Furrow microtubules and localized exocytosis in cleaving Xenopus laevis embryos

It is now well established that localized exocytosis occurs at or near the cytoplasmic bridge of dividing animal cells (reviewed by Straight and Field, 2000; Glotzer, 2001). That exocytosis is actually required to complete membrane separation at the end of cytokinesis is suggested in experiments with embryos of sea urchins (Conner and Wessel, 1999) and C. elegans (Jantsch-Plunger and Glotzer, 1999; Skop et al., 2001). In numerous organisms, this membrane insertion may have been amplified to produce the basolateral membrane domain contemporaneously with cleavage; for example, during cellularization in Drosophila (Sanders, 1975; Foe et al., 1993; Loncar and Singer, 1995) and during cleavage stage in a variety of large-egg organisms, including Loligo (Cartwright and Arnold, 1981), Brachydanio (Jesuthasan, 1998), sturgeon and various amphibians (Zotin, 1964; Sanders and Singal, 1975).

Amphibian embryos are distinctive among dividing cells for the comparatively large amount of membrane that is added continuously during the cleavage process (Selman and Perry, 1970). For example, in Xenopus, as much as 1.4 mm² of new plasma membrane appears in the cleavage plane during ∼15 minutes of furrow progression (Bluemink and de Laat, 1973). This new membrane is known to differ qualitatively from that of the original egg surface (Kalt, 1971; Sanders and Singal, 1975; Byers and Armstrong, 1986; Servetnick et al., 1990; Bieliavsky et al., 1992; Aimar, 1997). The main source of the new membrane appears to be a pool of oogenetically produced vesicles (Leaf et al., 1990; Servetnick et al., 1990) that contributes not only membrane lipids, but also glycoproteins, and extracellular matrix components that ultimately line the surface of the blastocoel (Kalt, 1971; Servetnick et al., 1990).

How membrane expansion is regulated during cleavage is not understood. In particular, it is not yet clear where exocytosis actually takes place. It has been known since Zotin's pioneering studies in sturgeon and amphibian embryos (Zotin, 1964) that vesicles accumulate in the spindle midzone beneath the advancing first cleavage furrow, suggesting that membrane addition might occur at the leading edge of the furrow, contemporaneously with its advance through the midzone. In support of this idea, Sawai's tracing of carbon particle motions on the surface of cleaving newt eggs suggested that membrane expansion occurs from the furrow base (Sawai, 1987). Alternatively, the persistence of a stable original-membrane domain at the leading edge of the furrow, including membrane glycoproteins (Byers and Armstrong, 1986) and collections of microvilli (Denis-Donini et al., 1976), has suggested that membrane addition occurs elsewhere along the cleavage plane, for example, at the margin between new and old domains or along its entire length. This idea has support from electron microscopic studies that show a variety of putative exocytotic vesicles at various sites along the cleavage plane, but not necessarily near its leading edge (Bieliavsky and Geuskens, 1990; Singal and Sanders, 1974).

In Xenopus, vesicles contributing basolaterally targeted U-cadherin to each early cleavage plane are distributed uniformly in the peripheral cytoplasm until cleavage (Angres et al., 1991). Thus, however it occurs, vesicle recruitment to the site of exocytosis must be locally regulated in the vicinity of each furrow. The possibility that microtubules might be involved in localized basolateral vesicle recruitment to the cleavage furrow was raised when Danilchik et al. (Danilchik et al., 1998) and Jesuthasan (Jesuthasan, 1998) identified novel microtubule-containing structures at the bases of cleavage furrows in Xenopus and Brachydanio, respectively. These structures are now referred to as furrow microtubule arrays (FMAs) to distinguish them from other microtubule-containing structures in the cleavage plane (e.g. midbodies and interzonal microtubules). Microtubule-disrupting experiments indicated that basolateral membrane growth in the cleavage plane requires microtubules (Danilchik et al., 1998; Jesuthasan, 1998). Similarly, Larkin and Danilchik found that furrow microtubules are required in cleaving sea urchin embryos to complete the closure of the cytoplasmic bridge, and proposed that one function of furrow microtubules is to direct vesicles toward a site of fusion at the cytoplasmic bridge to accomplish cell separation (Larkin and Danilchik, 1999). More recently, Skop et al. (Skop et al., 2001) and Shuster and Burgess (Shuster and Burgess, 2002) have identified microtubule-dependent membrane-addition events in the final stages of cytokinesis in C. elegans and sea urchin egg cleavage. This apparently general requirement for microtubules in the localization of membrane addition to terminate cytokinesis seems to reflect a direct recruiting of vesicles to the furrow base.

In the present report, scanning electron microscopy (SEM) and live-embryo confocal imaging revealed clusters of exocytotic fusion pores in the immediate vicinity of the expanding furrow base. Nocodazole treatment both randomized and reduced the number of these pores, indicating that microtubules are indeed required for a step in localizing vesicle exocytosis. To test whether microtubules are sufficient to direct this localization process, D₂O was used to generate ectopic microtubule monasters. D₂O-treated embryos underwent a rapid, nearly uniform expansion of the surface; SEM analysis indicated large numbers of exocytotic fusion pores randomly scattered across the entire surface. This effect of D₂O was abolished with nocodazole, confirming that microtubules near the surface, near the end of the cell cycle are sufficient to recruit exocytotic vesicles to the plasma membrane, and supporting the hypothesis that furrow microtubule bundles play a similar role in the furrows of dividing cells.

Adult Xenopus laevis females were induced to ovulate with human chorionic gonadotropin (Sigma; 800 units per frog) injected into the dorsal lymph sac 18 hours prior to use (16°C). Ovulated eggs were fertilized by mixing with fragments of testis macerated in MMR/3 (MMR: 100 mM NaCl, 1.8 mM KCl, 2.0 mM CaCl₂, 1.0 mM MgCl₂, 5 mM Hepes, pH 7.5). Eggs were dejellied in 2.5% cysteine in MMR/3 at pH 8.0. Eggs were then washed free of cysteine and cultured to appropriate stages in MMR/3. For many experiments, embryos were devitellinated manually using watchmakers' forceps. This procedure was best done between 45 and 60 minutes post-fertilization to avoid mechanical rupture. Devitellinated embryos were cultured in dishes coated with 2% agarose (Type V, Sigma) to prevent membrane adherence to the substrate. In experiments involving exposure to D₂O (see below), agarose was pre-equilibrated with similar concentrations of D₂O.

Stock solutions of nocodazole, cytochalasin B, cytochalasin D and latrunculin B (Sigma) in dimethylsulfoxide (DMSO) were diluted to concentrations of 5 or 10 μg/ml in MMR/3 and applied to embryos at specified times. Controls consisted of similar dilutions of DMSO in MMR/3. D₂O was diluted to 60% in MMR/3 and applied to embryos at specified times during early development.

For examining embryos via wholemount confocal immunocytochemistry, we adapted Gard's (Gard, 1993) protocol, as described previously (Danilchik et al., 1998). Embryos were fixed for 2-4 hours at room temperature or overnight at 4°C in Gard's fixative (3.7% formaldehyde, 0.25% glutaraldehyde, 0.2% Triton X-100, 80 mM PIPES, pH 6.8, 5 mM EGTA, 1 mM MgCl₂). After fixation, embryos were stored in methanol overnight at -20°C. Pigmented embryos were bleached in 10% H₂O₂ in methanol on a light table for 1-4 hours (bleaching time varied, depending on rate of progress of pigment fading). After bleaching, embryos were rehydrated in three consecutive rinses, for 10 minutes each, of: 50% MeOH in TBS, 25% MeOH in TBS, 100% TBS (1× TBS: 155 mM NaCl, 20 mM Tris-Cl, pH 7.4). To decrease background fluorescence caused by glutaraldehyde and autofluorescence of yolk platelets, embryos were placed in a reducing solution of 100 mM NaBH₄ in TBS for 4 hours (at room temperature) or overnight (at 4°C). NaBH₄ was removed by rinsing in NTBS five times over the course of 1 hour (NTBS: 1× TBS, with 0.1% NP-40).

Vitelline envelopes were removed prior to exposing fixed specimens to antibody. Some embryos were bisected with a razor blade fragment, cutting parallel to the animal-vegetal axis, either perpendicular to or paralleling the cleavage plane. Other embryos were processed intact for wholemount observation. Intact or bisected embryos were incubated with antibodies diluted in TBS with 10% fetal bovine serum and 5% DMSO overnight at 4°C, with agitation, with five one-hour washes in NTBS after each incubation. Primary antibodies included monoclonals against αβ-tubulin (Biogenesis; 1:1000), γ-tubulin (Sigma; 1:200), VSVG (clone P5D4; Sigma; 1:400) and sheep polyclonal antibodies against αβ-tubulin (Cytoskeleton; 1:200). Secondary antibodies were Alexa-546- or -488-conjugated goat anti-mouse or donkey anti-sheep (Molecular Probes, 1:100). After immunostaining and subsequent washes in NTBS, embryos were dehydrated in two rinses of MeOH, 30-60 minutes each, and then cleared in two rinses of Murray's clear (2:1::benzyl benzoate:benzyl alcohol).

The growth of new membrane during cell division was visualized in devitellinated embryos exposed to fluorescent soybean agglutinin for five minutes before first cleavage (Alexa 488 conjugate; 125 μg/ml; Molecular Probes). Unbound lectin was removed via several exchanges of MMR/3 prior to time-lapse recording.

To view membrane expansion, a suspension of activated charcoal particles was pipeted over devitellinated embryos prior to time-lapse recording. Time-lapse sequences were recorded using an NEC color CCD camera mounted on an Olympus SZH stereoscope. Images were captured on a Panasonic LQ-3031 optical disk recorder, digitized via an Xclaim VR 128 graphics card (ATI), and converted to TIFF image stacks for further analysis using Wayne Rasband's NIH Image (v. 1.62; http://rsb.info.nih.gov/nih-image), or to QuickTime (Apple) movies using Quicktime Pro or Adobe Premiere 5.1.

To display carbon particle motion as a kymograph, TIFF time-lapse image stacks were opened in NIH Image, assigned a 1-pixel virtual spacing between slices, and selected regions were then rotated 90° about the Y axis to display elapsed time along the X-axis as 1 pixel per captured frame.

Exocytotic fusion pore distributions in the cleavage furrow were measured in panoramic montages of digitized SEM images across relevant surfaces registered using Adobe PhotoShop. NIH Image's particle analysis package was then used to determine pore positions relative to the base of the furrow (leading edge). Pore counts in adjacent 10 μm × 50 μm regions of interest were determined and their local density was then plotted as a function of distance from the base of the furrow.

To examine membrane dynamics in vivo at high magnification, devitellinated embryos were bathed in MMR/3 containing 10 μM cytochalasin B and 10 μM FM1-43 (Molecular Probes) and examined via confocal microscopy (BioRad Radiance 2100; Nikon E800 upright microscope using a 60×/1.0 NA CFI₆₀ Fluor dipping objective). Focus on the surface of the animal hemisphere was readjusted frame by frame manually because the devitellinated embryos' height rises and falls rapidly throughout the cell cycle (Hara et al., 1980).

For scanning electron microscopy, intact or devitellinated embryos were fixed in 2.5% glutaraldehyde in 0.1 M Na-cacodylate, pH 7.5 overnight at 4°C. Fixed specimens were then transferred into 0.1 M Na-cacodylate, pH 7.5 at 4°C. If necessary, embryos were manually devitellinated at this point, and then dehydrated via 15-minute exchanges with 50, 75 and 95% ethanol, followed by three 30-minute exchanges with 100% ethanol. Specimens were rinsed twice with hexamethyldisilizane (Polysciences) for 30 minutes. Specimens were then allowed to dry overnight at room temperature in open vials. More rapid, vacuum-driven removal of the hexamethyldisilizane was found to be unsatisfactory, since it often resulted in tissue contraction or surface rupture. Dried specimens were mounted onto aluminum stubs using silver paste, sputtercoated with gold-palladium to 50 nm on a Hummer VII (Analect, USA), and then examined with a JEOL T330A scanning electron microscope.

In an earlier report (Danilchik et al., 1998), we suggested that furrow microtubules are involved in directing vesicles to a site of exocytosis in the growing cleavage plane. This hypothesis predicts that the site of membrane addition should be relatively near the furrow base, and that no significant membrane addition should take place elsewhere during furrow deepening. To investigate this idea, we manually removed the vitelline envelope prior to cleavage (Fig. 1A-D; Movie 1, see Supplementary Information), making the site of membrane addition accessible for direct observation or local manipulation via microinjection of function-blocking reagents. When FITC-conjugated soybean agglutinin (FITC-SBA) was bound to the egg surface, video time-lapse recording of the redistribution of label revealed the expansion of two broad, unlabelled surface zones on both sides of the furrow midline (Fig. 1E-H). At the same time, a thin line of original-surface components became concentrated at the furrow base, a phenomenon first noted by Byers and Armstrong (Byers and Armstrong, 1986). The lectin-binding in this region probably corresponds to the tips of microvilli that remain at the furrow base (Denis-Donini et al., 1976) (see also Fig. 4B,D). The original egg surface shows remarkable coherence, with little evidence of lateral mobility of surface components, and virtually no mixing with the unlabeled domain. Evidently little or no unlabeled membrane appears in the original surface. A variety of experiments (see below) using devitellinated embryos demonstrates that the major site of membrane insertion is within 50 μm of the furrow base, and that its appearance is microtubule dependent, consistent with a role for furrow microtubules in localizing exocytosis.

As was earlier shown (Bluemink and de Laat, 1973; Drechsel et al., 1997; Danilchik et al., 1998), localized membrane addition continues unabated after disruption of the contractile ring by treatment with cytochalasin B or function-blocked rho protein, with the result that a broad stripe of unpigmented surface develops at the presumptive cleavage plane. Here, we show similar results with another microfilament inhibitor, latrunculin B (Fig. 2A). The same embryo fixed, processed, and stained to reveal microtubules via confocal wholemount microscopy displays prominent bundles of microtubules in close proximity to the surface (i.e. a few microns) along the entire length of the disrupted furrow (Fig. 2B). Similar results were obtained with both cytochalasin B and D.

We used a method developed previously (Sawai, 1987) to follow the movement of small carbon particles dropped onto the surface of devitellinated, cleaving Xenopus embryos treated with cytochalasin B to expose the growing membrane domain. Video time-lapse recordings of particle motions (Fig. 3A-D) were reprocessed to produce kymographs displaying particle movement along the ordinate and elapsed time on the abscissa (Fig. 3E). Particle drift away from the furrow base was relatively steady, averaging 35 μm/minute. Particles landing near each other on the same side of the furrow traveled at nearly the same rate without drifting apart, indicating that little or no membrane expansion took place between them once they had moved beyond the furrow base. In contrast, particles landing on opposite sides of the furrow base traveled away from each other. Some particles, landing directly over the furrow base, remained there for several minutes before abruptly commencing to drift toward one side or the other. These results indicate that most of the membrane expansion occurs from a site or sites within ∼50 μm of the furrow base. Similar results (not shown) were obtained in embryos not treated with cytochalasin.

Scanning electron microscopy was used to examine the new surface of devitellinated, cleaving embryos for evidence of exocytosis. At low magnification (Fig. 4A,C), the new membrane domains appeared as a pair of relatively smooth triangular areas on either side of the cleavage furrow (Fig. 4A). At higher magnification, surveys revealed large numbers of circular pits or craters in the otherwise smooth new-membrane domain (Fig. 4B,D). Pits ranged in size from ∼0.5 to 2.0 μm in diameter. Two irregular stripes, roughly 10 μm wide, at either side of the furrow base itself were nearly devoid of pits (Fig. 4B,D). Similar analysis of embryos fixed at different times during first cleavage indicated that generally more than 75% of these pits were concentrated within 50 μm of the furrow base during the membrane expansion phase. Few pits were found near the margin between new and old membrane domains. For example, Fig. 5 (filled circles) plots pit density as a function of distance from the furrow base. In embryos treated with nocodazole during furrow formation, both the number and concentration of surface pits were significantly reduced (Fig. 5, open circles).

The above SEM observations are consistent with the idea that the pits represent fusion pores at sites of recent or ongoing vesicle exocytosis. As an independent test of this hypothesis, we searched for evidence of exocytosis in live embryos in which the plasma membrane was loaded with the fluorescent styryl dye FM1-43. Although FM1-43 is most commonly used in pulse-chase experiments to selectively label recycling exocytotic vesicles (Betz et al., 1996), it has a relatively high affinity for new membrane in the cleavage plane of Xenopus embryos (see below), making it useful for identifying sites of ongoing exocytosis. Devitellinated embryos undergoing first cleavage were incubated continuously in medium containing both cytochalasin B and FM1-43. As with latrunculin (Fig. 3), the cytochalasin disrupts furrow deepening, thereby making the membrane addition site accessible for viewing by confocal microscopy (Fig. 6). FM1-43 gradually accumulated in the plasma membrane, preferentially labeling the new membrane domain (Fig. 6, asterisk indicates less-labeled original-membrane of the animal hemisphere surface). Although some endocytosis has been detected in the new membrane domain of Xenopus (data not shown), consistent with recent work in zebrafish (Feng et al., 2002), the intense labeling along the margins between new and old membrane domains (Fig. 6) primarily reflects the large amount of membrane involved with microvilli and other protrusions in this region (Denis-Donini et al., 1976).

As embryos entered second cleavage, a new site of membrane addition appeared near the disrupted second furrow (arrow, Fig. 6). This expanding area was significantly less fluorescent than other regions within the new membrane domain, suggesting ongoing localized addition of unlabeled membrane. Time-lapse confocal recordings at high magnification in this region confirmed this idea. As an example, Movie 2 displays a particularly active site recorded near the arrow in Fig. 6. This movie segment shows the abrupt appearance of irregular dark patches, 10-20 μm² in area, which quickly diffused into the surrounding FM1-43-labeled membrane. Still frames representing three successive sequences from this time-lapse recording are presented in Fig. 7. In the sequence shown in panels A1-A4, the arrows indicate a single site of membrane expansion in which a patch of unlabeled membrane abruptly grew from ∼ 10 to 25 μm², before fading into the surrounding labeled surface. Similar single-fusion events are shown (Fig. 7B1-B4).

In some cases, circular structures, apparently stable exocytotic fusion pores, were seen to associate with the membrane for several minutes before abruptly flattening into the plane of the membrane and contributing unlabeled patches to it. For example, in the sequence shown in Fig. 7C1-C4, a 2.5 μm circular fluorescent structure (the same structure had also been present through the previous two sequences) suddenly disappeared to be replaced by an irregular patch of unlabeled surface (cf. arrows in Fig. 7C2-C3). The sudden introduction of the unlabeled patch evidently contributes surface area to the local membrane: by looping the movie sequence back and forth a few frames across this event, small surface particles can be seen to spread a short distance in all directions away from the patch.

A densitometric profile across the structure indicates that the membrane within the intensely labeled edges contains significantly less label than the surrounding plasma membrane (not shown). The brightly labeled edge evidently constitutes a barrier to lateral diffusion between the membrane of the vesicle and that of the cell surface until the abrupt flattening of the vesicle into the plane of the membrane.

To learn the size range of vesicles destined for exocytosis along the basolateral surfaces, we injected fertilized eggs with capped, synthetic mRNA encoding the full-length sequence of the basolaterally targeted viral coat glycoprotein of vesicular stomatitis virus (VSVG; plasmid gift of H.-P. Moore, Berkeley, CA) and fixed and processed them for wholemount immunocytochemistry using an anti-VSVG antibody (Sigma). Fig. 8 is a projection of a confocal image stack showing a collection of VSVG-labeled vesicles near the base of a cleavage furrow. The surface is also labeled, indicating recent exocytosis of similarly labeled vesicles at or near this site. Vesicles ranged in diameter from 0.6 μm to ∼2.0 μm, which is similar to the size of the presumed exocytotic pits shown above via SEM.

We reported previously that embryos exposed to high concentrations of D₂O form numerous ectopic monasters (Danilchik et al., 1998). The concomitant disturbance in cortical pigmentation had suggested new membrane addition, consistent with the idea that the presence of polarized microtubules near the surface is sufficient to provoke new membrane addition. To examine this possibility further, we repeated our earlier experiment, this time removing the vitelline envelope prior to incubation in D₂O. The experiment shown in Fig. 9 confirms that D₂O provokes a rapid, essentially random expansion of the embryo surface. Without the support of a vitelline envelope, embryos normally flatten slightly onto the substratum and assume an oblately spheroidal morphology (Bluemink and de Laat, 1973). In contrast, in the presence of D₂O, the surface of the embryos rapidly expanded at about the time of first cleavage (Fig. 9A,B). The overall spreading and thinning of the pigmented animal hemisphere suggests that the new surface area was introduced randomly across the entire surface. Scanning EM analysis (Fig. 10) confirms that exocytotic fusion pores appeared in large numbers in the smooth surface regions between clusters of microvilli (Fig. 10B), consistent with the idea that D₂O completely randomizes the site of membrane addition. Because microvilli were retained in large numbers throughout the D₂O-driven surface expansion, it is unlikely that the observed increase in surface area is simply due to microvillar shortening.

The D₂O-induced membrane expansion appears to be entirely microtubule-dependent, as indicated by the inhibition of expansion by 10 μM nocodazole (Fig. 11). To determine whether the nocodazole treatment actually disrupts microtubule formation in D₂O-treated embryos, control and treated embryos were fixed at 105 minutes post-fertilization, and examined via confocal microscopy for the presence of microtubules (Fig. 12). In an untreated embryo, aligned microtubule bundles of the FMA decorated the base of the first cleavage furrow (Fig. 12A). Elsewhere on the animal hemisphere of the same embryo, only a few microtubules were found near the surface (Fig. 12B). Vertically resectioning the same confocal stack (Fig. 12B, inset) revealed the normally low concentration of microtubules at the surface. In contrast, following treatment with D₂O, numerous microtubule monasters were found near the cell surface (Fig. 12C and vertical resection, inset). Nocodazole completely abolished microtubules in an embryo incubated in normal medium (Fig. 12D and inset), and effectively depolymerized microtubules near the surface of D₂O-treated embryos, even though monasters apparently persisted more deeply in the cytoplasm (Fig. 12E and inset). These results are consistent with a requirement for intact microtubules in membrane expansion; with the results of Figs 9,10,11 they indicate that the presence of microtubules near the surface is sufficient to induce membrane expansion.

D₂O-treated embryos were double-stained for both β- and γ-tubulin and examined via confocal microscopy. Monastral microtubules were seen to extend in all directions from γ-tubulin-rich centers (Fig. 13). Since γ-tubulin is a centrosomal protein that associates with the minus ends of microtubules (Oakley et al., 1990; Stearns et al., 1991; Stearns and Kirschner, 1994; Li and Joshi, 1995), we conclude that most of the microtubules interacting with the cortex in D₂O-treated embryos are oriented with plus ends facing the surface.

During the first cleavage in Xenopus, over 1.4 mm² of new plasma membrane is inserted along the advancing cleavage furrow (Bluemink and de Laat, 1973). A variety of maternally synthesized membrane proteins, including U-cadherin (Angres et al., 1991), β-integrin (Gawantka et al., 1992) and other basolaterally targeted membrane proteins (Servetnick et al., 1990), appear in this new membrane domain during or shortly after membrane deposition. It is generally assumed that the surface area increase derives from the exocytosis of post-Golgi vesicles bearing these maternal proteins. However, the time course of the appearance of specific membrane proteins in the basolateral domain may not always be contemporaneous with the bulk of membrane growth itself. In Xenopus, Fesenko et al. (Fesenko et al., 2000) demonstrated that three maternally synthesized tight-junction proteins display three distinct different temporal expression patterns that are independent of the membrane growth itself. Also, embryonically synthesized VSVG protein appears in basolateral membranes of nocodazole-blocked Xenopus embryos, indicating that vesicle addition can take place passively along surfaces that are no longer growing (Roberts et al., 1992). Thus, it is clear that the exocytosis of basolateral determinants is not synonymous with the bulk expansion of membrane as the cleavage furrow advances, and it is therefore important to study membrane growth as an independent process.

The present report addressed two related issues in basolateral membrane formation during Xenopus cleavage: the site of membrane insertion along the furrow, and the potential role of furrow microtubules in regulating this site. We found that the bulk of membrane addition proceeds via exocytosis from a site within ∼50 μm of the furrow base. This location is significant, since it overlies the ends of microtubules of the furrow microtubule array, and is therefore consistent with the idea that the microtubules mobilize the exocytotic vesicles involved in the membrane expansion. Also consistent with this hypothesis was the finding that ectopically placed microtubule plus ends near the cell surface in D₂O-treated embryos are sufficient to provoke ectopic new membrane addition. Therefore, we suggest that vesicles encountering furrow microtubules at the advancing furrow tip are transported toward the microtubule plus ends to become concentrated at either side of the base of the furrow.

Exocytotic fusion pores are transitory aqueous channels connecting the lumen of vesicles with the extracellular space during the process of exocytosis. As originally defined, pores are small structures, 20 to 100 nm in diameter (Breckenridge and Almers, 1987). We were initially inclined to dismiss the 2 μm pits seen via scanning EM as fixation artefacts. However, their nonrandom distribution, relatively uniform size range, and absence under conditions preventing new membrane addition (e.g. nocodazole) suggested that they might represent authentic exocytotic fusion pores. Similar structures in the 2 μm diameter range, observed via SEM in secretory alveolar type II cells, are accepted as authentic fusion pores, with channels remaining open for up to several hours (Haller et al., 2001). Similarly, large, stable cortical crypts or pits remain in the surface for several minutes following cortical granule exocytosis in zebrafish (Becker and Hart, 1999).

By independent methods, we visualized 2 μm hollow cavities in the growing surface of living embryos that resemble those seen in SEM. Time-lapse confocal microscopy of embryos stained with FM1-43 gave us a glimpse of the dynamic, transitory nature of these exocytotic fusion pores. Some apparently stable fusion pores remained at the surface for up to several minutes before flattening into the plane of the cell surface as a patch of unlabeled membrane. The area of each patch was approximately 10-20 μm², an amount of surface that could have been provided by individual vesicles ranging in diameter from ∼2.2 to 2.8 μm. We asked whether the density of fusion pores seen in cleavage furrows could provide a sufficient number of exocytotic events to account for the large amount of membrane introduced during cleavage. Bluemink and de Laat (Bluemink and de Laat, 1973) estimated that ∼2.3 mm² of new membrane is introduced ectopically to the surface following cytochalasin B treatment. Our embryos were smaller than those used by Bluemink and de Laat (1.16 mm diameter); we estimate the total new membrane in our cytochalasin-treated embryos (Fig. 6) to be ∼1.85 mm². Assuming that each fusion pore represents a potential delivery of 20 μm², as suggested from the FM1-43 labeling experiment (Fig. 7C), then ∼92,500 vesicles would be needed. Since membrane addition is continuous over a period lasting ∼15 minutes, only a limited number of fusion pores should be seen at any time in fixed SEM specimens. The profile of fusion pore density in Fig. 5 shows 27±3 fusion pores per adjacent 500 μm² regions nearest the furrow base. Again, if each pore represents 20 μm² of potential surface area, this density of exocytotic vesicles could represent ∼99,000 vesicles. In other words, the fusion pore clusters near the furrow base are of a density consistent with that capable of providing the entire 1.85 mm².

D₂O-stabilized microtubules appear to be fully capable of supporting organelle transport in Xenopus embryos, in a reportedly `randomized' fashion (Rowning et al., 1997). However, as we have shown here, D₂O-stabilized microtubules are evidently not randomized; instead, they form well-organized, ectopic monasters (see also Danilchik et al., 1998). The presence of centriole-like structures (van Assel and Brachet, 1966) and, as shown in this report, γ-tubulin at the foci of these monasters, indicate that the D₂O-stabilized monasters resemble conventional MTOCs, with microtubule plus ends radiating away from organizing centers. In effect, the entire cortex becomes enriched with overlapping arrays of microtubule plus ends, a situation that normally only occurs along the cleavage plane. Since the effect of D₂O on membrane expansion is blocked with nocodazole, we conclude that the exocytosis of vesicles depends on a plus-ward transport of vesicles. This conclusion is thus consistent with that of experiments demonstrating kinesin-dependence in the microtubule-based mobilization of vesicles toward sites of wound-induced exocytosis (Bi et al., 1997).

We still lack direct information about the polarity of microtubules in the FMA. From the SEM data shown here, we know that the site of membrane addition is just to either side of the furrow base, making it unlikely that the FMA transports vesicles toward the midline. Rather, the observed sites of exocytosis are consistent with vesicles being transported away from the midline. However, this leads to a surprising result: if exocytotic vesicles are indeed plus-directed, as suggested by the D₂O experiments, then the FMA bundles should be oriented with minus ends toward the midline. Such a microtubular arrangement is clearly in conflict with the known organization of plant-cell phragmoplasts (Staehelin and Hepler, 1996) and animal-cell midbodies (Eutenauer and McIntosh, 1980). To resolve this issue, we are presently investigating the polarity of FMA microtubules by following the motion of GFP-tagged EB1 in cleaving Xenopus embryos (E.E.B. and M.V.D., unpublished).

We thank Jerry Adey and Kay Larkin for help with the scanning electron microscopy. This work was supported by the National Science Foundation (IBN-0110985 and DBI-0070391) and by the National Aeronautics and Space Administration (NAG2-1199).