ABSTRACT
Primordial germ cells (PGCs) are the founder cell population of the gametes which form during the sexually mature stage of the life cycle. In the mouse, they arise early in embryogenesis, first becoming visible in the extraembryonic mesoderm, posterior to the primitive streak, at 7.5 days post coitum (d.p.c.). They subsequently become incorporated into the epithelium of the hind gut, from which they emigrate (9.5 d.p.c.) and move first into the dorsal mesentery (10.5 d.p.c.), and then into the genital ridges that lie on the dorsal body wall (11.5 d.p.c.). We have used confocal microscopy to study PGCs stained with an antibody that reacts with a carbohydrate antigen (Stage-Specific Embryonic Antigen-1, SSEA-1) carried on the PGC surface. This allows the study of the whole PGC surface, at different stages of their migration.
The appearance of PGCs in tissue sections has given rise to the conventional view that they migrate as individuals, each arriving in turn at the genital ridge. In this paper, we show that PGCs leave the hind gut independently, but then extend long (up to 40 μm) processes, with which they link up to each other to form extensive networks. During the 10.5 – 11.5 d.p.c. period, these networks of PGCs aggregate into groups of tightly apposed cells in the genital ridges. As this occurs, their processes are lost, and their appearance suggests they are now non-motile. Furthermore, we find that PGCs taken from the dorsal mesentery at 10.5 d.p.c. perform the same sequence of movements in culture. At first they are actively locomotory. They become linked to other PGCs via long processes and form clusters of non-motile cells. This aggregation together into closely apposed masses may be an important component of PGC migration from the gut into the genital ridges and would also allow signalling interactions between PGCs.
We show also that the first PGCs to emigrate from the hind gut, between 9 and 9.5 d.p.c., do so directly into the area where the genital ridges will form. This suggests that an adhesive interaction between these ‘pioneer germ cells’ and the target tissue may play a role in the localisation of PGCs into the genital ridges as they aggregate.
INTRODUCTION
Primordial germ cells (PGCs) are the founder population of the gametes. They are set aside at early stages of embryonic development, and later join the somatic tissue of the gonad by a combination of passive displacement and active migration (see Wylie and Heasman, 1993 for review). In the mouse, PGCs can first be identified at the late gastrula stage (7.5 days post coitum, d.p.c.) due to their expression of the enzyme alkaline phosphatase (Ginsberg et al., 1990). This useful marker of mouse primordial germ cells allows their subsequent distribution to be followed in tissue sections. At 8.5 d.p.c., the PGC population is localized to the base of the developing allantois and, by 9.5 d.p.c., they are in the epithelial lining of the hindgut. Between 9 and 9.5 d.p.c., they emigrate from the hindgut into its dorsal mesentery (10.5 d.p.c.) and from there into the genital ridges on the dorsal wall of the abdomen. During this period, PGC numbers increase from less than 100 to 25,000, indicating an approximate cell cycle time of 16-17 hours (Tam and Snow, 1981). Most PGCs have colonized the genital ridges by 11.5 d.p.c. Their morphology in sectioned material (Clark and Eddy, 1975) and their behaviour in culture (Donovan et al., 1986) suggest that PGCs actively migrate from the hindgut to the genital ridges.
PGCs isolated before, during, or after migration have distinctive adhesive properties in vitro (Donovan et al., 1986; ffrench-Constant et al., 1991). Primordial germ cells isolated during the 8.5-10.5 d.p.c. period adhere to certain feeder cell monolayers (Donovan et al., 1986), and to purified extracellular matrix molecules (ffrench-Constant et al., 1991; de Felici and Dolci, 1989). Moreover, it has been found that they are invasive (Stott and Wylie, 1986). Little is known, however, about their interactions with surrounding somatic cells during migration in vivo. An understanding of the changes in cellular associations of primordial germ cells, therefore, may shed light on the mechanism by which these cells are directed to the genital ridges.
During migration, primordial germ cells express on their cell surfaces the Stage Specific Embryonic Antigen 1 (SSEA1, Fox et al., 1981), a trisaccharide of the form galactose [β1-4]N-acetylglucosamine[α1-3]fucose (Gooi et al., 1981). We have used the monoclonal antibody, TG1 (Beverley et al., 1980), which recognises SSEA1 (see Donovan et al., 1987), to examine the 3-dimensional morphology of primordial germ cells in situ at different stages of migration. Using confocal microscopy to analyse the whole-mount samples, we have examined both the time and pattern of primordial germ cell emigration from the hindgut as well as their principal cellular associations. Their appearance in histological sections suggests that germ cells migrate to the genital ridges independently of each other and use cues from the surrounding somatic cells and extracellular matrix to guide them. A surprising fact to emerge from our work is that after their emigration from the gut, primordial germ cells extend long processes, which they use to associate with each other. The sequence of events seen by confocal microscopy between 9.5 and 11.5 d.p.c. suggests that PGC:PGC aggregation plays a role in the accumulation of these cells in the genital ridges. This interpretation is supported by the observation that primordial germ cells isolated from embryos during this same period and cultured in vitro, behave similarly. We also show that the first PGCs to emigrate from the hind gut do so directly into the region of the genital ridges. This suggests that an interaction between these ‘pioneer’ germ cells and the target tissue may be involved in the localisation of PGCs into the genital ridges as they aggregate.
MATERIALS AND METHODS
Whole-mount immunohistochemistry
Embryos were obtained from MF1 mice on 9 – 11 d.p.c. of timed pregnancies (day 0 is the day on which a vaginal plug is found). The yolk sacs were removed and the embryos fixed in freshly prepared paraformaldehyde (4% in PBS) for 2 hours at 4°C. To remove excess fixative, the samples were washed in several changes of PBS. At this stage, embryos were dehydrated into 100% methanol and were either stored at −20°C or embedded in PEDS wax for sectioning. Embryos at 10.5 d.p.c. or older were rehydrated to PBS and dissected further to expose primordial germ cell-containing tissues. To increase the access of the antibody to the tissues, the samples were incubated for 15 minutes at room temperature in PBS containing 2 mg/ml BSA, 0.1% Triton X-100, 0.02% sodium azide (PBTA). Non-specific antibody-binding sites were blocked by incubating the samples in PBTA containing 10% goat serum (PBTAS). The embryos were then incubated overnight at 4°C in an undiluted supernatant of the mouse hybridoma line TG1, which secretes a monoclonal antibody that reacts with SSEA1 (a gift from Peter Beverley). The samples were washed hourly for 5 – 6 hours in PBTA at room temperature before incubation overnight at 4°C with the secondary antibody, a fluorescein-conjugated goat anti-mouse Ig (Nordic) diluted 1/50 in PBTA. After extensive washing, the embryos were dehydrated through a methanol:PBS series to 100% methanol. All incubations to this point in the procedure were performed on a rotating platform. The embryos were cleared in benzoyl alcohol:benzoyl benzoate (1:2) and mounted in this mixture on cavity slides for viewing by confocal microscopy. The number of PGCs attached to each other was counted by following individual PGCs through z-series images in 0.5 μm steps. The standard error of the percentage (SE%) was calculated from the data according to the formula (pq/n)G. This can be used to assign confidence limits for the percentages such that there is a 68% chance that the population % lies between ±1SE% of the sample %, a 95% chance that the population lies between ±2SE% of the sample and a 99% chance that the population lies between ±3SE% of the sample %. Thick (20 μm) sections of wax-embedded embryos were dewaxed in acetone, blocked in PBS containing 10% goat serum/0.02% sodium azide (PSA) and stained as described for the whole mounts except that the antibody incubations were performed at room temperature for 3 hours and washes were with PSA. The sections were mounted in aqueous mounting medium containing an anti-quench agent (90% glycerol, 10% water, 100 mg/ml DABCO triethylenediamine (Sigma)).
All samples were analysed using a BioRad scanning-laser confocal microscope (MRC 600).
In vitro culture of primordial germ cells
The primordial germ cell-containing tissues (the urogenital ridges and the hindgut mesenteries) of 10.5 d.p.c. mouse embryos were dissected away from other tissues in Ca2+-Mg2+-free PBS. The tissue fragments were triturated repeatedly in a small volume of PBS to give a single-cell suspension. The cells were then washed in DMEM supplemented with 10% FCS, glutamine (4 mM), penicillin and streptomycin. They were pelleted and resuspended in the same medium before plating out onto a preformed STO cell monolayer (prepared as described previously (Donovan et al., 1986)) on poly-D-lysine-coated chamber slides (Lab-tek). The samples were fixed for 15 minutes at room temperature in 4% paraformaldehyde at various intervals after plating. They were stained, mounted and viewed essentially as described above for the sections except that there was no preblocking step, and washes and second antibody dilutions were in PBS.
Vital staining of primordial germ cells for aggregation assays
Primordial germ cells were isolated as described above and the single-cell suspension divided in two. Half the cells were washed into DMEM supplemented as described above. The remainder of the cells were labelled using the PKH26 Red Fluorescent General Cell Linker Kit (Sigma) according to the manufacturers instructions. All the cells treated in this way became labelled. The labelled cells were fractionated by Percoll (Pharmacia) centrifugation to eliminate labelled somatic cells. The labelled cell sample (250 μl) was added to an equal volume of 65% Percoll in PB1 containing 2.6% BSA (Barton et al., 1993) and then laid onto a cushion of 65% Percoll (300 μl) in an Eppendorf tube. 300 μl of 25% and then 200 μl of 20% Percoll were gently applied on top. The sample was then centrifuged at 270 g for 20 minutes. An 800 μl sample, consisting predominantly of somatic cells was removed from the top of the gradient and discarded. The remaining primordial germ cell-enriched fraction (enriched to 25%) was washed in DMEM plus supplements and then mixed back with the unlabelled cells before plating onto a STO cell monolayer. The samples were fixed and stained for SSEA1 as described above.
RESULTS
10.5 d.p.c. germ cells interact with each other to form large networks
A low-power confocal image of a 10.5 d.p.c. mouse embryo fragment stained for SSEA1 is shown in the mid-sagittal plane in Fig. 1A. In this focal plane, the germ cells can be seen concentrated in the dorsal part of the hindgut mesentery although a few germ cells are scattered more ventrally. The gut lumen is also stained by this antibody and it can be seen looping away from the dorsal body wall of the embryo. The targets of germ cell migration, the genital ridges, are not in the plane of focus in this image.
(A) Confocal image of a 10.5 d.p.c. mouse embryo fragment stained for SSEA1 showing the localisation of primordial germ cells. The anterior end of the embryo is to the right-hand side of the image. The ventral body wall of the embryo has been dissected away to reveal the primordial germ cell-containing tissues. Black open triangles, primordial germ cells; white arrow heads, hindgut lumen; m, hindgut mesentery. Scale bar, 500 μm. (B) Four confocal images cells from a 10.5 d.p.c. mouse embryo stained for SSEA1. Each image is separated from the last by 3 μm in the Z axis. In frame 1 three apparently separate cells are visible. By moving 3 μm out of the plane of the page, however (frame 2), it can be seen that cell ‘x’ has a long process with which it associates with cell ‘y’ (frame 3). In frame four it can be seen that cell ‘y’ touches cell ‘z’. Scale bar, 20 μm. (C,D) Confocal images of 10.5 d.p.c. mouse embryos stained for SSEA1, showing two kinds of germ cell process. (C) A lamellipodia-like process with microspikes protruding from its surface (scale bar, 10 μm). (D) Two cells associating with each other via a long filopodia-like process (scale bar, 20 μm).
(A) Confocal image of a 10.5 d.p.c. mouse embryo fragment stained for SSEA1 showing the localisation of primordial germ cells. The anterior end of the embryo is to the right-hand side of the image. The ventral body wall of the embryo has been dissected away to reveal the primordial germ cell-containing tissues. Black open triangles, primordial germ cells; white arrow heads, hindgut lumen; m, hindgut mesentery. Scale bar, 500 μm. (B) Four confocal images cells from a 10.5 d.p.c. mouse embryo stained for SSEA1. Each image is separated from the last by 3 μm in the Z axis. In frame 1 three apparently separate cells are visible. By moving 3 μm out of the plane of the page, however (frame 2), it can be seen that cell ‘x’ has a long process with which it associates with cell ‘y’ (frame 3). In frame four it can be seen that cell ‘y’ touches cell ‘z’. Scale bar, 20 μm. (C,D) Confocal images of 10.5 d.p.c. mouse embryos stained for SSEA1, showing two kinds of germ cell process. (C) A lamellipodia-like process with microspikes protruding from its surface (scale bar, 10 μm). (D) Two cells associating with each other via a long filopodia-like process (scale bar, 20 μm).
We have optically sectioned (0.5 μm sections) labelled embryos at higher magnifications and find that germ cells extend an array of processes from their surfaces. These processes appear to link the germ cells together (Fig. 1B,D).
We have examined the complete surfaces of 70 individual primordial germ cells in the mesenteries of six 10.5 d.p.c. embryos in ‘Z series’ images and have found that 90% (1SE%=3.6) are linked either via side-by-side associations or via fine processes. The processes vary in length from 3 μm (microspike-like, Fig. 1C) to 40 μm (filopodia-like, Fig. 1D) and are of apparently random orientation. We describe germ cells associated in this way as being ‘networked’.
Network formation is established after the germ cells emerge from the hindgut endoderm, and leads to their aggregation in the genital ridges
We have compared the appearance of primordial germ cells in a temporal series of whole mounts beginning at 9 d.p.c. (there are no antibody markers for younger primordial germ cells) in order to determine when the germ cell network is established and what happens after it forms. A confocal image of primordial germ cells in the hindgut endoderm of a 9 d.p.c. mouse embryo is shown in Fig. 2A. At this stage, the hindgut, in which the germ cells are embedded, is next to the dorsal aorta. Although some of the germ cells have an elongate appearance suggestive of motile cells, few associate with each other at this stage. Fig. 2B,C shows slightly later embryos (9.25-9.5 d.p.c.) where we have found germ cells with processes projecting out of the gut endoderm into the surrounding tissue, close to the site where the genital ridges will develop. These ‘pioneer’ germ cells may therefore contact the gonadal region as early as 9 d.p.c. By 9.5 d.p.c., most germ cells have migrated out from the epithelial lining of the gut and it is at this time that networks first become evident (Fig. 2D). We examined complete Z-series images from 243 individual primordial germ cells in nine embryos, and found that 35% (1SE%=3.0) are networked by 9.5 d.p.c. Fig. 3A shows a genital ridge and its associated mesonephros from an 11.5 d.p.c. mouse embryo. The germ cells at this stage are now within the genital ridges. The appearance of these cells is quite distinct from those seen at 10.5 d.p.c. PGCs from 11.5 d.p.c. embryos are present as large aggregates of rounded cells, which associate with maximal contact of their surfaces and show few or no processes (Fig. 3B). Comparison of the 9.5, 10.5, and 11.5 d.p.c. images shows that the germ cell networks seen at 10.5 d.p.c. are part of a sequence of events whereby germ cells emerge from the hind gut separately and come together into closely apposed masses in the genital ridges.
Confocal images from 9 – 9.5 d.p.c. embryos stained for SSEA1 (A) Primordial germ cells in the hindgut endoderm of a 9 d.p.c. mouse embryo. Scale bar, 100 μm (B) A 9.25-9.5 d.p.c. mouse embryo showing primordial germ cells with process projecting out of the hindgut endoderm. Scale bar, 100 μm (C) A 50 μm thick section through a similar aged embryo to that shown in B, with a single primordial germ cell sending a process out of the gut endoderm into a site adjacent to the developing mesonephros. The gut lumen is out of the plane of focus and appears as a blur in this image as does the end of the primordial germ cell process. Scale bar, 40 μm. (D) Confocal image of a 9.5 d.p.c. mouse embryo showing germ cells coming together into a network. The plane of focus is such that the hindgut lumen clearly visible. Scale bar, 40 μm. Open triangles (either black or white) indicate primordial germ cells. Black arrowheads indicate the hindgut lumen. White arrowheads indicate the periphery of the hindgut endoderm. da, dorsal aorta; s, somite, mn, developing mesonephros.
Confocal images from 9 – 9.5 d.p.c. embryos stained for SSEA1 (A) Primordial germ cells in the hindgut endoderm of a 9 d.p.c. mouse embryo. Scale bar, 100 μm (B) A 9.25-9.5 d.p.c. mouse embryo showing primordial germ cells with process projecting out of the hindgut endoderm. Scale bar, 100 μm (C) A 50 μm thick section through a similar aged embryo to that shown in B, with a single primordial germ cell sending a process out of the gut endoderm into a site adjacent to the developing mesonephros. The gut lumen is out of the plane of focus and appears as a blur in this image as does the end of the primordial germ cell process. Scale bar, 40 μm. (D) Confocal image of a 9.5 d.p.c. mouse embryo showing germ cells coming together into a network. The plane of focus is such that the hindgut lumen clearly visible. Scale bar, 40 μm. Open triangles (either black or white) indicate primordial germ cells. Black arrowheads indicate the hindgut lumen. White arrowheads indicate the periphery of the hindgut endoderm. da, dorsal aorta; s, somite, mn, developing mesonephros.
Confocal images from an 11 d.p.c. mouse embryo stained for SSEA1 (A) A genital ridge and associated mesonephros. The primordial germ cells are packed into the tissues of the developing gonad. Scale bar, 100 μm. (B) Rounded up primordial germ cells, which have a non-motile appearance and which directly associate with each other, maximizing their cell surface contacts. They are quite distinct from the primordial germ cells of younger embryos. Scale bar, 10 μm.
Confocal images from an 11 d.p.c. mouse embryo stained for SSEA1 (A) A genital ridge and associated mesonephros. The primordial germ cells are packed into the tissues of the developing gonad. Scale bar, 100 μm. (B) Rounded up primordial germ cells, which have a non-motile appearance and which directly associate with each other, maximizing their cell surface contacts. They are quite distinct from the primordial germ cells of younger embryos. Scale bar, 10 μm.
10.5 d.p.c. germ cells form networks and aggregate in vitro
The interactions between PGCs could occur passively or actively. They are undergoing cell division during migration and so connected cells might be siblings that remain adherent to one another. Indeed, networks could be formed passively by initially adherent siblings being partially pulled apart by the extension of the mesentery, thus forming the networks seen at 10.5 d.p.c. Alternatively, a single PGC may actively extend processes until it encounters another PGC and then selectively adhere to it. To determine the involvement of morphological movements in network formation, we isolated PGCs from 10.5 d.p.c. embryos, disaggregated them mechanically and plated them onto a substratum of a confluent monolayer of irradiated STO fibroblasts. The cultures were fixed and the germ cells stained for SSEA1, 2 hours, 24 hours and 48 hours after plating. Examples of the cell conformations are shown in Fig. 4. We find that during the first 2 hours in culture the PGCs adhere to the STO cell monolayer, flatten out and start to extend processes. After 24 hours in culture, some of them have formed networks. Processes extending up to 50 μm, and indis-tinguishable morphologically from those seen in the embryo, have been observed between PGCs in culture. By 48 hours, however, clusters of at least 8 PGCs are apparent in which they are closely apposed and rounded up, identical in appearance to those found in 11.5 d.p.c. embryos. In a separate experiment, in which the density of the cells seeded was halved so that we could assess the proportion of cells associating with time, we found in four replicate samples that, after 2 hours in culture, 15.1% (1SEM=1.7) of PGCs were networked. After 24 hours, this number increased to 65% (1SEM=1.94) and, by 48 hours, the number of PGCs in networks had risen to 86.25% (1SEM=1.8) in the culture. Thus progressively more PGCs are found in clusters than as single cells attached to STO cells, suggesting that PGCs may preferentially adhere to each other. Clusters of up to 7 cells were observed in this experiment. Such clusters have also been documented previously and shown to increase in size with time (Godin et al., 1991). The fact that clusters of PGCs form in vitro suggests that morphological movements within the embryo are not essential for network formation.
The hindgut mesentery and associated mesonephroi were dissected from embryos at 10.5 d.p.c. After disaggregation, the cells were plated onto a STO cell monolayer at a density of 8 – 10 embryo equivalents per well. At intervals of 2 (A, scale bar, 50 μm), 24 (B-D, scale bars, 10, 10 and 25 μm, respectively) and 48 (E, scale bar, 10 μm) hours, the cells were fixed and stained for SSEA1. Examples of the conformations of the cells found at these times are shown.
The hindgut mesentery and associated mesonephroi were dissected from embryos at 10.5 d.p.c. After disaggregation, the cells were plated onto a STO cell monolayer at a density of 8 – 10 embryo equivalents per well. At intervals of 2 (A, scale bar, 50 μm), 24 (B-D, scale bars, 10, 10 and 25 μm, respectively) and 48 (E, scale bar, 10 μm) hours, the cells were fixed and stained for SSEA1. Examples of the conformations of the cells found at these times are shown.
To distinguish between the roles of mitosis and cell aggregation in cluster formation, we labelled a proportion of the cells with a rhodamine-conjugated dye before plating them onto the STO monolayer. The labelled cells were subjected to a Percoll gradient enrichment procedure to eliminate excess somatic cells, since this makes it difficult to distinguish the labelled from unlabelled germ cells. The cultures were fixed at 24 hours and the germ cells stained for SSEA1. Fig. 5 shows that, in some cases, aggregates of PGCs forming in culture contain both labelled and unlabelled cells, and therefore cannot have arisen by cell division alone.
Germ cell-containing tissue from 10.5 d.p.c. embryos was isolated as described in the legend to Fig. 4. Half the disaggregated cells were labelled with a rhodamine-conjugated dye (PKH26). The cells from the labelled sample were enriched for germ cells and then mixed with the unlabelled sample before plating onto a STO cell monolayer. The cultures were fixed and stained for SSEA1 after 24 hours. Examples of ‘mixed’ clusters are shown. PGC1 is both PKH26-positive (red) and SSEA1-positive (green), PGC2 is PKH26-negative but SSEA1-positive, and PGC3 is an aggregate of both PKH26-positive and -negative PGCs.
Germ cell-containing tissue from 10.5 d.p.c. embryos was isolated as described in the legend to Fig. 4. Half the disaggregated cells were labelled with a rhodamine-conjugated dye (PKH26). The cells from the labelled sample were enriched for germ cells and then mixed with the unlabelled sample before plating onto a STO cell monolayer. The cultures were fixed and stained for SSEA1 after 24 hours. Examples of ‘mixed’ clusters are shown. PGC1 is both PKH26-positive (red) and SSEA1-positive (green), PGC2 is PKH26-negative but SSEA1-positive, and PGC3 is an aggregate of both PKH26-positive and -negative PGCs.
DISCUSSION
The object of this work was to identify structures in the embryo with which primordial germ cells interact, and analyse changes in cell shape during migration using confocal microscopy. The advent of the confocal microscope has enabled three-dimensional analysis to be performed, and with associated real time imaging techniques, analysis of migratory cells in living tissues is now possible in some organisms. Clearly, such experiments rely on the availability of suitable markers and, in this respect, SSEA1 is particularly useful as it is distributed over the entire PGC surface. An obvious and surprising observation is that progressive germ cell-germ cell association occurs during the 2-day period of migration from the hind gut to the genital ridges. As they emigrate from the hindgut, primordial germ cells are not generally associated with each other. 1 day later, however, when the germ cells are in the mesentery, they are connected by long processes. 1 day later still, they are aggregated into clusters of closely apposed cells in the genital ridges. This sequence of events also occurs when 10.5 d.p.c. primordial germ cells are cultured in vitro, suggesting that it is not simply due to activity of the surrounding tissues. It is well documented that primordial germ cells actively divide during their migration (see Wylie and Heasman, 1993 for review). It may be argued, therefore, that clusters of PGCs represent sibling cells. We have demonstrated, however, by a differential staining procedure, that cell aggregation does play a role in germ cell cluster formation.
Another significant observation is that the first primordial germ cells to emigrate from the gut do so when there is no dorsal mesentery. The genital ridges begin to form medial to the mesonephros as small thickenings on the dorsal abdominal wall at 9.5 d.p.c. (see Clark and Eddy, 1975). Thus the first primordial germ cells to emigrate from the hindgut extend processes into a site very close to, or actually in, the developing genital ridge.
Our observations raise two important points concerning primordial germ cell migration, neither of which is apparent by examination of the cells by conventional microscopic methods. Firstly, germ cells do not migrate independently of each other, but instead form an extensive network of connected cells. Secondly, the progression of events seen in vivo and in vitro suggests that PGC-PGC adhesion may play a role in their accumulation in the genital ridges. Previous work on primordial germ cells and other migratory cell types has focused on heterotypic attachments between the migratory cells and the cells and extracellular matrix surrounding them (Donovan et al., 1986; de Felici and Dolci, 1989; ffrench-Constant et al., 1991; Hynes and Lander, 1992). Our work, however, demonstrates the occurrence of interactions between the migratory cells themselves. Does the process of germ cell-germ cell aggregation result in their accumulation in the genital ridges? And if so, how? One explanation is that the primordial germ cells leaving the gut endoderm first, during 9 –9.5 d.p.c., enter directly into the site where the genital ridges will develop and anchor themselves to the target tissue. As later primordial germ cells emerge from the gut, they migrate in the elongating mesentery and extend long processes which attach to other primordial germ cells. Aggregation would then draw the primordial germ cells into the genital ridges.
This model of primordial germ cell migration is supported by their behaviour when isolated during the migratory phase and cultured on feeder cells. The primordial germ cells adhere, spread and move on the surfaces of the embryo fibroblasts. During this period, they associate with each other, such that after 2 days in culture, only a small proportion are found as single cells. The majority are found in aggregates of closely apposed and apparently non-motile cells. Thus, although primordial germ cells stick to somatic cells and use them as a substratum for migration, it appears that they may adhere preferentially to each other.
Furthermore, these observations raise the possibility that germ cell-germ cell contacts (rather than germ cell-somatic cell contacts) play a role in switching off the migratory phenotype. Signalling interactions between PGCs would be allowed by their close apposition, and may play an important role in the migratory and/or proliferative behaviour of these cells.
It will be important to identify the adhesion molecules mediating germ cell-somatic cell adhesion and germ cell-germ cell adhesion. There are a number of candidate molecules. N-cadherin, for instance, has been reported to be present on chicken germ cells (Hatta et al., 1987) but so far cadherins have not been identified on mouse germ cells. Xenopus germ cells express a cell surface glycolipid during the period when they leave the gut, a factor which bears a carbohydrate moiety that has been found to play a role in cell-cell adhesion during the blastula stage (Turner et al., 1992). SSEA1 itself has been implicated in cell-cell adhesion in the early mouse embryo (Bird and Kimber, 1984) and thus may play a role in homotypic germ cell-germ cell adhesion. SSEA1 is involved in compaction at the morula stage, during which cells maximize their contacts forming a tight ball of cells. This process has a close resemblance to that which we see during primordial germ cell migration.
In summary, these observations are incompatible with the view, suggested by their appearance in tissue sections, that PGCs migrate as individuals, and arrive at the genital ridges in turn, like runners finishing a race. Instead, the principal movements seem to involve PGCs interacting with each other, first via long processes, and then by aggregation. It will be important to identify the molecules involved in this homotypic adhesion event, since we would predict them to be critically important in PGC migration.
ACKNOWLEDGEMENTS
We are grateful to the Wellcome Trust for financial support for this work. We also thank Aaron Crawford, Julie Cooke and Colin Sharpe for useful discussions, constructive criticisms and proof reading.