Time-lapse film analysis of cytoplasmic streaming during late oogenesis of Drosophila

The analysis of maternal-effect mutants in Drosophila has shown that mature oocytes contain the information specifying the axial coordinates in the egg (Bull, 1966; Lohs-Schardin & Sander, 1976; Nüsslein-Volhard, 1977). While the determination of the axial coordinates may be the result of gradients building Up during oogenesis or very early embryogenesis (Nüsslein-Volhard, 1979), the pole plasm contains local determinants which function autonomously in transplantation tests (Illmensee, Mahowald & Loomis, 1976). The information-carrying molecules must have been synthesized in either case during oogenesis and deposited at the appropriate sites in the oocyte. Little is known about the mechanisms that would allow such specific deposition of molecules in the oocyte.

The analysis of oogenesis has been hampered, because of the lack of media that would allow oogenesis to continue in vitro. Only in case of the paedo-genetically reproducing gall midge Heteropeza conditions were found which permit normal development so that oogenesis in vitro could be analysed by means of time-lapse films (Went, 1977). The oocytes were shown to pulsate rhythmically and follicles were seen to rotate in the ovary prior to their release into the body cavity. The reasons for these follicular motions are unknown.

In Robb’s medium (Robb, 1969) ovarian follicles of Drosophila undergo apparently normal development during late vitellogenesis in vitro (Petri, Mindrinos, Lombard & Margaritis, 1979). This observation led us to study the follicular development in vitro by means of time-lapse films allowing direct visualization of cytoplasmic movements.

Female wild-type flies (Oregon R) were dissected in Robb’s medium and follicles isolated with tungsten needles. The follicles were placed in a small incubation chamber on a siliconized slide or in a shallow depression slide and covered with a coverslip. Alternatively, a flow-through chamber (Vollmar, 1972) which allows for sufficient oxygen supply and prevents desiccation was used.

The follicles were filmed using bright-field optics. 16 mm films (Kodak Plus-X Reversal) were prepared using a Bolex H 16 reflex camera attached to a LeitZ microscope. The film was finally analysed using a Kodak analyst projector. Since the morphology of the follicle changes only slightly during individual film sequences, the observed cytoplasmic streaming was plotted on photographs prepared from a single frame of the particular film sequence studied.

Cytoplasmic streaming in a total of 63 follicles of different developmental stages was analysed (stages 7–9/7 follicles; stage 10A/9; stages 10–B12/25; stage 13/8; stage 14/14). The observed time pattern of cytoplasmic streaming during development is highly reproducible, but ooplasmic streaming during stages 10–12 may vary with respect to speed and direction of movement (see under ‘Results’).

When follicles are filmed for 30–45 min in Robb’s balanced saline solution instead of Robb’s medium the same pattern of cytoplasmic streaming is obtained.

When stage-10 follicles were left to develop in vitro misshaped chorionic filaments and a remaining nurse-cell cap were typically observed (Petri et al. 1979). Judged by morphological criteria follicles older than stage 10 developed normally in vitro after filming.

When vitellogenic follicles (stage 10) are isolated and cultured in Robb’s medium for up to 11 h they develop into mature oocytes (stage 14) and, furthermore, the time course of this development in vitro is comparable to that in vivo (Petri et al. 1979).

We chose to isolate follicles of each developmental stage and to film cytoplasmic streaming only for 30–45 min following the isolation of the follicle since in this way possible artefacts as a result of longer incubations in vitro might be avoided or reduced. However, we have no evidence that this precaution was necessary.

Stage-9 and younger follicles do not develop to maturity in vitro, presumably due to a lack of essential growth factors in the medium. It is, therefore, important to show that these follicles remain viable in vitro for the period of filming. To test the viability of follicles (stage 10 and younger) after culturing in vitro for 30–45 min the follicles were injected into female flies homozygous for the female-sterile mutation ovarian tumor (ovt). In these flies no ovarian follicles are formed (Dr E. Gateff, personal communication) and, therefore, the implant can easily be detected in the host fly at the end of the in vivo incubation. The injected follicles were allowed to develop in vivo for 48 h (previtellogenic stages) or 24 h (stages 9 and 10). Finally, the host flies were dissected and the development of the implanted follicles assessed (Table 1). The results show that a large percentage of the injected follicles completed their development in vivo. The smaller percentage of successful implantations with stage-9 and -10 follicles as compared to previtellogenic follicles (Table 1) is most probably due to technical difficulties since mid-vitellogenic follicles are so large that they can easily be punctured or slit open by the sharp edges of the injection pipette. Occasionally parts of damaged follicles were recovered; in most cases these fragments consisted of the posterior pole containing ooplasm surrounded by columnar follicle cells.

In three cases the implanted follicles had not yet reached the final stage of oogenesis when their host fly was dissected (Table 1, asterisked). These follicles completed oogenesis in vitro and hence in these cases the follicles had gone through three changes of in vitro/in vivo culture and yet they developed into stage-14 oocytes with apparently normal morphology.

With respect to protein synthesis no quantitative or qualitative change during 1 h of culture in vitro could be detected when stage-10 follicles were labelled with [³⁵S]methionine immediately after their isolation or after pre-incubation for 1 h and the radioactive polypeptides analysed on SDS-gels (not shown).

In late stage-9 follicles no cytoplasmic streaming can be observed but the ooplasm shows random oscillatory motions in all directions (Fig. 1). In the nurse cells characteristic back and forth movements can be observed which again do not result in any lasting displacement of cytoplasm. At this developmental stage there is no visible indication of cytoplasmic transport from the nurse cells to the oocyte.

At stage 10A the ooplasm begins to stream (Fig. 2). The central area of the oocyte could not be analysed with the methods used because of the thickness of the follicle and strong absorption of light by the yolk platelets. However, the observed pattern of cytoplasmic movements clearly suggests that the streaming extends into the axial area as well (Fig. 1, arrows). Since at stage 10A the nursecell cytoplasm does not visibly stream into the oocyte, the observed ooplasmic movements do not seem to be the result of cytoplasmic influx from the nurse cells.

After the follicle cells at the nurse-cell/oocyte border have completed their centripetal migration (stage 10B) a very different picture emerges. Nurse-cell cytoplasm can be seen streaming into the oocyte through the four ring canals which connect the oocyte with four neighbouring nurse cells (King, 1970). At first only the cytoplasm of these four nurse cells is poured into the oocyte while the more anteriorly located nurse cells are not affected (not shown). Later, however, when the nurse cells degenerate rather rapidly, cytoplasm flows towards the central area forming a fast and massive stream which reaches the oocyte within a minute (Fig. 3). When the nurse-cell cytoplasm passes the gap left by centripetally migrated follicle cells, it gains the fastest speed and the streaming can even be observed directly under the microscope. Nurse-cell cytoplasm flowing towards the central cytoplasmic stream is occasionally seen to be pushed back into a nurse cell (Fig. 3, dashed lines). Minutes later when the built-up pressure is apparently equilibrated, the cytoplasm flows out the same way it was first pushed back, and merges with the central stream of cytoplasm.

The flow of cytoplasm through the intercellular bridges connecting the nurse cells with each other is indicated by the observed pattern of cytoplasmic streaming (Fig. 3). However, at this stage the cell membranes start to break down (Cummings & King, 1970), presumably a prerequisite for the formation of the large and fast-moving central cytoplasmic stream.

During the development of a follicle from stage 10B to stage 11 the volume of the nurse cells was found to decrease at a rate of about 13000 μm³/min.

The diameter of the cytoplasmic stream passing through the four intercellular bridges connecting the nurse cells with the oocyte was difficult to determine in stage-10B follicles since the depth of focus was not small enough to measure the streaming through each cytoplasmic bridge separately. As a result, the areas between the intercellular bridges, where there is no streaming, cannot be measured reliably since immobile regions and fast-moving cytoplasmic streams may alternate on different levels of focus. However, at stage 11, when the intercellular bridges move closer together, the cytoplasmic stream was found to be about 12–13 μm in diameter.

From the above data the speed of the cytoplasm passing through the cytoplasmic bridges connecting the nurse cells with the oocyte can be predicted to be about 1·8 μm/sec. When the speed of streaming was measured directly in film sequences it was found to be 1·9 ±0·4 μm/sec. Figure 3 shows an example of a particularly fast-moving cytoplasmic stream (about 2·3 μm/sec.).

Since the calculated and the measured speed of cytoplasmic streaming is roughly the same it seems likely that the bulk of the nurse cell cytoplasm and not just large-sized cytoplasmic inclusions is affected by the streaming.

The ooplasm is also in continuous motion. The cytoplasm either flows in a circular fashion (Fig. 4) or in a way described earlier (Fig. 2). In general, the geometry and speed of ooplasmic streaming does not seem to follow any strict rules. By following the cytoplasmic streaming near the periphery of the growing oocyte in time-lapse films, we calculated that a cytoplasmic particle travelling the circular way may complete one round in about 20–40 min (Fig. 4).

When the nurse cell breakdown is almost completed (stage 12), the cytoplasmic streaming from the nurse cells to the oocyte ceases and the ooplasmic movements slow down. The cortical cytoplasm at the anterior end of the oocyte (facing the degenerating nurse cells) gains in thickness and can measure up to 15 μm. It is not known whether the increased cortex width is due to the specific deposition of nurse-cell products. Nurse cells of this late stage were previously shown to synthesize several stage-specific proteins (Gutzeit & Gehring, 1979), but it is not known whether these proteins become incorporated into the growing cortex.

At stage 13 even the cytoplasmic movements within the oocyte come to a halt and no streaming can be detected anywhere in the ooplasm during the final stage of oogenesis.

An inherent problem of in vitro research is the difficulty in assessing the relevance for the in vivo situation. However, we feel confident that our observations reflect by and large the in vivo situation since (1) stage-10 follicles develop normally by morphological criteria in vitro; (2) the developmental time required to complete oogenesis is comparable after in vitro and in vivo culture (Petri et al. 1979); (3) young follicles (stage 10 and earlier) are not irreversibly damaged by the in vitro treatment since they continue developing in vivo and, finally, (4) protein synthesis in stage-10 follicles is quantitatively and qualitatively stable for at least 60 min in vitro.

Once the follicle cells at the nurse cell/oocyte border have completed their centripetal migration, the nurse-cell cytoplasm pours into the rapidly growing oocyte where strong cytoplasmic streaming leads to thorough mixing within minutes. Therefore, the movements do not appear to be involved in the transport of molecules to specific sites in the oocyte. During late stage 12 and stage 13 the cytoplasmic movements cease, first in the nurse cells and later in the oocyte. The ooplasmic streaming can be observed not only before the nurse cytoplasm pours into the oocyte but also after this process is completed. Therefore, it appears that the ooplasmic movements and the cytoplasmic streaming in nurse cells at the time of their rapid degeneration are independently controlled processes.

The reported observations have interesting implications with respect to the site-specific localization of molecules during oogenesis. Because of the rapid ooplasmic streaming during stages 10 to 12, this period appears to be ill-suited for the specific localization of molecules. Consistent with this notion is the finding that polar granules appear already in stage-9 oocytes at the posterior pole of the follicle (Mahowald, 1962). In the wasp Pimpla the oosome is already localized at the posterior pole at the beginning of vitellogenesis long before the nurse-cell cytoplasm streams into the oocyte (Meng, 1968). This also holds true for the germ plasm in follicles of the ants Camponotus and Formica (Bier, 1952). Prelocalized receptors might, of course, bind specific molecules carried around in the stream of cytoplasm and thereby acquire the necessary factors for their function. The pole plasm may be a case in point, since transplantation tests show that it becomes competent during late vitellogenesis to induce pole cells in young embryos (Illmensee et al. 1976).

In the Drosophila oocyte only those molecules which are localized in the approximately 5–7 μm wide cortex at stage 11 may be unaffected by the cytoplasmic streaming.

The polar granules at the posterior pole of late vitellogenic follicles are presumably included in the immobile cortical cytoplasm. In mature eggs of Drosophila hydei polar granules were found to be located in the peripheral 5–10 μm of the egg (Mahowald, 1973) which approximates the cortex thickness (up to 7 μm) of the smaller sized stage-11 follicles of Drosophila melanogaster (Fig. 4). The large oocyte nucleus, however, must be anchored in some unknown fashion to keep its place.

Nüsslein-Volhard (1979) suggested that the anteroposterior and the dorsoventral coordinates of the embryo are defined by two gradients. If gradients are set up by concentration differences of diffusible molecules (Meinhardt, 1977) it seems reasonable to assume that such gradients can only be established in the absence of cytoplasmic streaming. These conditions are met prior to stage 10A and after stage 12.

We wish to acknowledge the invaluable advice and support of Dr H. Vollmar and Prof. K. Sander during the course of this work. We thank the Deutsche Forschungsgemeinschaft for financial support (SFB 46).