halted at around 50% epiboly; however, a constriction of the embryo still occurs in the DC germ ring (Kane et al., 1996).
Perhaps the most striking F-actin structure that we visualized in both fixed and live embryos was the 10- to 16-μm-wide punctate band of actin in the periphery of the E-YSL at the EVL margin (see Figs. 1–5) that appeared at shield stage and persisted until the end of epiboly. A similar but wider (i.e., ∼40–50 μm) and less compact punctate zone of cortical actin has been reported in another teleost fish embryo, medaka (Oryzias latipes) at 90% epiboly (Cooper and Kimmel, 1998). This zone of cortical actin was reported to be associated with a dense concentration of microvilli produced on the highly convoluted surface of the YSL. Similarly, in zebrafish, we showed that the punctate band of actin corresponds to numerous surface folds (see Fig. 5A,B). In addition, by bathing embryos in a fluorescent dextran to label endocytic vesicles (Cooper and D' Amico, 1996), we demonstrated that this punctate actin band is colocalized with a region of membrane internalization (Fig. 5C–E). Such a zone of endocytosis was first reported during epiboly in Fundulus embryos by Trinkaus (1984), who suggested that this process might play a major role in epiboly by removing the yolk cell membrane in front of the advancing EVL margin (Trinkaus, 1984; Betchaku and Trinkaus, 1986). This region of endocytosis has been confirmed more recently in zebrafish by Solnica-Krezel and Driever, (1994), who demonstrated that this process continues normally in embryos treated with nocodazole and, thus, is not dependent on microtubules. Although the exact role of actin in the endocytic process is not yet fully understood, it has been suggested that it might be involved in generating the contractile forces required to induce the invagination of the plasma membrane, to pinch off the invaginations to produce vesicles, and/or to drive the detached vesicles away from the plasma membrane (for recent reviews, see Qualmann et al., 2000; Jeng and Welch, 2001; Schafer, 2002). We suggest that a combination of these constrictive processes and the direct internalization of the yolk cell membrane might provide the main driving force required to move the connected margin of the EVL toward the vegetal pole.
In Fundulus, the zone of endocytosis occurs throughout epiboly (Trinkaus, 1984). We suggest that, in zebrafish, the punctate actin band (and thus the zone of endocytosis) only forms at shield stage, because it is from then to the end of epiboly that the bulk of the yolk cell membrane has to be removed; therefore, a highly organized system of rapid membrane internalization is required.
Model for the Role of Actin in Zebrafish Epiboly
Whereas the F-actin bundles in the periphery of the EVL cells and the actin mat at the vegetal pole are present throughout epiboly (although the latter diminishes as epiboly progresses), the ring of actin at the EVL leading margin and the punctate actin band in the E-YSL only make an appearance after 50% epiboly, i.e., when the embryonic equator has been overrun by the advancing blastoderm margin. Thus, we propose that the second half of epiboly is driven in part by a different mechanism than the first half and that the new structures that we describe appear in response to changes in the geometry of the somewhat spherical embryo. This process would not be a novel mechanism in morphogenesis as a similar process, involving leading cell migration followed by purse string closure, has been described during ventral enclosure of the hypodermis in C. elegans (Williams-Masson et al., 1997). Furthermore, Keller and Trinkaus (1987) demonstrated that, in Fundulus epiboly, the EVL adjusts to the changing geometry of the spherical embryo by reducing the number of marginal cells ∼6-fold by means of a combination of shape change (with cells becoming tapered at the marginal boundary), and active rearrangement. They subsequently suggested that these morphogenetic changes might act as a driving force in the spreading of the EVL over the yolk. Cell rearrangement has also been demonstrated to occur during epiboly of the EVL in medaka (Kageyama, 1982), and we suggest that either cell rearrangement or shape change, or possibly both, may also play a role in the latter stages of zebrafish epiboly. It is clear that, in zebrafish, before 50% epiboly the blastoderm has to spread over the expanding surface of the animal hemisphere. This spread might occur by a combination of the doming action of the yolk (which may physically push the cells of the blastoderm over and around it; Keller et al., 2003) and active migration of the leading EVL cells (Zalik et al., 1999). However, once the equator is crossed, the situation is reversed, and the blastoderm has to spread over a diminishing surface area as it converges on the vegetal pole. Thus, we suggest that in zebrafish a combination of the actin rings at the EVL and DC leading margins acting as tightening purse strings, as well as possible active cell rearrangement and shape change of the EVL marginal cells, might be required to draw these cells in toward the vegetal pole. In addition, the punctate band of F-actin is associated with a zone of endocytosis. This coordination may help to drive epiboly by means of the constriction of the E-YSL and the exertion of tension on the attached margin of the EVL margin as well as by means of the removal of the yolk cell membrane ahead of the advancing blastoderm as epiboly progresses (Betchaku and Trinkaus, 1978, 1986).
Role for Calcium in Modulating the Actin Structures During Late Epiboly
We have shown previously that a rhythmic series of intercellular Ca2+ waves traverse the blastoderm margin of zebrafish embryos, first appearing at ∼65% epiboly and subsequently arising every 5–10 min throughout the latter half of epiboly (Gilland et al., 1999; Webb and Miller, 2003). Thus, the appearance of the Ca2+ waves follows the appearance of the EVL and DC actin rings as well as the punctate E-YSL actin band. This finding suggests that once in place, individual Ca2+ transients might play a role in the function of these structures. Furthermore, the introduction of the Ca2+ chelator DBB into the blastoderm margin at early shield stage inhibits the formation of the punctate actin band, disrupts the formation of the EVL and DC marginal actin rings, and thus blocks epiboly (Fig. 4). This finding suggests that Ca2+ signals might also play a role in the formation of these structures. The introduction of a similar concentration of DMB, however, has no effect on either epiboly or the formation of the punctate actin band. It has been reported that Ca2+ plays a crucial role in actomyosin purse-string assembly and closure during wound healing (Bement et al., 1999). Thus, the EVL and DC marginal rings might be stimulated to contract by the rhythmic marginal Ca2+ waves, therefore, helping to occlude the yolk cell. Furthermore, although reports have been somewhat contradictory (reviewed by Gundelfinger et al., 2003), it has also been suggested that Ca2+ might modulate actin-based endocytosis. Créton et al. (1998) have reported a continuous elevated level of Ca2+ at the blastoderm margin during epiboly. Thus, we suggest that this process might induce endocytosis in this zone.
FITC–phalloidin and cytochalasin B were purchased from Sigma Chemical Co. (St. Louis, MO). Nonmuscle rhodamine actin, dithiothreitol (DTT), and ATP were obtained from Cytoskeleton, Inc. (Denver, CO), whereas the rhodamine–phalloidin, FITC-labeled dextran (10 kDa), DBB, and DMB, both tetra-potassium salts, were purchased from Molecular Probes (Eugene, OR).
Egg Collection and Dechorionation
Zebrafish (Danio rerio) were maintained on a 14-hr light/10-hr dark cycle to stimulate spawning (Westerfield, 1994). Fertilized eggs were collected and dechorionated as described in detail elsewhere (Leung et al., 1998).
Labeling F-Actin in Fixed Embryos
Dechorionated embryos were maintained at 28°C in 30% Danieau's solution (19.3 mM NaCl; 0.23 mM KCl; 0.13 mM MgSO4 · 7H2O; 0.2 mM Ca(NO3)2; 1.67 mM Hepes, pH 7.2) through the cleavage and blastula stages. At ∼30% epiboly (4.7 hpf), ∼shield stage (6 hpf), ∼85% epiboly (8.5 hpf), and ∼95% epiboly (9.5 hpf), embryos were fixed with 4% paraformaldehyde in phosphate-buffered saline (PBS; Westerfield, 1994) overnight at 4°C, after which they were washed thoroughly with PBS. All subsequent wash steps were for 5 min and were carried out with gentle shaking. Embryos were washed twice with PBS containing 0.1% Tween-20 (PBT) and then once with PBT containing 1% DMSO (PBTD). They were then incubated with blocking buffer (PBTD containing 10% bovine serum albumin; BSA) at room temperature for 2 hr.
To label F-actin, embryos were incubated with either 1 μg/ml FITC–phalloidin or 0.08 μg/ml rhodamine–phalloidin, at room temperature for 1 hr in the dark, after which they were washed extensively with PBTD containing 1% BSA (PBTD/BSA). FITC–phalloidin and rhodamine–phalloidin were prepared as stock solutions of 0.1 mg/ml and 80 μg/ml in methanol, respectively, and then diluted in PBTD/BSA just before use.
Fluorescence was observed by using either a Bio-Rad MRC-600 laser scanning system, mounted on a Zeiss Axioskop, or a Leica TCS SP2 laser scanning confocal unit mounted on a Leica DM IRE2 microscope. Images were collected by using Zeiss water immersion Achroplan 20×/0.5NA, 40×/0.75NA, and 63×/0.9NA objectives on the Bio-Rad MRC-600 system, whereas Leica HC Plan-Fluotar 10×/0.3NA and Plan-Apochromat (oil) 40×/1.25NA objectives were used on the Leica TCS SP2 system. On the Bio-Rad system, FITC and rhodamine fluorescence was visualized by using 488-nm excitation/520-nm emission and 568-nm excitation/585-nm emission, respectively. On the Leica confocal microscope, FITC and rhodamine fluorescence was visualized by using 488-nm excitation/ 500- to 535-nm emission and 543-nm excitation/ 560- to 700-nm emission, respectively.
Labeling Live Embryos With Rhodamine–Actin and FITC–Dextran
To label total actin, dechorionated embryos were injected at the single-cell stage into the yolk at the base of the blastodisc with ∼2 nl of rhodamine–actin (stock at 2 mg/ml in 0.4 mM DTT, 0.16 mM ATP, and 2.5 mM Tris-HCl, pH 8.0). The microinjection pipettes and the pressure injection system used for injecting embryos are described in Webb et al. (1997). Embryos were then maintained at 28°C in the dark until ∼30% epiboly (4.7 hpf). Images were acquired throughout epiboly by using the Bio-Rad MRC600 laser scanning confocal system described earlier.
To label endocytic vesicles, embryos were maintained at 28°C in the dark until either 30% epiboly or 60% epiboly (∼4.7 hpf or ∼7 hpf). They were then incubated with the FITC-labeled dextran (10 kDa) at 10 mg/ml in 30% Danieau's solution for 30 min (Cooper and D' Amico, 1996). After five brief washes with 30% Danieau's solution, images were obtained both immediately and after 10 min by using the Bio-Rad MRC-600 laser scanning confocal system described earlier. The percentage colocalization between the punctate actin band and the zone of endocytosis was determined using the “Measure Colocalization” function in Metamorph version 6.1 (Universal Imaging Corp., Downingtown, PA).
Treatment With Cytochalasin B
Embryos at 50% epiboly (5.3 hpf) were incubated with either cytochalasin B (stock of 5 mg/ml prepared in DMSO) at 10–20 μg/ml in 30% Danieau's solution or with DMSO diluted 1:500–1:250 in 30% Danieau's solution (controls) at 28°C in the dark. These embryos were then fixed at 8.5 hpf, i.e., when the DMSO-control embryos had reached ∼85% epiboly, and labeled with rhodamine–phalloidin as described in the F-actin labeling protocol described above.
Fluorescent images were collected by using the Leica TCS SP2 laser scanning confocal unit described earlier. The percentage epiboly and other dimensions of the cytochalasin B–treated and control embryos were measured using the Leica confocal software (version 2.5, Build 1040). These data were then imported into Prism 3 (GraphPad Software, Inc., San Diego, CA) to calculate means ± SEM and plot graphs, as well as determine the degrees of significance by using the Student's t-test for unpaired samples.
Microinjection of BAPTA-Type Buffer Solutions
The following BAPTA-type calcium buffers were used: DBB (Kd of 1.5 μM) and DMB (Kd of 0.15 μM). DBB and DMB were prepared as stock solutions of 200 mM in double-distilled water containing 42 mM CaCl2 and 5 mM Hepes, set at pH 7.0 with KOH. These were then mixed with 400 μM rhodamine-labeled dextran (Rh–dextran; 70 kDa), to give a final concentration of 150 mM. The Rh–dextran, used to confirm the correct localization of the spreading injectate, was visualized by using a Zeiss Axiovert 135 fluorescence microscope with a Zeiss Plan Neofluar 20×/0.5 NA objective and 540- to 560-nm excitation/580-nm emission.
Dechorionated embryos were maintained at 28°C until early shield stage (∼6 hpf) at which time they were injected with 2 nl of DBB/Rh–dextran or DMB/Rh–dextran through the vegetal pole into the center of the yolk cell near the base of the blastoderm. As the water percentage of early zebrafish embryos has been reported to be ∼68% (Leung et al., 1998), the final buffer concentration could be carefully adjusted such that microinjection of 150 mM calcium buffer would produce a final concentration of ∼5 mM within the embryo.
Injected embryos were allowed to develop for a further 2 hr (i.e., until 8 hpf) at 28°C before being fixed with 4% paraformaldehyde and the F-actin labeled with FITC–phalloidin. The effects of the buffer injection on the actin cytoskeleton were examined by using the Bio-Rad confocal microscope system described earlier. Images were imported into Metamorph version 6.1 for quantification of the actin structures, after which the data were imported into Prism 3 for analysis as described above.
Imaging of Membrane Surface Foldings Using Nomarski Optics
Dechorionated embryos were maintained at 28°C in 30% Danieau's solution throughout the cleavage and blastula stages. At ∼75% epiboly (8 hpf), embryos were placed in our custom-made holding/viewing chambers (described in detail in Webb et al., 1997). Membrane surface folds were visualized by using a Zeiss Axiovert 135 microscope with a Zeiss Plan Neofluar 100×/1.3NA oil objective. Thirty-five millimeter photomicrographs were taken by using a Zeiss MC-80 photographic unit attached to the microscope. Images were then scanned into Corel PHOTO-PAINT 11 for subsequent enhancement with the “Contrast Enhancement” and “Contour – Find Edges” functions.
Measuring the Area of the Yolk Cell Membrane
The area of the yolk cell membrane (YCM) was measured in embryos at sphere stage, i.e., just before the onset of epiboly, and again at shield stage. To simplify the calculations, embryos at both stages were assumed to be perfect spheres. Thus, the following calculation was used: area of YCM = total surface area of embryo minus surface area of blastoderm. In purely spherical terms, this is equivalent to the following: surface area = total surface area of sphere minus surface area of a spherical cap. Thus, YCM area = 4πr2 – 2πrh, where r = radius of embryo and h = vertical height of the blastoderm from animal pole to the blastoderm/yolk cell margin. The equations for calculating the surface area of a sphere and spherical cap were from Zwillinger (2003).
The zebrafish were supplied by the Zebrafish International Resource Center, supported by grant RR12546 from the NIH-NCRR.