ABSTRACT
Cell migration is a common feature of development. In order to understand more about the factors that controf these movements we have embarked on further analysis of the migration of Drosophila germ cells. This process involves passage of the germ cells across the gut pri mordium and migration toward the mesoderm where the somatic gonad forms. We are particularly interested in the early phase of this migration when the germ cells interact with the amnioproctodeal invagination, the developing gut, before entering into association with the mesoderm. We will summarize the results of our and other studies of these events before describing a number of enhancer trap lines which show expression in the amnioproctodeal invagination during the early phase of germ cell migration. These reveal more about the com plexity of this tissue and suggest this tissue is capable of guiding the early phase of germ cell migration.
INTRODUCTION
Drosophila melanogaseter germ cell migration was first studied in some detail by Sonnenblick (1941) and Rabi nowitz (1941). These studies and those of later workers, par ticularly Poulson. Mahowald, and Technau (Illmensee and Mahowald, 1974; Poulson and Waterhouse, 1960; Technau and Campos-Ortega, 1986) described the migratory route and behavior of the germ cells as they arrive at the gonad. The understanding of the nature of the processes around the mid 1980s is well summarized in (Campos-Ortega and Hartenstein, 1985). We are interested in further analysis of the events of early migration because this is a situation in which a directed movement of cells during development can be analyzed using both experimental manipulation and genetic approaches.
A related matter, the determination and fate of the pole cells has been the subject of considerable study. The formation of the pole cells is directed by the germ plasm, a Jeterminant localized at the posterior pole of the embryo. A number of components of this plasm have been identified for review see Lasko, 1992). This factor directs these cells to enter the germ line. This seems to be the only fate available to these cells (Technau and Campos-Ortega, 1986). The mode of action of the determinant, both in the formation of morphological pole cells and in the determi aation of the germ line per se is not well understood. Only recently has Kobayashi produced data suggesting that a particular splice is made productively only in those pole cells that enter the germ line (Kobayashi et al., 1993).
The early events of germ cell formation and migration are a.mooned in Fig. 1. The germ cells form at the posterior pole of the egg before cellularization of the blastoderm in response to the posteriorly localized determinant which performs two functions: specifying germ cells and setting up the pattern of the posterior abdominal region of the embryo (reviewed in St-Johnston and Nilsslein-Volhard, 1992). As the blastoderm layer forms, the cells that lie under the germ cells are assigned an endodermal fate and will eventually go on the form the midgut. These fates are determined by the torso signaling system and are independent of the posteri orly localized determinant. At the same time cells along the ventral mid-line are assigned a mesodermal fate by the dorsal signaling system (reviewed in Rushlow and Arora, 1990). Some of these cells will eventually go on to form the somatic tissues of the gonad.
The geometry of the cellular blastoderm is transformed by the events of gastrulation and germ band extension. The mesodermal cells form the ventral furrow as they enter the embryo, before coming to lie laterally in the interior. At about the same time the cells at the posterior pole begin to change shape and form the polar plate. This invagination moves forward and deepens during the fast phase of germ band elongation becoming the amnioproctodeal invagina tion, the primordium of the gut. This structure then elongates both by growth of the cells and by recruitment of ectoder mal cells into the hindgut primordium (see Skaer, I 993 for a review). During these processes the germ cells enter the amnioproctodeal invagination and come to lie at the blind posterior end of this pocket.
During late stage 10 of development the germ cells pass across the endoderm and begin to migrate around on the surface of this tissue, moving between it and the ensheath-ing yolk sac. By the end of this phase the majority of germ cells lie on the ventral side of the germ band close to the mesoderm. As the germ band begins to retract the germ cells migrate from the endoderm into the mesoderm where they associate with the gonadal mesoderm. The cells that form this mesoderm arc distinguished by their expression of the transposon 412 (Brookman et al., 1992). It is particularly interesting to note that this structure assembles normally even in the absence of germ cells.
During this migration the behavior of the germ cells is regulated in (wo generally important ways. First, the time of onset of migration is controlled so that cell movement begins at a particular stage in development. Second, this movement is directed so that the cells move from inside the gut primordium to their eventual site of differentiation (the somatic gonad). In order to understand more about these behaviors of the germ cells we have taken three distinct approaches: (I) real lime visualization of germ cells during migration. (2) analysis of germ cell migration in well characterized mutants altering the pattern of the embryo and (3) screening the genome for mutants specifically altering the behavior of germ cells during their migration.
The first of these sets of experiments has involved labeling of host germ cells with nuclear-tagged rhodamine dextran and transplantation of fluorescein-labeled cells into the amnioproctodeal invagination of these labeled embryos (Jaglarz and Howard. 1994). In this way we can visualize both host and transplanted cells using an automated low light video microscope. We have transplanted blastoderm cells and find that they do not move from the amnioproctodeal invagination with the germ cells of the host embryo. Instead they remain in the gut whilst transplanted germ cells migrate along with the host cells (jaglarz. and Howard. 1994). This experiment shows that the initial migration is not simply the consequence of the physical displacement of the contents of the gut. Instead it seems that there arc specific interactions between the germ cells and their environment, the developing gut, which initiate the migration.
In keeping with this suggestion is the fact that even in the absence of the mesoderm, which will eventually form the somatic gonad, the migration of the germ cells begins. Remarkably not only is the migration initiated, but it is correctly oriented toward dial side of the gut which would normally face the mesoderm (Jaglarz and Howard. 1994). Although not demonstrating conclusively that signals in the developing gut both initiate the migration of germ cells from the gut and guide these cells toward the mesoderm, these data arc provocative and suggest that this explanation of the early events of germ cell migration is plausible. Consequently we are particularly interested in the developing amnioproctodcal invagination.
In particular we are searching for genes that arc expressed here al the correct stage of development. Mutations in these genes could conceivably produce several different phenotypes. First, they may result in no defects in cither gut development or germ cell migration. This could occur cither if the gene’s function is served by other complementary genes expressed in the same way. or if the expression is unrelated to any function, and is in some sense frivolous. Secondly, mutations in these genes may result in severe disruption of the development of the gut. l or instance all the cells of the gut may die. In this case it will not be surprising il the migration of the germ cells is perturbed. A third sort of outcome may be that the development of the gut itself is perturbed but that the migration of the germ cells is unaffected. This could be for a number of reasons, for instance the gene may act in gut development after the germ cells have left it and entered the mesoderm. A fourth outcome, and the one that we are most interested in. would be disturbance of the migration of the germ cells without any substantial change in the structure of the gut. These might identify signals emanating from the gut and guiding germ cells toward the mesoderm.
If such a mutation were to exist it would be relatively difficult to detect al the stage when germ cells are moving toward the mesoderm on the amnioproctodcal invagination since even in wild type embryos there is a stage in the migration when the germ cells are disposed relatively chaotically on the surface of the amnioproctodcal invagination. The expected phenotype would be persistence of this chaotic form beyond the appropriate stage in development in an embryo which was otherwise morphologically normal. Unfortunately the exit and migration toward mesoderm occur over a relatively short period of development, at the extended germ band stage, and there are no facile morphological markers which enable us to assign a particular specimen to a particular stage within this period. Consequently our initial analysis has focused on a much later stage of development (shortened germ band) when any errors in germ cell migration arc fairly clear since by this time most of the germ cells are normally found clustered in the gonad at the level of the fifth abdominal segment. A mutation that randomized the early migration on the gut because it interfered with the putative signals from gut to germ cells but did not disturb migration in the mesoderm or assembly of the somatic gonad would result in an embryo which would still form a gonad but which would have an unusually large number of scattered germ cells.
In order to find genes involved in producing signals in the gut we have chosen to look at enhancer trap lines which are expressed there. A number of labs have done large enhancer trap screens and arc making their material available. We have been fortunate enough to have the help of Laura Mathies and Matthew Scott who are studying gut development and had identified a number of enhancer trap lines expressed in the gut. They were kind enough to pass on information about these lines to us. These lines were generated both in screens done in the Scott lab and in the Spradling lab.
RESULTS AND DISCUSSION
Here we report results from IO of these lines which arc listed in Table 1. We documented the expression pattern of beta- galactosida.se in a collection of embryos staged between 5 and 9 hours and on the distribution of germ cells in later embryos in a staged collection from 12–15 hours.
As expected all of these lines show expression of beta- galactosidasc in the amnioproctodeal invagination (Figs 2 and 3). The intensity of this staining is very variable. The weakest signal is seen in the line 1(3)2093 (Fig. 2A) where it is only barely detectable as the amnioproctodeal invagination forms. With the possible exception of 1(3)2093 all the lines show expression of beta-gal in the stomodeal invagination as well as the amnioproctodeal invagination. This is entirely consistent with the homology between (he terminal regions of the embryo demonstrated genetically by the existence of the terminal class of mutants (St-Johnston and Nüsslein-Volhard. 1992). Interestingly, each of these lines show spatially modulated expression within the amnioproctodeal invagination revealing differences along both the proximodistal and dorsoventral axes. For example 1(2)5315 is expressed preferentially in the dorsal side of distal part of the amnioproctodeal imagination (Fig. 2B) whereas 1(2)6430 is expresses in the more anterior part of the amnio-proctodeal invagination around the primordia for the Malpighian tubules (Fig. 2F).
We have not attempted a complete analysis of the dynamics of each of these expression patterns. However, we note that each pattern develops in a characteristic way and show two stages in the development of the expression of 1(2)0255 (Fig. 2D.H). This expression begins at the distal end of the amnioproctodcal invagination (Fig. 211) before developing into a more uniform pattern (Fig. 2D). In addition to showing expression in the amnioproctodcal invagination and the stomodcai invagination three of these lines show marked expression in other parts of the embryo (Fig. 3) suggesting cither that these enhancer traps arc inserted near regulatory elements for two different tissuespecific functions or that the function being regulated by the enhancer(s) is not specific to the terminal tissues.
These data indicate that at the lime that the germ cells leave the amnioproctodcal invagination it is well enough differentiated to be capable of providing signals both to initiate and guide the migration of the germ cells.
Il was of great interest to determine what, if any. consequence for the behavior of germ cells these enhancer trap inserts had. The slocks were not marked so that we could distinguish homozygous enhancer insert embryos from their sibs. We therefore examined >100 embryos from each later collection looking for evident defect in gonad assembly. Apart from rare defects in development which are probably due to random factors such as aneuploidy due to the balancer chromosomes, we saw only relatively normal embryos al this late stage in development (12-15 hours) in 9 of the 10 lines examined. An example of this morphology is shown in Fig. 4A. We conclude that these inserts have not disrupted functions necessary for normal germ cell migration. In contrast, another line 1(3)2093 showed defects in gonad assembly Fig. 4B. In these mutant embryos clusters of germ cells do form al about the normal site of gonad assembly (arrow’ in Fig. 4B). However, most of the germ cells arc dispersed in the posterior 1/3 of the interior of the embryo. These embryos arc relatively normal in appearance and form segments, secrete cuticle and assemble a nerve chord.
This phenotype can be seen as the result of some early disturbance of germ cell migration with only some of those cells which fortuitously enter the appropriate part of the mesoderm thereafter finding the somatic gonad. There are two extreme ways in which this disturbance in germ cell migration could be generated: the developing gut may have degenerated relatively early in its development thereby randomizing the germ cell movements, or alternatively, the gut may develop perfectly normally except that it lacks the ability to signal position to the germ cells causing them to migrate randomly al this stage. We favor the first sort of interpretation since we have never observed a properly formed gut in our collections. The beta-galactosidase staining of these embryos is relatively weak, just barely above background (Fig. 2A) and is consequently not very provocative. In addition the chromosome is lethal, consistent with an extreme failure in gut formation. This phenotype is more extreme than that seen in the well characterized mutant fork head which prevents proper development of the midgut (Weigel. 1989). In this case the majority of the germ cells leave the amnioproctodcal invagination and enter the developing gonad undisturbed (Warrior and Howard, unpublished data). This suggests that the gene disrupted in this enhancer trap line may act early in the differentiation of the gut. Il is interesting to note that in early (3- 5 hour) collections from this stock a number of embryos were observed in which the majority of the germ cells have not yet migrated to the mesoderm and appeared randomly distributed on the surface of the amnioproctodcal invagination. This observation is provocative. However, more work is necessary characterizing the development of these mutants using molecular markers for gut development and stocks where the mutant embryos can be unambiguously distinguished, to define further its role both in gut development and to understand the reasons for the dispersion of germ cells seen in the mutant embryos.
ACKNOWLEDGEMENTS
We would like to thank Laura Mathies. Malt Scott. Steve Beck endorph and Alan Spradling for information and fly stocks. Tony Brown and Phil Ingham for their patience and Chris Rushlow, Tom Gridley, Colin Stewart and an anonymous referee lor helpful comments on the manuscript.