Imaging the Migrating Border Cell Cluster in Living Drosophila Egg Chambers

Abstract

Cell migration is a key process in animal development and central to the spread of cancer. Border cell migration in Drosophila egg chambers is an excellent general model for cell migration, but lacks techniques for studying this process in living cells. Here, we describe a simple and effective method of preparing egg chambers in halocarbon oil. The movement and behavior of the migrating border cells can reproducibly be followed in up to 25 egg chambers simultaneously by time-lapse microscopy using a variety of green fluorescent protein markers on a widefield microscope over a period of 4 hr. Our studies reveal a remarkably linear migration route of the border cell cluster and highly dynamic activity within their cluster. Migrating cells rapidly alter their relative positions and generate transient protrusions. These activities are likely to play key roles in the mechanism of migration and cannot readily be analyzed using fixed samples. Developmental Dynamics 236:2818–2824, 2007.

Introduction

Cell migration plays key roles during animal development and in the spread of cancer cells by metastasis, but the mechanisms involved are only beginning to be understood. Many systems where cell migration occurs, share important features. For example, in all known invasive cell migrations, the Cadherin signaling pathway plays a major role. Unfortunately, most systems for studying cell migration are complex and experimentally intractable, making detailed functional analyses of invasive cell migration very difficult. In contrast, the migration of a group of cells, termed border cells, during stage 9 of oogenesis in Drosophila offers a very tractable model system (Rorth, 2002; Montell, 2006). The border cell cluster consists of 6–10 highly specialized cells. Within the cluster, the two polar cells are more centrally located and surrounded by the outer cells, which seem to have migratory behavior. The cluster moves as an organized group from the anterior of the egg chamber through the nurse cells to the oocyte. Once at the border with the oocyte, the migratory cells line-up and change direction, moving along the border to a dorsoanterior position. Here, they are responsible for the correct development of the micropyle, a structure that subsequently allows entry of sperm to the egg. At least three signaling pathways are involved in the process of border cell migration, Ec-dyson for the timing, JAK/STAT for the cluster to become migratory, and Pvf1 to keep the border cell cluster on course (Montell, 2003). It was previously estimated that the border cell cluster takes approximately 6 to 6.5 hr to travel the approximately 150 μm total distance involved. However, detailed studies of the timing and mechanism have been severely hampered by the fact that it was not feasible to follow the migratory process in living egg chambers.

Here, we report a rapid method of dissection and culturing of ovaries in halocarbon oil, which supports their normal development for up to 4 hr. Using this method, we reveal a re-markably linear migration route of the border cell cluster and a highly dynamic behavior of cells within the cluster, with the migrating cells changing relative positions rapidly and generating transient protrusions. While this manuscript was in preparation, two alternative techniques for preparing egg chambers in aqueous solution were reported independently (Prasad and Montell, 2007; Bianco et al., 2007). These methods also allow the imaging of migrating border cells and revealed similar cellular behavior to those we describe here. However, the method of mounting living egg chambers reported here is more easily performed and allows clearer imaging by fluorescence and differential interference contrast (DIC) microscopy than preparation in aqueous solution. The egg chambers can readily be maintained in focus throughout the imaging period, because, unlike egg chambers mounted in aqueous solution, they are attached to the coverslip surface during the dissecting process. We report the migration of the border cells to be a highly motile process, with extensions of the outer border cells reaching in all directions and the individual cells of the cluster frequently changing their relative positions within the cluster.

Results

In Vivo Imaging of the Migrating Border Cell Cluster

To observe the migration of the border cell cluster, we initially used a fly line expressing a fusion between a nuclear localization signal and green fluorescent protein (nlsGFP). This strategy shows fluorescence in the nuclei of all cells, including the nurse and follicle cells. However, the nuclei in the border cell cluster are much smaller than those of the polyploid nurse cells, so they can easily be identified by nlsGFP expression in a mid-stage 9 egg chamber (Fig. 1A). Within the border cell cluster, the polar cells show an enriched expression of nlsGFP, as they stop dividing early in oogenesis. Therefore, the polar cells are readily identified by their central positions and a higher expression level of nls- GFP. Following preparation of egg chambers in halocarbon oil (see the Experimental Procedures section), we could reproducibly image the migration of the border cell cluster in living egg chambers for more than 2 hr and over a distance of more than 30 μm (Fig. 1C–I).

Fig. 1. In vivo imaging of border cell migration. Nuclear localization signal and green fluorescent protein (nlsGFP) is expressed in nuclei of the nurse cells, the surrounding follicle cells, and the migrating border cell cluster.

A: Mid-stage 9 egg chamber, in which the nuclei of diverse types can be visualized. The border cell cluster is clearly distinct (marked with arrowhead), with the two polar cells in the middle of the cluster expressing slightly more nlsGFP and six visible surrounding border cells. In all images presented, the egg chambers are orientated with the anterior to the left and the posterior to the right. The border cells migrate from the anterior to the posterior. B: Schematic drawing of a mid-stage 9 egg chamber. The nurse cell nuclei and the smaller oocyte nucleus are shaded in gray. The polar cells in the migrating border cell cluster are indicated in blue. The follicle cells and the outer migratory cells, which are derived from follicle cells, are depicted in pink. C: An egg chamber dissected in halocarbon oil allows clear imaging of the border cluster, here visualized at time 0. D: One hundred sixty minutes later, the border cell cluster has traveled 48 μm (see also Supplementary Movie S1). E–I: Enlarged images of the migration taken every 40 min. The border to the oocyte is to the right. The border cell cluster migrates on a straight line through the landscape of nurse cells, as all images are taken in the same Z focal plane. The cluster itself is very dynamic.

We observed that the process of border cell migration could be divided into three phases: The first stage was the detachment of the border cell cluster from the epithelial cell layer. This stage was followed by an initial fast migration through the anterior 30-70% of the length of the egg chamber, with clusters showing an average speed of 0.3 μm/min (n = 3). However, the speed of cluster migration was very variable. During this process, the cells within the cluster are highly motile (Supplementary Movie S1, which can be viewed at http://www.interscience.wiley.com/jpages/1058-8388/suppmat). Finally, a period of slower migration was observed with an average speed of 0.17 μm/min (n = 4) as the cluster approached the oocyte border. These findings are in agreement with the two phases of active migration recently described (Bianco et al., 2007).

We only once observed the detachment of the border cell cluster from the anterior follicle cell sheet. This finding might be due to the fact that insulin, which is needed for this process (Prasad and Montell, 2007; Bianco et al., 2007), was not present in the halocarbon oil used as mounting medium. We never filmed the entire migration in a single follicle, largely because imaging was generally limited to 2–3 hr. However, we have filmed for up to 4 hr, and it is certainly possible that continuous imaging could be maintained over even longer periods.

When dissecting ovaries in halocarbon oil, we pull individual ovarioles over the coverslip until they are freed from all surrounding fibrous material. This method has distinct advantages to the recently published alternative methods (Prasad and Montell, 2007; Bianco et al., 2007). The egg chambers are attached to the coverslips, which enables the filming of border cells in one specific Z section without the need to adjust the focus during the imaging period. In addition, halocarbon oil has a refractive index similar to that of glycerol, making it more favorable for optical imaging than aqueous mounting.

We noted that the migration of the border cell cluster occurs along a remarkably straight line parallel to the anteroposterior axis. This was evident as the processes of migration could almost always be filmed without alterations to the plane of focus (Supplementary Movie S1). This finding is surprising, as it might have been anticipated that the migrating cells would have been deflected from their course by the much larger nurse cells.

Border Cells Frequently Change Their Positions Within the Cluster

The ability to observe the movement of the cluster in real-time led us to test whether individual border cells maintain their relative positions during migration. We, therefore, tracked the individual nlsGFP‐marked border cells by in vivo imaging during migration. We observed that the outer migratory cells move around the polar cells during migration, causing the relative positions of the cells within the cluster to change with time (Fig. 2A–D; Supplementary Movie S2). In the faster phase of border cell migration, the motility of the border cells also seems to be higher. We speculate that this striking motility is directly related to the physical mechanism of migration of the cluster.

Fig. 2. Individual cells can be followed during migration. Enlargement of border cell cluster from Figure 1A and its migration followed over time.

A–D: An individual cell (marked by arrowheads) rolls around the polar cells during migration (see also Supplementary Movie S2). D: The outer border cell, that was followed in A–C moves into a different focal plane (marked by black arrowhead). To the right of the images, the diagrams indicate different cells by color: pink are outer migratory cells, blue are the mostly central polar cells, and in green is one outer border cell, which clearly changes its position relative to the polar cells during migration.

Border Cell Migration Is Not Influenced by Fluorescent Imaging or by the Preparation of Egg Chambers

To test whether the migration of the border cells in nlsGFP flies is comparable to that in wild-type flies, we dissected egg chambers of Oregon R flies and observed these by DIC imaging. The individual cells of the cluster were detectable in DIC, and the polar cells could be identified by their position and their migratory behavior (Fig. 3A). Clusters that migrated for more than 30 μm had velocities of 0.35 μm/min (n = 4), similar to those observed in egg chambers expressing nlsGFP. Border cell migration could be observed at all locations along the migration path. When the clusters approached the oocyte border, the oocyte appeared to bulge toward the cluster to facilitate contact (Fig. 3C; Supplementary Movie S3). In many cases, it appeared that the cluster then ceased active migration and was absorbed by the expanding border of the oocyte (data not shown). This finding frequently appeared to be initiated by a projection from the oocyte border (arrow in Fig. 3C; and Supplementary Movie S3). This finding presumably reflects signaling between the approaching cell cluster and the oocyte border.

Fig. 3. Border cell migration is not influenced by fluorescent imaging or by the preparation of egg chambers.

The border cell cluster can be imaged in vivo by differential interference contrast (DIC) microscopy. A–D: Time course of migration followed in DIC imaging. A: The polar cells (see arrows) and surrounding border cells are clearly distinguishable. B: The arrangement of individual cells changes, while migration proceeds. C: Structures are frequently visible (see arrow) that appear to project from the oocyte border toward the approaching border cell cluster. D: The cluster frequently appears to be “absorbed” into the oocyte border. E,F: The microtubule network of dissected egg chambers appears normal as visualized by Tau-green fluorescent protein (GFP). Furthermore, Tau-GFP is not expressed in the border cells; therefore, the cluster appears as a black hole (marked by arrows). Here, the border cell cluster migrated 77μm in 2 hr, which was the fastest movement we imaged.

The microtubule (MT) network is sensitive to physical and metabolic perturbations. To test whether the MT network is affected by our dissection and imaging methods, we analyzed a fly line expressing the MT-binding protein bovine Tau-GFP (Micklem et al., 1997). Tau is not expressed in the border cell cluster, as the expression of the GFP fusion is under the control of the maternal α4 tubulin promoter, which drives female germline-specific expression. The migrating border cell cluster, therefore, appears as a dark region, surrounded by the MT network of the nurse cells (Fig. 3E,F). We observed no defects in the MT network in Tau-GFP lines in which border cell migration was imaged in real-time. The border cell cluster shown in Figure 3F migrated a distance of 77 μm in 110 min (Supplementary Movie S4), which correlates with a speed of 0.7 μm/min. As in other lines tested, we found the speed of migration of border cells expressing Tau‐GFP to be variable, but with an average of 0.4 μm/min (n = 8).

Dynamic Protrusions Extend in All Directions From the Border Cell Cluster

In fixed material, we observed an extensive MT network in the migrating border cells, which was also readily detected in protrusions extending out from the cluster. No polarity could be determined for MTs in these protrusions in fixed samples, and we, therefore, examined flies expressing a fusion between GFP and EB1, which localizes at the plus end of growing MTs (Shimada et al., 2006). In migrating border cells, EB1-GFP was enriched in the cytoplasm of the polar cells and was present at lower levels in the cytoplasm of the surrounding migratory cells (Fig. 4B–F). EB1-GFP was also readily detected in the protrusions, but no clear polarity could be ascertained.

Fig. 4. Dynamic protrusions extend in all directions from the border cell cluster.

A: Protrusions of the migrating border cell cluster contain a distinct microtubule network in fixed material, here visualized with anti-Tubulin. B: When the border cell cluster is still attached at the anterior, two long extensions can be seen, which are presumably identical to the previously described actin-rich long cellular extension (LCE). C: A long, microtubule-rich extension extends from the anterior of the border cell cluster until the cluster has advanced approximately 30% along the migration path. D–F: The border cell cluster expressing EB1GFP was followed over time. Protrusions were observed to extend in several different directions. In some cases, EB1GFP appeared to accumulate at the tip of the protrusions.

It was previously reported from observation of fixed material that a single migratory border cell from the cluster generates a long cellular extension (LCE; Fulga and Rorth, 2002). The LCE is an actin-rich structure that forms at early stage 9 and is proposed to play an important role in invasive migratory behavior. In living stage 9 egg chambers, clusters that lie within the first third of the total migration distance were associated with extensions that could be as long as half the egg chamber (Fig. 4B). These extensions are likely to be the same as the previously reported LCE. At this stage, we generally did not observe protrusions in other directions. At later stages of migration, we also observed more numerous but shorter and highly dynamic protrusions. These originated from all migratory border cells, with the notable exception of polar cells (Fig. 4D–F; Supplementary Movies S5 and S6). The protrusions were enriched in the direction of migration, but extended in all directions. In some egg chambers, the protrusions were observed to shrink and disappear following imaging for 3–4 hr. When this occurred, the border cell cluster abruptly ceased migrating, and in many cases, this was followed by death of the egg chamber, which can be easily determined by the deformation of the nurse cell nuclei or when the nurse cell nuclei stop moving and rotating, which also was observed by Bianco et al. (2007).

In addition, migrating cell clusters, which were still in the first half of the path, were associated with a MT bundle that extended all the way back to the anterior of the egg chamber, marking the path the border cell cluster had taken (Fig. 4C). These anterior extensions sometimes reached halfway through the nurse cell region of the egg chamber. Strikingly, individual cells were frequently left behind on this bundle. Sometimes these cells could be seen to rejoin with the cluster, moving at a particularly high speed (data not shown). The mechanical basis of these effects remains to be determined.

Discussion

Here, we have reported a simple method of dissecting and mounting egg chambers in halocarbon oil, allowing the migration of the border cell cluster to be followed in real-time in living cells. We also describe several GFP markers that allow the rearrangement and dynamic cellular extensions of the cells to be followed in detail for up to 4 hr. Analysis of a Tau-GFP line showed that the MT network in dissected egg chambers is not disturbed by the preparation, when compared with previous descriptions of Tau-GFP in halocarbon mounted egg chambers (Micklem et al., 1997).

Following the border cell cluster in living cells revealed different phases of migration. First, the border cells detach from their surrounding epithe‐lial neighboring cells in the anterior follicle cell layer. We were able to image the actual detachment only infrequently, possibly because stimulation from external factors, including insulin (Prasad and Montell, 2007; Bianco et al., 2007), is required to trigger release. The second phase, which was most reliably observed, was a rapid migration through half to two thirds of the nurse cell region. In the last, more passive phase, the cluster appeared to be less motile and the migration was generally slower than that over the first two-thirds of migration. In some cases, the border cell cluster migrated to the border of the oocyte. In other cases, the border cells did not migrate all the way to the oocyte border, but instead were contacted by outgrowths from the oocyte. This finding strongly suggests that the approaching cell cluster signals to the oocyte border, but the putative signaling molecule(s) involved remain to be identified.

Using nlsGFP to identify the nuclei of the migrating border cell cluster, we observed that the outer border cells roll around the polar cells, which are in the center of the cluster. This finding could also be observed by DIC imaging. Using the growing MT plus end marker EB1GFP, we observed that long extensions from the migrating border cell cluster, which were previously described as actin-rich LCEs, also contain a high density of MTs. At higher magnification, we could identify single EB1GFP particles (unpublished observations). However, more sophisticated particle-tracking analyses will be required to determine whether the EB1GFP particles move in one direction only, or are enriched at the leading edge of these protrusions when they are in the process of extending. Analyses in living cells, revealed the presence of shorter and more dynamic protrusions emanating from the cluster at later stages in the migration process.

The roles played by the protrusions and dynamic cell movements within the cluster in the mechanism of migration are currently unclear. We propose that the protrusions could provide traction, while the intracluster movement contributes to the motile force needed to propel the migrating cluster through the nurse cells. Another striking feature of the migration was its linear path. The border cell cluster showed no sign of being deflected from a straight line of migration by the presence of the much larger, polyploid nurse cells. If the migrating cells are following a chemoattractant gradient of PVF1 as previously proposed (Duchek et al., 2001), it might have been expected that this gradient too would be affected by the position of the nurse cells, leading to their circumnavigation by the border cells.

While the present work was in preparation, Prasad and Montell (2007) and Bianco et al. (2007) described aqueous-based mounting methods, requiring the addition of insulin. These techniques also allowed the visualization of the migrating cell cluster in living oocytes and revealed very similar cellular behaviors to the ones we describe here, despite the markers used being distinct. One difference between our technique and their aqueous mounting methods is that, while we are unable to image the first step of migration, the addition of insulin overcomes this difficulty. Nevertheless, the dissection, mounting, and imaging of egg chambers in halocarbon oil, we have described here, offers several advantages over aqueous media for the imaging of the subsequent steps of migration. First, the egg chambers remain attached to the cover slip surface, allowing injection of reagents while the egg chamber remains in one plane of focus. This avoids the need to readjust the focus during imaging, which is required with aqueous mounting. The lack of drift of dissected egg chambers, also allows up to 25 egg chambers to be imaged in parallel during one timelapse experiment, using a precisely controlled xyz stage. Second, halocarbon oil has a higher refractive index than water, so that higher numeric aperture, oil immersion objectives can be used more readily for imaging with greater optical clarity. Third, in halocarbon oil, it is possible to remove all the tissue surrounding each egg chamber and to image each egg chamber in isolation, thus reducing stray light from surrounding tissue and improving the quality of the resulting imaging.

Live cell imaging of border cell migration will allow analyses of the mechanisms of border cell migration in wild-type or GFP-marked lines, as well as in lines mutant for factors that play important roles in migration in more detail. Furthermore, it is now feasible to undertake an in vivo screen for factors that are involved in the organized migration of the border cell cluster. It will also be possible to investigate in vivo the properties of potential chemoattractant or -repellent compounds.

Experimental Procedures

Egg Chamber Dissection and Live Imaging

Freshly eclosed flies were placed on fresh standard cornmeal–agar medium with yeast extract sprinkled on top at 25°C for exactly 2 days. Anesthetized females were then submerged in halocarbon oil Series 95 and the ovaries dissected out. The ovaries were immediately placed in a fresh drop of halocarbon oil Series 95 on a Borosilicate cover glass (No. 1 thickness). Using forceps, the ovary was pressed to the coverslip at its end containing mature eggs and then the surrounding fibrous network was gently torn with two to three strokes of a Tungsten dissecting needle. The Tungsten needle was then wrapped very gently around the germaria and egg chambers up to stage 5. Holding on to these younger egg chambers, individual ovarioles were then gently pulled out of the ovary, until they attached to the glass without any surrounding material in proximity. By this preparation, stage 8-9 egg chambers were undamaged. Imaging was performed on a widefield Deltavision microscope (from Applied Precision on an Olympus inverted IX70 microscope using a Coolsnap HQ CCD camera from Roper Scientific). Images were acquired with 20×/0.75NA unless otherwise stated. Up to 25 living egg chambers were imaged in parallel, using a precisely controlled xyz stage for revisiting individual egg chambers on one coverslip.

Fly Strains

Wild-type flies were Oregon R; nls- GFP contains four copies of a poly‐Ubiquitin promotor-driven nuclear GFP transgene (Davis et al., 1995). EB1GFP flies are also under the control of the polyUbiquitin promotor (Shimada et al., 2006). Tau-GFP is under the control of the maternal α4 tubulin promoter (Micklem et al., 1997).

Immunostaining

Ovaries were dissected and then fixed in 4% paraformaldehyde in phosphate buffered saline (PBS) with 0.5% NP40 and heptane, followed by washes with 0.2%Tween in PBS and 1% Triton in PBS, preincubation in 5% goat serum 0.2% Tween in PBS and fluorescein isothiocyanate–conjugated Tubulin antibody from Sigma (clone DM1A) followed by washes in PBS containing 0.2% Tween.