Germ cell-autonomous Wunen2 is required for germline development in Drosophila embryos

In many animals, PGCs are specified and segregated from somatic cells at an early stage in development. PGCs are often formed in regions that are distant from the site where the gonad will form, and subsequently migrate through the embryo towards the future gonad. It has become clear that interactions between PGCs and surrounding somatic cells are essential for the proper migration of PGCs to the gonads.

Drosophila PGCs, which are known as pole cells, are formed at the posterior end of the blastoderm embryo. During gastrulation, pole cells are carried inside of the embryo with the invaginating posterior midgut primordium. Pole cells then actively pass through the midgut epithelium to enter the interior of the embryo, and move dorsally along the midgut surface towards the mesoderm. Next, pole cells migrate from the midgut to the mesoderm, where they make contact with the somatic gonadal precursors (SGPs). Pole cells adhere tightly to the SGPs and finally coalesce into the embryonic gonad (Starz-Gaiano and Lehmann,2001).

Nanos and its co-factor Pumilio (Pum) are well-characterized maternal factors that are essential in a cell-autonomous manner for the proper migration of pole cells. Pole cells lacking Nanos or Pum are able to pass through the midgut. After exiting the midgut, however, they remain clustered on its outer surface, fail to enter the mesoderm and are lost during subsequent development (Asaoka-Taguchi et al., 1999; Forbes and Lehmann,1998; Kobayashi et al.,1996). In addition, Nanos or Pum are essential for transcriptional regulation in pole cells. Pole cells lacking either of them show premature activation of several germline-specific enhancers that are normally expressed at much later stages (Asaoka et al.,1998; Asaoka-Taguchi et al.,1999; Kobayashi et al.,1996), as well as the ectopic expression of genes normally active only in somatic cells (Deshpande et al.,1999). The product of the polar granule component(pgc) gene also regulates transcriptional repression in pole cells. In embryos with reduced pgc function, pole cells ectopically express soma-specific genes such as zerknullt and tailless. In these pole cells, there is a premature phosphorylation of the Ser2 residue in the C-terminal domain (CTD) repeats of RNA polymerase II(Deshpande et al., 2004; Martinho et al., 2004). As this phosphorylation reflects the active transcription state(Dahmus, 1996), it has been suggested that pgc regulates global transcription in pole cells. As with nanos and pum, in embryos with reduced pgcfunction, pole cells fail to migrate into the gonads and die after exiting the midgut (Nakamura et al.,1996).

In addition to these cell-autonomous mechanisms, intercellular communication between pole cells and somatic cells is crucial for pole cell development and migration. For example, trapped in endoderm-1(tre1) acts in the passage of pole cells through the midgut. tre1 encodes a G-protein-coupled receptor that is expressed in pole cells, and is thought to respond to a signal from the midgut to direct trans-epithelial migration (Kunwar et al.,2003). Furthermore, pole cells also receive directional cues from somatic cells that guide their migration to the gonads. Enzymes involved in isoprenoid biosynthesis are essential for attracting pole cells toward the gonads (Santos and Lehmann,2004; Van Doren et al.,1998a). Conversely, zygotic wun and wun2, which encode a putative ectoenzyme, lipid phosphate phosphatase (LPP), are expressed in the ventral region of the midgut, and provide a repulsive environment for pole cell migration (Burnett and Howard,2003; Starz-Gaiano et al.,2001; Zhang et al.,1996; Zhang et al.,1997). In addition to repelling pole cells, Wun and Wun2 activities also affect pole cell survival. Overexpression of either of them in somatic tissues leads to the drastic loss of pole cells during their migration(Burnett and Howard, 2003; Starz-Gaiano et al., 2001). However, it remains elusive how zygotic Wun and Wun2 activities in somatic cells exert their functions on both pole cell migration and survival. We report that wun2 has a maternal function that is required in pole cells for their survival. The wun2 function in pole cells was different from those of nanos, pum or pgc, because transcriptional regulation in early pole cells was intact in the maternal wun2 mutant embryo. Furthermore, we provide evidence that pole cell survival requires a balance of LPP activities in pole cells and somatic cells. These results indicate that Wun2 in pole cells competes with somatic Wun and Wun2 for pole cell survival.

Isogenized w; FRT42B males were treated with 25 mM ethyl methanesulfonate (EMS) (Sigma) for 18-24 hours using standard methods. Mutagenized males were crossed with w; T(2;3)ap^Xa/CyOvirgin females. Single male progeny with the genotype w; FRT42B,*/T(2;3)ap^Xa or CyO (where the asterisk indicates the mutagenized chromosome) were crossed with w;T(2;3)ap^Xa/CyO virgin females to establish balanced stocks. The average frequency of lethal hits per chromosome was 1.2.

To generate germline clones, w; FRT42B, */T(2;3)ap^Xafemales were crossed with y w hs-FLP; FRT42B ovo^D1/CyOmales. The progeny were heat-shocked twice daily at 37° C for 2 hours during the early pupal stage. Embryos from females with the genotype y w hs-FLP; FRT42B, */FRT42B ovo^D1 crossed to wild-type males were collected, fixed and stained for Vasa. Using this strategy, 1156 independent mutagenized lines were screened.

The N14 chromosome was outcrossed to an unmutagenized FRT42B chromosome to recombine away second mutation(s). A lethal mutation on the N14 chromosome was placed between FRT42B and curved by meiotic recombination mapping. N14 was crossed to the deficiency kit for 2R (Bloomington Stock Center) and tested for complementations based on lethality and on the maternal-effect mutant phenotype. During this process, the maternal-effect phenotype of the N14 mutation was found to be separable from the lethal mutation, and was uncovered by five overlapping deficiencies in the chromosome region 45C-D: Df(2R)Np5, Df(2R)w45-30n, Df(2R)w73-1, Df(2R)wun-GL and Df(2R)w45-19g. The breakpoints of these deficiencies were determined by semi-quantitative Southern hybridization using PCR-amplified genomic fragments as the probes.

A series of genomic DNA fragments in the N14 candidate locus was isolated from a Drosophila genomic DNA library in λFIXII (a gift from B. Suter, McGill University) and subcloned into pCaSpeR3. These constructs were introduced into flies by a standard method, and were checked for their ability to rescue the N14 mutant phenotype.

Immunostaining was carried out as described(Kobayashi et al., 1999),except that the embryos were devitellinized by hand in the tracer experiment using the caged fluorescein. The following primary antibodies were used:rabbit and rat anti-Vasa, rabbit anti-Nanos (K.H.-N. and A.N., unpublished),rabbit anti-cleaved caspase 3 (Asp175) (Cell Signaling Technology), mouse anti-RNA polymerase II H5 (Babco) and rabbit anti-β-galactosidase(Cappel). Antibody detection was performed using either a biotinylated secondary antibody followed by the ABC Kit (Vector Laboratory) and DAB staining, or Alexa Fluor 488-, 568-, 594- and 660-conjugated secondary antibodies (Molecular Probes). Fluorescence signals were observed under a laser-scanning confocal microscope (Leica TCS-SP2 AOBS). In situ hybridization with DIG-labeled RNA probes was carried out as described(Kobayashi et al., 1999). For the nanos, gcl or pgc probes, hybridized signals were detected with an alkaline phosphatase-conjugated anti-DIG antibody (Roche). For wun2, hybridized signals were detected with a horseradish peroxidase-conjugated anti-DIG antibody (Roche), amplified with the TSA Biotin System (Perkin Elmer) and visualized with alkaline phosphatase-conjugated streptavidin (Vector Laboratory). To detect β-galactosidase expression,embryos were stained with X-Gal as described(Sano et al., 2001).

A photoactivatable lineage tracer, DMNB-caged fluorescein (Molecular Probes), was injected into early cleavage stage embryos at a concentration of 1 mg/ml. The injected embryos were allowed to develop until the cellular blastoderm stage. To photoactivate the tracer, pole cells were exposed to a two-second pulse of UV irradiation from an epifluorescence microscope fitted with a DAPI optical filter set. Embryos with pole cells that were successfully marked with fluorescein were allowed to develop until the indicated stages. These embryos were then fixed and stained with antibodies.

EP2650 is a viable insertion that is located ∼30 bp upstream of the wun2 locus. Excision lines were generated by crossing virgin EP2650 females with males possessing Δ2-3 transposase. Genomic DNA from the excision lines was prepared and used as PCR templates to screen lines that carried a deletion within the wun2 locus. The breakpoints of the deletion were determined by sequencing the PCR products.

To test the phenotype of wun^m+z–wun2^m–z– mutant embryos, wun2^N14/Df(2R)w45-19g females were crossed to wun^CE/CyO, P{en-lacZ} males. Df(2R)w45-19guncovers both wun and wun2 loci, and wun^CE allele lacks zygotic expression of both wunand wun2 (Starz-Gaiano et al.,2001). Embryos that received a wun^CEchromosome were identified by the loss of β-galactosidase expression. In this experiment, half of the lacZ^– embryos showed a less severe pole cell death phenotype than that observed in the wun2^m–z– embryos. Thus, we scored embryos with weaker pole cell death as well as mis-migrated pole cell phenotypes as wun^m+z– wun2^m–z– embryos. For the overexpression of wun2 both in pole cells and the mesoderm, twist-Gal4/+; nanos-Gal4-VP16/+ females were crossed to EP2650 males. Half of the embryos obtained from this cross were expected to overexpress wun2 in both pole cells and the mesoderm. These embryos were identified by a strong pole cell migration defect, owing to the overexpression of wun2 in the mesoderm.

To identify new maternally acting genes required for pole cell development,we performed a genetic screen for maternal mutations causing defects in this process. We carried out a screen for EMS-induced mutations on the chromosome arm 2R using the FLP/FRT/DFS system, which produces homozygous germline clones in heterozygous mothers (Chou and Perrimon,1996). Embryos derived from homozygous germline clones were stained for Vasa, a marker protein for the germline. From the screening of 1156 independent lines, we obtained one line, which we named N14,that exhibited developmental defects in pole cells during their migration.

Deficiency mapping revealed that N14 was uncovered by Df(2R)w45-19g. In embryos derived from N14/Df(2R)w45-19gmothers (hereafter referred to as N14^m– embryos),normal numbers of pole cells were formed (stage 5, Fig. 1A,B; Table 1). These pole cells were carried into the embryo along with the invagination of the posterior midgut,passed through the midgut epithelium, and migrated dorsally along the surface of the midgut (Fig. 1C,D). However, at stage 11, when pole cells normally associate with the mesoderm,the number of Vasa-positive pole cells was dramatically reduced in N14^m– embryos(Fig. 1E,F). In these embryos,the remaining pole cells, if any, associated with the surface of the midgut(Fig. 1F), and in subsequent development, few or no pole cells were incorporated into the gonads(Fig. 1G,H; Table 1). N14^m– embryos showed no discernible morphological defects in somatic tissues and developed into adults. However, consistent with the loss of pole cells during embryogenesis, over 80% of the adult females developed from N14^m– embryos had agametic ovaries(Table 1). This defect in pole cell development was not rescued by a paternally supplied wild-type copy of the N14 gene, and zygotic N14 mutation did not affect the maternal N14 mutant phenotype. These results indicate that maternal N14 function is required for the maintenance of pole cells during their migration to the gonads.

The above observations allow two possible interpretations: that the pole cells are eliminated after exiting the midgut in N14^m– embryos, or that they survive but lose pole cell-specific markers, such as Vasa. To discriminate these alternative possibilities, we examined the fate of pole cells in N14^m– embryos by using a photoactivatable lineage tracer, caged fluorescein. We injected caged fluorescein into cleavage-stage embryos and photoactivated it in pole cells at the cellular blastoderm stage. We successfully marked the majority of the pole cells, and traced the marked cells until they formed the embryonic gonads in wild-type embryos (data not shown).

In N14^m– embryos, pole cells, which were marked with both fluorescein and anti-Vasa antibody, were observed on the surface of the midgut at stage 10 (Fig. 2A-C). However, at stage 11, the fluorescein-marked cells rapidly disappeared (Fig. 2D-F). We occasionally observed fluorescein-marked pole cells with no or very faint Vasa signals on the surface of the midgut (Fig. 2D-F; arrowheads). At stage 11, the pole cells appeared to have experienced a significant reduction in size accompanied by an increase in the intensity of fluorescein signal (Fig. 2D-F; arrows), and few fluorescein-marked pole cells were detectable by stage 12 (Fig. 2G-I). From these observations, we conclude that pole cells in N14^m– embryos die after they pass through the midgut epithelium.

We next examined whether pole cell death in N14^m–embryos was caused by apoptosis. To detect apoptotic cells, we used an antibody against cleaved caspase 3. It has been shown that an antiserum against the same epitope reacts with the active form of the Drosophila effecter caspase, Drice, and labels dying cells in Drosophila (Brennecke et al.,2003; Yu et al.,2002). At stage 11, pole cells marked by the uncaged fluorescein with reduced Vasa signals were frequently observed in N14^m– embryos. However, active caspase 3 immunoreactivity was never detected in these dying cells(Fig. 3; arrowheads). We did observe a few caspase 3/Vasa double-positive cells(Fig. 3; arrows). Even in wild-type embryos, however, a small number of mis-migrated pole cells, which were usually seen in or on the surface of the midgut, were double-stained for the active caspase 3 and Vasa. The average number of active caspase 3/Vasa double-positive cells per stage 11 embryo was 1.7±2.2 in the N14^m– mutant (n=24) and 2.2±2.4 in wild-type (n=40). These observations indicate that caspase 3-mediated cell death occurs in mis-migrated pole cells even in wild-type embryos, and that it is not sufficient to explain the dramatic increase in pole cell death in N14^m– embryos. These results suggest that pole cell death in N14^m– embryos occurs via a caspase 3-independent pathway.

A reduction in pole cell number during migration is also observed in embryos from females mutant for nanos, pum or pgc(Asaoka-Taguchi et al., 1999; Kobayashi et al., 1996; Nakamura et al., 1996). As these maternally acting genes have roles in earlier events in pole cell development, such as transcriptional repression(Asaoka-Taguchi et al., 1999; Deshpande et al., 1999; Martinho et al., 2004), we examined whether such events are also disrupted in N14^m– embryos.

First, we analyzed the distributions of several maternal RNAs and proteins that localize in pole plasm. At the blastoderm stage, distributions of nanos (Fig. 4A,E), germ cell-less (gcl)(Fig. 4B,F) and pgcRNAs (Fig. 4C,G), and the Nanos protein (Fig. 4D,H) in N14^m– embryos were indistinguishable from those in wild-type embryos, indicating that these pole plasm components are normally incorporated into pole cells in N14^m– embryos.

We next examined whether transcriptional control in early pole cells is disrupted in N14^m– embryos. It has been shown that transcriptional repression in pole cells correlates with the absence of phosphorylation of the Ser2 residue in the CTD repeats of RNA polymerase II(Seydoux and Dunn, 1997). To test global transcriptional activity in pole cells, we stained embryos with the antibody H5, which specifically reacts with the phospho-Ser2 version in the CTD of RNA polymerase II. In wild-type blastoderm embryos, H5 staining was not detected in pole cells, in contrast to the marked staining observed in somatic cells (Fig. 5A). This pattern of H5 staining was not altered in N14^m–embryos (Fig. 5B), indicating that global transcriptional repression in pole cells remains intact in N14^m– embryos.

Finally, we asked whether the N14 mutation affects zygotic gene expression in pole cells. We introduced a nanos-Gal4-VP16 transgene into an N14 mutant background to supply a potent transcriptional activator in pole cells, and analyzed the Gal4-VP16-dependent UAS-lacZ target gene expression in pole cells(Van Doren et al., 1998b). In N14^m– embryos, we did not detect precocious expression of β-galactosidase in early pole cells(Fig. 5C-E). The expression ofβ-galactosidase was normally repressed until stage 8, and was initiated in pole cells from stage 9 onwards (Fig. 5C-E). After stage 12, owing to the pole cell loss in N14^m– embryos, the number of embryos expressingβ-galactosidase was diminished (Fig. 5E). These results indicate that the regulation of zygotic gene expression in pole cells is normal in N14^m– embryos. Thus, the N14 mutation is unlikely to affect early events in pole cells, i.e. the sequestration and maintenance of pole plasm components, global transcriptional repression and the initiation of zygotic gene expression,leading us to conclude that N14 represents a novel class of maternal mutations affecting pole cell survival.

Through complementation tests, we identified five overlapping deficiencies that uncovered the N14 mutant (see Materials and methods). The breakpoints of these deficiencies defined the N14 locus within a∼100 kb genomic region containing nine identified or predicted genes(Fig. 6A). To identify the gene responsible for the N14 mutant phenotype, we generated a series of transgenes containing genomic DNA fragments in this region(Fig. 6A). Of these, a ∼17 kb genomic DNA fragment, which contained two genes, wun2 and CG13955, completely rescued the N14 mutant. To determine which gene was responsible for the N14 mutant phenotype, we next made constructs, in which only one of the genes was intact. Only the 8 kb HincII fragment (Pwun2-8k), containing the entire wun2 locus, rescued the N14 mutant. Furthermore, a genomic fragment that had a partial deletion in the wun2 RNA-coding region(Pwun2-8kΔ) failed to rescue the N14 mutant. Finally,we found that the N14 mutant chromosome had a nonsense mutation in the wun2 gene at the 111th Trp codon(Fig. 6B). Thus, we refer to the N14 mutant as the wun2^N14 allele.

To generate new wun2 alleles, we mobilized a P element insertion, EP2650, which locates ∼30 bp distal to the 5′ side of the wun2 locus. As a result, we isolated a deletion of wun2(wun2^Δ), in which one-third of the wun2locus was deleted (Fig. 6B). The wun2^Δ mutant was homozygous viable, and lacked wun2 RNA expression (see below). In embryos derived from wun2^Δ/Df(2R)w45-19g and wun2^Δ/wun2^N14 mothers, pole cells formed normally, but most of them died after exiting the midgut;phenotypes essentially the same as those of the N14^m– embryos(Table 1). The maternal wun2^Δ mutant phenotype was rescued by introducing the Pwun2-8k transgene (data not shown). These data demonstrate that maternal wun2 is essential for pole cell survival.

wun2 encodes an LPP that dephosphorylates a number of phospholipids in vitro (Burnett and Howard,2003; Starz-Gaiano et al.,2001). We next asked whether the phosphatase activity of Wun2 is responsible for maternal Wun2 function. It has been shown that the putative catalytic residues of Wun2, His274 or His326, are essential for its activity(Starz-Gaiano et al., 2001). We mutated His274 or His326 into Lys (H274K or H326K) and examined the ability of such mutant transgenes to rescue the wun2^N14 phenotype. The two mutant Pwun2-8k transgenes with H247K or H326K failed to rescue the wun2^N14 phenotype (data not shown), indicating that the effect of Wun2 on pole cell survival is dependent on its phosphatase activity.

We next examined the distribution of wun2 RNA during early embryogenesis. wun2 RNA was detected ubiquitously in cleavage-stage embryos (stage 2, Fig. 7A). Although the signal in the somatic cell region became undetectable by the cellular blastoderm stage, it remained at high levels in pole cells (stage 4, Fig. 7B,B′). At stage 5, wun2 signal was also detected in the somatic region in a posterior stripe pattern (Fig. 7C). These observations were consistent with a previous report(Renault et al., 2002). In embryos from wun2^Δ/Df(2R)w45-19g females crossed to wild-type males, wun2 RNA was undetectable in pole cells,but was expressed in the posterior stripe(Fig. 7D). These results indicate that wun2 RNA in pole cells is supplied maternally, while it is expressed zygotically in the posterior stripe.

To examine whether wun2 activity is required in pole cells for their survival, we tested whether the pole cell-specific zygotic expression of wun2 rescues the maternal wun2 mutant phenotype. For this purpose, wun2^N14 mutant females carrying the nanos-Gal4-VP16 transgene were crossed with males possessing EP2650, which expresses wun2 under UAS control(Starz-Gaiano et al., 2001). We found that, in embryos resulting from this cross, pole cells escaped from cell death (Fig. 8D-F). Although these pole cells were moderately dispersed, as in wun2⁺ embryos that overexpressed wun2 in pole cells (Fig. 8G-I), they were incorporated into the gonads. These results indicate that the expression of wun2 in pole cells is sufficient to rescue pole cell death in wun2^m– embryos, and that wun2 is required in pole cells for their survival.

It has been shown that zygotic wun2 acts redundantly with wun in somatic tissues to guide pole cell migration. In embryos that are double mutants for wun and wun2, pole cells are dispersed on the midgut at stage 10, and fail to migrate to the mesoderm(Starz-Gaiano et al., 2001; Zhang et al., 1996; Zhang et al., 1997). Interestingly, somatic Wun and Wun2 are also capable of affecting pole cell survival. Overexpression of either of them in somatic tissues causes dramatic pole cell loss after stage 11 (Burnett and Howard, 2003; Starz-Gaiano et al., 2001). As this latter phenotype is very similar to that observed in wun2^m– embryos, we asked whether there are any functional links between maternal wun2 and zygotic wun and wun2 in regard to pole cell survival.

First, we examined the effect of the loss of both zygotic wun and wun2 on pole cell death caused by a maternal wun2 mutation. Embryos lacking both maternal wun2 and zygotic wun and wun2 (referred to as wun^m+z–wun2^m–z– embryos) showed a partial rescue of the pole cell death phenotype of the maternal wun2 mutation(Fig. 9A,B). In wun^m+z– wun2^m–z– embryos, an average of 10.5±3.5 pole cells were survived at stage 14-15(n=28), although they failed to migrate toward the gonads as in the wun^m+z– wun2^m+z– embryos. This result revealed that maternal wun2 interacts genetically with zygotic wun and wun2 in regulating pole cell survival.

We next examined whether the overexpression of wun2 in pole cells affects the pole cell loss phenotype caused by the overexpression of wun2 in somatic cells. We crossed females carrying both the nanos-Gal4-VP16 transgene and a mesoderm driver twist-Gal4with males possessing EP2650 to overexpress wun2 both in pole cells and the mesoderm. The embryos that overexpressed wun2 both in pole cells and the mesoderm had an average of 10.4±4.0 pole cells(n=64), while embryos overexpressing wun2 in the mesoderm alone had an average of 4.4±3.6 pole cells (n=99) per embryo at stage 14-15 (Fig. 9C,D). This result revealed that the overexpression of wun2 in pole cells suppressed the pole cell loss caused by the overexpression of wun2 in somatic cells. These observations show that pole cell survival requires a balance between LPP activities in pole cells and somatic cells. This further suggests that Wun2 activity in pole cells competes with the Wun and Wun2 activities in somatic cells to promote pole cell survival.

The two related LPPs, Wun and Wun2, act redundantly in somatic cells to provide a repulsive environment to steer pole cell migration(Starz-Gaiano et al., 2001; Zhang et al., 1996; Zhang et al., 1997). The directional migration of pole cells is disrupted only when the expression of both wun and wun2 is lost in somatic cells. Wun and Wun2 also influence pole cell viability, because the overexpression of either of them in somatic cells causes a severe loss of pole cells(Burnett and Howard, 2003; Starz-Gaiano et al., 2001). These observations had led to the proposal that Wun and Wun2 were indistinguishable in their somatic function, which is required for pole cell migration and survival. However, our results show that Wun and Wun2 have separable functions, and that Wun2 plays a novel cell-autonomous role in pole cell survival. Furthermore, we have shown that a balance between the LPP activities in pole cells and the surrounding somatic cells is crucial for pole cell survival. We propose that this balance is also involved in the directional migration of pole cells.

Several maternal mutations that affect pole cell survival have been isolated. These include nanos, pum and pgc(Asaoka-Taguchi et al., 1999; Forbes and Lehmann, 1998; Kobayashi et al., 1996; Nakamura et al., 1996). However, our results indicate that the phenotypes of maternal wun2mutant embryos are different from those of nanos, pum or pgcmutant embryos. For example, although nanos, pum and pgc are required for transcriptional repression in early pole cells(Asaoka et al., 1998; Asaoka-Taguchi et al., 1999; Deshpande et al., 2004; Deshpande et al., 1999; Martinho et al., 2004),maternal wun2 plays no role in either transcriptional repression or in the onset of zygotic gene expression in pole cells(Fig. 5). Thus, it is likely that maternal wun2 is required for pole cell survival at a different developmental step from nanos, pum and pgc.

Surprisingly, dying pole cells in wun2^m– embryos were negative for cleaved caspase 3 (Fig. 3). We also found that the overexpression of the caspase inhibitor p35 (Hay et al., 1994) in pole cells did not rescue the pole cell death phenotype in wun2^m– embryos (data not shown). Furthermore, we have never detected TUNEL-positive pole cells in wun2^m– embryos (data not shown). Thus, the pole cell death in wun2^m– embryos seems to occur via a mechanism different from typical caspase-dependent apoptosis. It has become evident that caspase-independent cell death pathways do exist(Golstein et al., 2003; Leist and Jäättela,2001; Lockshin and Zakeri,2002), and we suppose that such caspase-independent cell death might be occurring in these pole cells. Further morphological study may reveal how pole cell death occurs in wun2^m– embryos.

Wun2 belongs to a conserved family of LPPs(Burnett and Howard, 2003; Starz-Gaiano et al., 2001). LPPs are integral membrane proteins that dephosphorylate a number of bioactive lipid phosphates in vitro, such as lysophosphatidic acid, phosphatidic acid,sphingosine-1-phosphate and ceramide-1-phosphate. These lipid phosphates act as extracellular signaling molecules and/or intracellular second messengers and affect a variety of cellular processes, including cell survival and motility (Brindley et al.,2002; Sciorra and Morris,2002). LPPs can attenuate cell activation by dephosphorylating bioactive lipid phosphates and/or they can generate alternative signals from dephosphorylated lipids, such as diacylglycerol, sphingosine and ceramide. The active sites of LPPs are exposed either on the outer surface of the plasma membrane or on the luminal surface of intracellular organelles, depending on their subcellular localization (Brindley et al., 2002; Sciorra and Morris,2002). It has been proposed that LPPs promote the incorporation of lipid phosphate substrates into the outer leaflet of the membrane before their dephosphorylation (Roberts and Morris,2000). Although the exact subcellular distribution of the endogenous Wun2 protein remains elusive, it localizes to the plasma membrane in Drosophila embryos when overexpressed(Starz-Gaiano et al., 2001). Therefore, Wun2 is likely to be an ecto-enzyme that promotes the uptake and the dephosphorylation of extracellular lipid phosphate substrates.

Our data indicate that maternally supplied Wun2 acts in a cell-autonomous manner to promote the survival of pole cells (Figs 7, 8). By contrast, zygotically expressed Wun and Wun2 act in somatic cells to direct pole cell migration(Starz-Gaiano et al., 2001; Zhang et al., 1996; Zhang et al., 1997). Furthermore, we have shown that pole cell survival requires a balance between LPP activities in pole cells and somatic cells(Fig. 9). Considering that Wun and Wun2 are likely to function as ecto-enzymes(Burnett and Howard, 2003; Starz-Gaiano et al., 2001),the same extracellular lipid phosphate could be degraded by both Wun2 in pole cells, and Wun and Wun2 in somatic cells. Thus, we propose that Wun2 activity in pole cells competes with somatic Wun and Wun2 activities for the uptake and the dephosphorylation of a common substrate. When Wun or Wun2 is overexpressed in somatic cells, the extracellular substrate would be depleted in the hemocoel surrounding the pole cells, so that Wun2 in pole cells is unable to produce the survival signal any longer. In embryos overexpressing Wun2 in both somatic cells and pole cells, increased Wun2 activity in pole cells leads to the increased incorporation of the substrate, promoting pole cell survival. In wun^m+z– wun2^m–z– embryos,pole cells escaped from cell death (Fig. 9B), suggesting that pole cells are capable of producing the survival signal even in the absence of pole cell-autonomous Wun2 activity. There are eight LPP genes, including wun and wun2, in the Drosophila genome; wun mRNA is also maternally supplied in cleavage-stage embryos (Renault et al.,2002). Although wun mRNA does not become concentrated in pole cells, it is conceivable that trace amounts of Wun and/or other LPP become partitioned into pole cells, promoting their survival in the wun^m+z– wun2^m–z– embryo. However, in normal embryos, such activity would have only a subtle effect on pole cell survival, because it would promote the survival of only a small number of pole cells.

Somatic LPP activity also functions to repel pole cells. An extracellular substrate might direct somatic cells to produce a repellant molecule, while directing pole cells to produce a distinct survival signal. However, based on our finding that LPP activity in somatic cells competes with that in pole cells for an extracellular substrate, we favor the idea that LPP activity in somatic cells provides a repulsive environment that directs pole cell migration by depleting a substrate that is required by pole cells for their survival. We expect that similar LPP-mediated mechanisms of cell migration and survival may be widely used, as LPPs play important roles in various developmental processes such as axon growth, axis patterning and extra-embryonic vasculogenesis in mammals(Bräuer et al., 2003; Escalante-Alcalde et al.,2003). Future work will focus on identifying the endogenous substrate for Wun2 and resolving the mechanism by which Wun2 exerts its effects on pole cell survival.

We thank K. Howard, R. Lehmann, P. Rørth, B. Suter, Exelixis, and the Bloomington Drosophila Stock Center for fly stocks and reagents;and H. Sano for help during the screening. We are also grateful to D. Sipp and N. Fuse for helpful discussions and comments on the manuscript. This work was supported in part by a Grant-in-Aid from the Ministry of Education, Culture,Sports, Science and Technology, Japan, and a CREST project grant from the Japan Science and Technology Agency.