Proc. Natl. Acad. Sci. USA Vol. 92, pp. 2199-2203, March 1995 Developmental Biology

Cell membrane formation during the cellularization of the syncytial blastoderm ofDrosophila (embryonic development/ formation/electron microscopy)

DRAGUTIN LONCAR* AND S. J. SINGERt Department of Biology, University of California, San Diego, La Jolla, CA 92093-0322 Contributed by S. J. Singer, December 2, 1994

ABSTRACT The early blastoderm of Drosophila is a syn- to permit entry of a water-soluble fixative into the space cytium in which about 6000 nuclei become localized in the between the vitelline and plasma membranes. Permeation of peripheral cytoplasm. During cycle 14 interphase, a wave of the fixative through the plasma membrane then occurred, but membrane formation encircles each nucleus inside its own the resultant fixation of the cytoplasmic contents was not plasma membrane, thereby generating an intact epithelial reproducibly adequate in our hands. In a small but useful layer. The details of this process of cellularization have been addition to this procedure, we have microinjected fixative unclear. Using an improved method offixation ofthe embryos through the puncture. The original periplasmic contents of the for electron microscopy, we show by morphological observa- space between the punctured vitelline and intact plasma tions that a large number of membrane-bounded, electron- membranes were thus rapidly and uniformly displaced by transparent vesicles, ofdiameters ranging from 0.05 ,im to 0.5 fixative, which, after permeation of the plasma membrane, ,um, are present in the periplasm and become redistributed resulted in rapid and uniform fixation of the cytoplasmic during cellularization so as to provide the membrane mass contents. By virtue of this procedure, we believe that we have required at each phase of the process. We recognize three been able to visualize ultrastructural details of the cellular- phases. In the first two phases, the vesicles that were present ization process not previously described. in the apical periplasmic space at earlier stages become Our morphological results strongly suggest that the source concentrated and aligned between the nuclei. The vesicles then of the membranes formed throughout cellularization consists undergo concerted but not precisely synchronous fusion to predominantly of small intraembryonic membrane-bounded, form double membranes, starting at furrows in the plasma electron-transparent vesicles, which are repositioned at several membrane of the embryo and extending about 7 ,um into the stages in the process to become fused with the forming mem- periplasmic space. Subsequently, in the third phase vesicles branes. Several consecutive phases in the continuous process are recruited to the basal periplasmic space but do not become of cellularization were observed and are described. aligned between the nuclei as in the first phase. We presume that these vesicles fuse individually with the growing ends of the double membranes until encirclement of each nucleus is MATERIALS AND METHODS complete. We speculate that these vesicles are all derived from Drosophila embryos.Drosophila eggs were provided through the Golgi apparatus and are moved about in the blastoderm the kindness of James Kadonaga, employing the detailed by interactions with components of the cytoskeleton. conditions and procedures described elsewhere (5). In brief, collecting plates were kept in collecting chambers at 25°C for The development of the early embryo of insects such as a half hour and then removed from the chambers but left at Drosophila is unusual in that the zygote undergoes a series of 25°C. The embryos were then sampled every half hour for 4 hr. 14 nuclear divisions before any cellularization arises (except They were then dechorionated in 50% sodium hypochlorite for the pole cells). The first 9 nuclear divisions occur rapidly and fixed (see below), and specimens were prepared for elec- in the interior of the embryo, after which most of the nuclei tron microscopy. This procedure resulted in a collection of migrate to the periphery ofthe embryo where they divide more progressive developmental stages. Within the time interval slowly another 5 times. The syncytial blastoderm then has after mitotic division 14 (2.5-3.5 hr), more precise ordering of about 6000 somatic nuclei organized in the peripheral peri- the developmental sequence was based on morphological plasm underneath the plasma membrane. At this stage, during criteria such as nuclear positions and shapes and the state of the interphase of cycle 14, a synchronized wave of membrane membrane furrow formation as observed in the electron formation appears to descend into the periplasmic space, micrographs (3, 6). initiating at a hexagonal cytoskeletal cage surrounding each Fixation and Electron Microscopy. After dechorionation, nucleus underneath furrows in the plasma membrane, and Drosophila eggs and embryos were immobilized on double- eventually encircling and compartmentalizing each nucleus. sided tape and covered by a drop of fixative (mixture of 2% of This membrane encirclement produces the cellular blasto- glutaraldehyde and 2% paraformaldehyde in 0.15 M NaCl/0.01 derm. (For reviews, see refs. 1 and 2.) M phosphate, pH 7.4). The viteiline membrane was gently The process of membrane formation during cellulariza- punctured in the fluid-filled pockets (3, 4, 7) in the pole of the tion-e.g., the source of the membrane components, and the eggs by a fine micropipette, and then a small amount of fixative mechanisms underlying membrane formation-is not well ("10 ,u) was injected into the pocket between the vitelline and understood. In large part, this stems from the fact that plasma membranes. This process replaced the fluid in the space ultrastructural studies of the process by electron microscopy between the vitelline and plasma membranes with fixative, the have been impeded by technical problems. In several previous excess fluids flowing out past the micropipette puncture. Eggs studies (e.g., refs. 3 and 4) the impermeable vitelline layer of and embryos with such perforated vitelline membranes but intact the dechorionated embryo was punctured once by a fine needle plasma membranes were kept in fixative for another 5 hr. During

The publication costs of this article were defrayed in part by page charge *Present address: Department of Biology, School of Medicine, payment. This article must therefore be hereby marked "advertisement" in University of California, Davis, CA 95616. accordance with 18 U.S.C. §1734 solely to indicate this fact. tTo whom reprint requests should be addressed. 2199 Downloaded by guest on September 26, 2021 2200 Developmental Biology: Loncar and Singer Proc Natl. Acad Sci. USA 92 (1995) that time, both the vitelline and plasma membranes became rigid plasma membrane to a distance of about 7 ,tm into the peri- and fixed. Manually devitellinized eggs and embryos were then plasm at this time. Each forming double membrane is still kept in fixative overnight, washed in 0.1 M phosphate buffer (pH connected at its basal end by an enlarged furrow canal situated 7.4) and then postfixed in S04 (1% solution in 0.1 M phosphate about halfway down the length of the nuclei. Transparent buffer) for 1 hr. They were then washed in phosphate buffer, vesicles are again accumulated in the apical periplasm above dehydrated through ascending concentrations of alcohol and the nuclei (Fig. 1 C), occupying about 19% of the volume of propylene oxide, and embedded in Epon by a standard proce- that space (Table 1). No significant accumulations of vesicles dure. Epon-embedded material was cut in sections of 70-nm are observed between the nuclei or just beyond the furrow thickness, contrasted with uranyl acetate and lead citrate, and canals. The basal periplasm exhibits only a few vesicles at this examined in a JEOL 1200 EX electron microscope at 80 keV. phase. Morphometry. The volume density (%) of vesicles was mea- (iii) Phase III (3-3.5 hr). The space between neighboring sured in the periplasm of zygotes and blastoderms by the nuclei now contains a fully formed double membrane (Fig. procedures developed by Weibel and coworkers (8, 9). The 1D), extending about 10 ,tm from the plasma membrane and photographs were enlarged to a final magnification of x 4800. ending in a looped-out furrow canal. The apical periplasm is A multipurpose test system after Weibel (M 168) was modified now largely depleted of the transparent vesicles seen in Fig. and used for morphometric estimation. Volumetric density of 1 C, and the basal periplasm shows a marked accumulation of vesicles (%) in Table 1 refers to the density of vesicles in the such vesicles (Fig. 1D; Table 1). The double membranes periplasm of the zygotes and blastoderms excluding nuclei. continue to elongate, extending to distances of 20-25 ,tm from the plasma membrane (Fig. 1E). The furrow canals then RESULTS become enlarged roughly perpendicular to the internuclear double membranes, enveloping each nucleus. Eventual fusion Electron Microscopic Ultrastructural Preservation. The of two extending furrow canals occurs (cells labeled BC in Fig. method of fixation described in Materials and Methods that was 1E) to complete the process of cellularization. During this used in this study has successfully preserved ultrastructural time, large numbers of transparent vesicles occupy about 21% detail and substructure in the embryonic cytoplasm, including of the volume of the basal periplasm of the forming cells (Fig. mitochondria, nuclei, yolk granules, vesicles, and membranes 1E; Table 1). There are now very few vesicles in the apical (Figs. 1 and 2). These results validate the adequacy of our cytoplasm. fixation procedure. Embryos at the next stage of development, around 3.5-4 hr, Vesicles and Membrane Formation During Cellularization. enter and develop several cell layers (Fig. 1F). Overview oftheprocess at low magnification. We first survey the Transparent vesicles are now almost entirely absent from the results at low magnification in Fig. 1 and then examine certain outermost layer of cells but are seen in the cells of the features in more detail at higher magnification in Fig. 2. At an underlying layer. early stage in embryonic development prior to nuclear migra- Details at higher magnification. Membrane-bounded vesicles tion to the periphery (Fig. 1A), the peripheral cytoplasm ofvarious diameters line up during phase I (Fig. 2A and B) and contains many large yolk granules (Y) and a large number of undergo fusion (Fig. 2 C-F) to form a double membrane dur- small electron-transparent membrane-bound vesicles that are ing phase II. The process of vesicle fusion appears to be dispersed in an unorganized manner. The vesicles occupy concerted but not precisely synchronous. The double mem- about 12% of the periplasmic volume by morphometric anal- brane not only grows and becomes more extended with time ysis (Table 1) and have a range of diameters from 0.05 ,um to but appears to undergo a maturation process. Shortly after 0.5 ,um. Subsequent mitotic divisions of the nuclei and their formation, the two membranes exhibit separations that are not eventual movement from the central part of the blastoderm uniform along their lengths (Fig. 2F), but with time the two into the periplasmic space that occurred in successive stages membranes become uniformly closely juxtaposed along their are not shown. By the end of cycle 13 the periplasm increased entire lengths. to a distance of about 20 ,tm and contained numerous small Surface Membrane of the Blastoderm. At all stages prior to vesicles similar in size, number, and spatial distribution to the and during cellularization, the membrane surface of the de- vesicles in the early stage shown in Fig. 1A. The process of veloping blastoderm is relatively smooth and, in particular, cellularization occurred in the interphase of cycle 14 (stage 5 does not exhibit any significant villous formations (Fig. 1A-E). of ref. 6); although it was, of course, continuous, it is conve- The surface furrows are transient structures that form between nient to recognize several distinctive phases in the process. the nuclei and retract during different cycles of nuclear (i) Phase I (-2.5 hr). The nuclei, of nearly spherical shape, division in the periplasm. It is of interest that the lining up of have by now been translocated to the periphery, and the yolk the transparent vesicles between successive nuclei occurs at a granules have been moved out of the peripheral cytoplasm into time when there are no surface furrows (Fig. 1B; Fig. 2A and the interior, a distance of about 20 ,um. The vesicles in the B), which form only around the time that the vesicles are apical periplasm become redistributed and their numbers are undergoing fusion (Fig. 1 C; Fig. 2 C and D). increased in the space between the nuclei (Table 1), where they become lined up to a distance of about 7 ,um from the plasma membrane (Fig. 1B). No marked furrowing of the plasma DISCUSSION membrane is seen at this phase such as is observed shortly Cellularization of the syncytial blastoderm ofDrosophila mela- afterwards (Fig. 1 C). The vesicle concentration simulta- nogaster is a remarkable process involving a precisely staged neously markedly decreases in the apical periplasm directly and nearly synchronous growth of plasma membranes that above the nuclei and in the basal regions of the periplasm. The completely envelop the nearly 6000 somatic nuclei that had clear inference from these observations is that the vesicles previously been localized to the periphery of the embryo. In present randomly throughout the periplasm just prior to phase this highly organized series of events occurring during the I are somehow reorganized and recruited into a linear array interphase of cycle 14 and requiring -1 hr, an intact cellular situated between the nuclei at the beginning of cellularization. epithelium of the blastoderm is generated from the . (ii) Phase II (2.5-3 hr). By this phase, the nuclei were now The nature of this process of membrane formation has been elongated (Figs. 1 C-E), and uniformly positioned in the unclear. Our electron microscopic results demonstrate, how- periplasmic space. Between the nuclei, double membranes are ever, that these membranes form by the concerted and suc- in the process of forming by the fusions of the lined-up vesicles. cessive fusions of numerous preformed intraembryonic vesi- The partially formed membranes extend from furrows in the cles that are recruited to and within the appropriate regions of Downloaded by guest on September 26, 2021 Developmental Biology: Loncar and Singer Proc. Natl. Acad ScL USA 92 (1995) 2201

FIG. 1. Low-magnification electron micrographs near the surface (S) of the zygote (stage 1) (A), the syncytial blastoderm (at successive times during stage 5) (B-E), and the cellular blastoderm (stage 6) (F). (Bars = 2 ,um.) (A) Yolk granules (Y) of different electron transparency are randomly scattered throughout this region (as in the entire cytoplasm, not shown), except for a thin periplasmic rim (P) close to the surface. Numerous transparent small vesicles are randomly distributed throughout the cytoplasm. (B) Early in the interphase of nuclear cycle 14 (-2.5 hr of development), at the beginning of phase I (see text). The transparent vesicles (V) become concentrated and aligned between the spherical nuclei (N) for a distance of #7 ,um from the surface. The basal periplasm (B) exhibits only a few vesicles. (C) Onset of phase II (see text), between 2.5 and 3 hr of development. Membrane furrows (F) have formed between the now elongated nuclei (N), originating at newly formed indentations of the surface (S) and terminating in enlarged furrow canals (C). Numerous vesicles (V) are again accumulated in the apical periplasm, but now in a random distribution; the basal periplasm (B) shows only a few vesicles. Nucleoli (n) are visible within the nuclei. (D) Beginning of phase III (see text), at 3-3.5 hr of development. The furrows (F) of fully formed double membranes extend about 10 ,um from the surface indentations to the furrow canals (C). Vesicles (V) are now accumulated, but in a random distribution, in the basal periplasm (B), but few are present in the apical region. (E) End of phase III, approaching the completion of cellularization. The furrow membranes (F) are now elongated to about 25 ,um. Furrow canals (C) of adjacent furrows have in some cases (cells labeled BC) fused with one another to produce an intact cell. In other cases, fusion has not yet occurred (arrows). The basal periplasms contain numerous vesicles (V), but the apical regions contain few. Yolk granules (Y) are sequestered to the cytoplasm underlying the forming cells. (F) The cellular blastoderm at about 3.5 hr of development. The cells, all delineated by fully formed membranes, are now arranged in two or three layers. Downloaded by guest on September 26, 2021 2202 Developmental Biology: Loncar and Singer Proc. NatL Acad ScL USA 92 (1995)

FIG. 2. Further details of the successive stages of membrane formation. Each vertical panel contains at the top (A, C, and E) a lower-magnification picture of the surface region (S) of the same specimen examined at higher magnification in the bottom of the panel (B, D, and F, respectively). (A, C, and E are turned about 90° clockwise from the orientations in Fig. 1.) B rep- resents phase I (see text) and the alignment of the vesicles (V) that are clearly membrane-bounded (ar- mt> rows). Note that in A this specimen still has spherical nuclei (N) and shows as yet no surface furrowing. D r shows the subsequent concerted vesicle fusions (aster- tz@; isks) between regions of already-formed double mem- branes (arrows). F indicates the following stage of nearly fully formed furrow membranes (F), with alter- nating regions of closely opposed (arrows) and some- what more separated double membranes. M, mitochon- dria; R, ribosomes; B, the basal direction of membrane growth. [Bars = 2 ,um (A, C, and E), 0.5 ,um (B), 0.4 ,um (D), and 0.2 ,.tm (F).]

the periplasm. These membrane-bounded vesicles are highly the interior. A large number of the transparent vesicles are electron-transparent and range from 0.05 ,um to 0.5 ,um in randomly distributed in the apical periplasmic space (as in Fig. diameter. They are among the most abundant types of struc- 1A). We recognize three consecutive phases in the subsequent tures observed within the periplasm. In different phases of the cellularization process. In phase I, the transparent vesicles in cellularization process, these vesicles occupy'from 5% to near- the periplasmic space (Fig. 1A) are concentrated and aligned ly 20% of specific regions of the periplasmic volume (Table 1). within the spaces between the nuclei (Figs. 1B and 2B) for a Just prior to the onset of cellularization, after mitotic cycle distance of about 7 ,um from the surface of the embryo into the 13, the spherical'nuclei are present in the periplasmic space, interior. In the next half hour or so, in phase II, these aligned and the yolk granules have been removed from this space into vesicles undergo concerted fusions with one another (Fig. 2D), starting with vesicle fusions at the newly formed furrows in the Table 1. Amount of vesicles in the periplasm during early membrane at the embryonic surface (Figs. 1 C and 2C). Even- Drosophila development tually, closely opposed continuous double membrane struc- Vesicle volume tures form (Fig. 1 C and D; Fig. 2F), terminating at their basal Developmental stage* density, (%) ends in a connecting furrow canal (Fig. 1 C and D), as has previously been reported (3). Zygote (stage 1) 12 3 These events ofvesicle alignment and fusions in phases I and Syncytial blastoderm (stage 4) 10 ± 4 II generate only'about half or less of the plasma membrane Syncytial blastoderm (stage 5) required for complete cellularization. The forming membrane Phase I is continuous with the plasma membrane ofthe blastoderm but Apical-internuclear space 29 ± 5 extends only part of the distance into the periplasmic space, Basal periplasm 5 3 and at this stage the nu'clei remain in direct contact with the Phase II syncytial cytoplasm. The vesicles that are depleted from the Apical periplasm 19 4 apical periplasm in phase I appear then to be replenished Basal periplasm 5 3 (Table 1; compare Fig. 1 Cwith Fig. 1B). However, these apical Phase III vesicles are randomly distributed and have never been ob- Apical periplasm 4 4 served to line up or to undergo concerted fusions as in phases Basal periplasm 21 ± 6 I and II. Although these apical vesicles subsequently largely Cellular blastoderm (stage 6) 5 t 4 decrease in number (Table 1; compare Fig. 1D with Fig. 1 C), *Determined according to ref. 6. their fates are not directly observed. Their decrease in the Downloaded by guest on September 26, 2021 Developmental Biology: Loncar and Singer Proc. Natl Acad Sci USA 92 (1995) 2203 apical periplasm is, however, followed by an increase in phase certainly sufficient to provide the necessary total membrane III in the number of similar randomly distributed vesicles in the mass. Indeed, no membranous structures other than the basal periplasm (Fig. 1 D and E; Table 1). From these results, vesicles are present in quantities that might alternatively we propose that in phase III of cellularization, the double provide the membrane required. membranes formed in phases I and II are further extended by In previous morphological studies (3, 10), multiple villous a succession of single fusions of individual vesicles with the projections were observed on the surface of the blastoderm furrow canals, until at the final stage of the process, two prior to and throughout the first, slow step of cellularization adjacent furrow canals (Fig. 1E) undergo fusion with one but were no longer seen after the second, fast step. It was another to completely enclose each nucleus within a fully suggested that the flattening out of these villous projections continuous single plasma membrane (Fig. 1F). In this last interiorized the membrane mass required for the fast step. stage, remnants of the double membranes that are not utilized However, in our transmission electron micrographs (Fig. 1 in the formation of the final single membranes must be A-D; Fig. 2A, C, and E) we never observed villous projections accounted for, but we have no direct observations of their at the blastoderm surfaces at any stage. It is possible that the fates. villous projections may have been artifacts of the fixation Previous morphological studies (3) have suggested that procedures used in the earlier studies. membrane formation during cellularization occurs in two A question posed by our results and conclusions is, where do distinct steps with different rates. In the first and slower step, the transparent vesicles originate that we propose are used for membrane invaginations between nuclei form, extending from cellularization? In most ordinary cells, new plasma membrane the plasma membrane of the blastoderm about halfway (-10 mass arises by fusion of vesicles derived from the Golgi ,um) to its final penetration. In the second and faster step, the apparatus with already existing plasma membrane. Of great rest of the process of membrane extension and nuclear encir- interest, therefore, is the recent demonstration (11) at the light clement occurs. In addition to such morphological evidence, microscopic level of resolution that punctate structures that genetic studies have shown that there are mutations which are immunofluorescently labeled for the Golgi-associated independently affect either the slow or fast steps of membrane protein p3-COP are located in the periplasm of the Drosophila formation, supporting the idea that two different mechanisms syncytial blastoderm exactly where the transparent vesicles operate in the two steps (1). Our observations are consistent that we have observed at the much higher resolution of the with this idea. The first step of membrane formation (our electron microscope are located, even to the extent of their phases I and II) is associated with the lining up of the vesicles apparent internuclear linear arrangement (figure 3E in ref. 11) between the nuclei extending part way into the periplasm, during the first phase of cellularization. It would make sense followed by their concerted fusions. The second step of mem- if indeed the transparent vesicles that we have observed in our brane formation (our phase III) appears to involve only studies to be plasma membrane precursors were derived from repeated single fusion events at the furrow canal. In fact, since the Golgi apparatus and were then translocated (2) via cy- the compositions of the vesicles are unknown, it is conceivable toskeletal-mediated transport to the periplasm of the blasto- that, despite their similar appearance, the vesicles used in the derm and subsequently lined up in the internuclear spaces. first and second steps are different. The vesicles used in the Immunofluorescence and immunoelectron microscopic stud- first step (phases I and II) presumably were the same or closely ies should help to characterize the composition of the trans- similar to the vesicles already seen in large numbers (Table 1) parent vesicles relative to that of.the Golgi apparatus and of the in the periplasmic space in earlier stages of blastoderm devel- forming membranes, as well as the possible relationships of opment (Fig. 1A), prior to nuclear migration to the periphery. cytoskeletal elements to the vesicles and their redistributions. This suggests that these vesicles are the products of maternal gene transcription. On the other hand, the vesicles used in the We are grateful to Mr. Lance Washington for his help and advice second step of membrane formation (phase III) might origi- in connection with the use of the JEOL 1200 EX electron microscope. nate only after zygotic gene transcription has been turned on We are indebted to Dr. James Posakony for several helpful discussions. (see ref. 1). Therefore, even if the sets of vesicles used in the The assistance of Ms. Lucy DeVito in the collection and manipulation first and second steps are both Golgi-derived (see below), they of the Drosophila eggs and embryos was most valuable. The work was might nevertheless be different in composition. Such differ- supported by National Institutes of Health Grant GM 15971 to S.J.S. ences could account for the different mechanisms of mem- brane formation in the two steps. 1. Schweisguth, F., Vincent, A. & Lepesant, J.-A. (1991) Biol. Cell 72, 15-23. Any proposed mechanism of membrane formation during 2. Schejter, E. D. & Wieschaus, E. (1993) Annu. Rev. Cell Biol. 9, cellularization must account at least qualitatively for the 67-99. amount of membrane needed. If we assume that each of the 3. Fullilove, S. L. & Jacobson, A. G. (1971) Dev. Biol. 26, 560-577. cells in the newly created cellular blastoderm is a cylinder of 4. Sanders, E. J. (1975) Cell Tissue Res. 156, 463-474. 4.5-,tm diameter and 20-,um length, the top surface of which 5. Kamakaka, R. T. & Kadonaga, J. T. (1994) Methods Cell Biol. 44, is supplied by the already formed plasma membrane of the 225-235. embryo, the ratio of new membrane area required to that ofthe 6. Campos-Ortega, J. A. & Hartenstein, V. (1985) The Embryonic original plasma membrane of the blastoderm is about 17. If we Development of (Springer, New York). take 0.2 ,um as the average diameter of the transparent vesicles, 7. Fullilove, S. L., Jacobson, A. G. & Turner, F. R. (1978) in The a total of about 2200 vesicles would be required per cell to Genetics and Biology ofDrosophila, eds. Ashburner, M. & Wright, T. R. F. (Academic, London), pp. 106-227. provide this new membrane area. In specimen sections of 8. Weibel, E. R. (1969) Int. Rev. Cytol. 26, 235-302. about 0.1-,um thickness used in the electron microscope, this 9. Weibel, E. R. (1979) Stereological Methods: Practical Methodsfor would come to a total of about 50 vesicles per cell per section Morphometry (Academic, London), Vol. 1. over the entire process of cellularization. Visual inspection of 10. Turner, F. R. & Mahowald, A. P. (1976) Dev. Biol. 50, 95-108. Fig. 1 B-E indicates that the number of vesicles present in the 11. Ripoche, J., Link, B., Yucel, J. K., Tokuyasu, K. & Malhotra, V. periplasmic space in each phase of membrane formation is (1994) Proc. Natl. Acad. Sci. USA 91, 1878-1882. Downloaded by guest on September 26, 2021