The maternal gene nanos has a central role in posterior pattern formation of the Drosophila embryo

Three groups of maternal effect genes, the anterior, posterior and terminal groups, determine the longitudinal pattern of the Drosophila embryo (reviewed in Nüsslein-Volhard et al. 1987). The activities of the anterior and posterior group of genes are required for the establishment of two morphogenetic centers localized at the anterior and posterior poles of the egg. Both centers have been defined by transplantation experiments as well as by the phenotypes of maternal genes that alter or abolish the function of these centers (Frohnhdfer and Nüsslein-Volhard, 1986; Lehmann and Nüsslein-Volhard, 1986). The anterior center controls head and thoracic development, and the posterior center is required for abdominal development. The terminal group of genes is required for the development of the unsegmented terminal regions of the embryo, the acron and telson, which develop independently of anterior and posterior centers (Klingler et al. 1988).

At least nine maternal effect genes belong to the posterior group of genes and mutations in these genes result in very similar if not identical phenotypes. Embryos from females mutant for any one of these posterior group genes do not form abdomens (Fig. IB). The common phenotype suggests that each of the posterior group genes encodes an essential component of one pattern-forming process. However, on the basis of additional phenotypic traits, we can distinguish three classes within the posterior group of genes: The mutant phenotypes of pumilio (pum, Lehmann and Nüsslein-Volhard, 1987) and nanos (nos, Nüsslein-Volhard et al. 1987 and below) are restricted to the abdominal region. In contrast, mutations in five genes, oskar (osk, Lehmann and Nüsslein-Volhard, 1986), vasa (vas, Schüpbach and Wieschaus, 1986a), valois (vis, Schüp-bach and Wieschaus, 1986a), staufen (stau, Schüpbach and Wieschaus, 1986a) and tudor (tud, Boswell and Mahowald, 1985) share an additional characteristic previously described as the ‘grandchildless’ phenotype. These mutants fail to form pole cells, the germ cell precursors, due to lack of polar granules that are normally found in the posterior pole plasm (Mahowald, 1962). The dual phenotype suggests that pole plasm formation and abdomen determination have common developmental requirements. Finally, cappuccino (capu, Manseau and Schüpbach, 1989) and spire (spir, Manseau and Schüpbach, 1989) share the abdominal and the pole plasm phenotype with the abovementioned genes, but in addition capu and spir mutants also affect chorion shape and dorsoventral pattern, which implies a more general role of these two genes in the establishment of egg cell polarity (Manseau and Schüpbach, 1989).

A functional link between posterior pole plasm and abdominal segmentation was first demonstrated by transplantation experiments involving the genes osk andpum (Lehmann and Nüsslein-Volhard, 1986, 1987). It has been shown that wild-type pole plasm harbors an activity that can rescue the abdominal phenotype of both osk and pum. The site of localization of rescuing activity (the posterior pole) is not congruent with the target site (the abdomen), but separated by the telson, which is not affected in posterior group mutants. The posterior pole thus serves as a source of a morphogenetic signal which is required at a distance for the segmentation of the abdominal region. Since rescuing activity can not be recovered from the prospective abdominal region, we have proposed that the activity might be stored in the form of RNA at the posterior pole, while the spreading protein product is either not sufficiently abundant and/or not accessible for transplantation (Lehmann and Nüsslein-Volhard, 1986).

In this study, we have used genetic and cytoplasmic transplantation experiments in order to establish functional relationships between the different posterior group genes. The common phenotype that results from mutation of the posterior group genes suggests that all nine genes have an indispensable function for abdomen formation. This phenotypic similarity, however, does not allow any conclusion about the step each gene affects within the pathway. We have investigated the properties of seven posterior group genes: nos, pum, osk, vas, vis, tud, and stau. The posterior group genes capu and spir have been described elsewhere (Manseau and Schüpbach, 1989) and were not included in this study. In the first set of experiments, we use cytoplasmic transfer to show that all posterior group mutants affect the same abdominal signal. In the second set of experiments, we use double mutants to show that some of the posterior group genes regulate the distribution of the posterior signal. The third set of experiments shows that the gene nanos is the only candidate to encode the posterior signal.

New alleles of stau and vas were induced by 35 HIM EMS on a chromosome marked with b pr cn (for markers see Lindsley and Grell, 1968). New mutations were identified on the basis of non-complementation in trans to a chromosome carrying the mutations stau^HL and vas^PD. New alleles of pum, osk and nos were induced by 35 mM EMS in a chromosome marked with st e. New mutations were identified on the basis of noncomplementation in trans to a chromosome carrying an inversion (In(3R)Msc), which breaks in pum, a deficiency (Df(3R)pXM66), which uncovers osk, and a point mutation for nos (nos^L7).

The first nos allele (nos^L7) was isolated fortuitously by K. Tietze in the laboratory of Campos-Ortega. In subsequent screens, we isolated three new nos alleles on the basis of noncomplementation of the original allele: nos^RD and nos ^RC are strong, is weak (see Lehmann, 1988). The alleles nos¹¹¹⁷, nos^RC and nos^7iw correspond to the alleles nos¹³, nos¹⁸ and nos¹⁷, respectively in Tearle and Nüsslein-Volhard (1987). Two alleles (nos^RC and nos⁸⁰’) in transheterozygous combinations as well as in trans to a deficiency for this region affect oogenesis. All nos alleles are viable over representative lethal mutants obtained in saturation screens of a small deficiencies including the nos locus (Vassin and Campos-Ortega, 1987; Alton et al. 1988). Cytologically, nos maps proximal of DI in chromosome band 91F13. It is uncovered by Df(3R)Dl-A143 and Df(3R)Dl-X43 but not by Df(3R)Dl-12, Df(3R)Dl-KX12 and Df(3R)Dl-HD23 (Alton et al. 1988; Vassin and Campos-Ortega, 1987). By meiotic recombination nos maps proximal to the lethal complementation group l(Dl-X43) C3=91F-B. nos complements one representative allele of each of the lethal complementation-groups within the DI region identified by the laboratories of M. Muskavitch (Alton et al. 1988) and J. Campos-Ortega (Vassin and Campos-Ortega, 1987).

The phenotype and properties of pum have been described in detail (Lehmann and Nüsslein-Volhard, 1987). We isolated a weak pum allele, pum²¹ and semi-lethal mutations which fail to complement the maternal pum phenotype, suggesting a requirement for pum at later stages of development.

The phenotype and properties of osk have been described in detail (Lehmann and Nüsslein-Volhard, 1986). All new osk alleles, osk⁵⁴, osk⁸⁸, osk¹²³, osk¹⁵⁰ exhibit the strong abdominal phenotype as described in Lehmann and Nüsslein-Volhard, 1986. We also isolated lethals in the osk region, all of which complement osk, suggesting that the osk gene cannot mutate to lethality. This finding supports the notion that osk is highly specific for the process of pole plasm formation and abdominal development.

The original allele, stau^HL shows a variable abdominal segmentation phenotype (Schüpbach and Wieschaus, 1986a). We isolated three additional alleles: stau^D3 (strong), stau^G2 (medium strength similar to stau^HL, Schüpbach and Wieschaus, 1986a), and stau^C8 (temperature-sensitive). The strongest allele, stau^D3, exhibits a very strong, fully penetrant abdominal segmentation phenotype. stau^C8 behaves like a partial lack-of-function (hypomorphic) allele in trans to strong stau alleles. The stau gene is unique among the maternal genes affecting anteroposterior pattern in that it affects both the anterior and the posterior center.

The original allele of vasa, vas^PD, shows the abdominal and pole plasm phenotype, when homozygous and in trans to a deficiency (Schüpbach and Wieschaus, 1986a). We isolated additional vas alleles on the basis of non-complementation of vas^PD: vas^D1, vas^q6, vas^q7, vas^D5 (strong alleles); vas^O11 (medium strength similar to vas^PD, Schüpbach and Wieschaus, 1986); vas^O14 (temperature-sensitive), vas^O14 behaves like a partial lack-of-function (hypomorphic) allele in trans to strong vas alleles. The four strong alleles result in nearly complete female sterility. Homozygous females or females heterozygous for such an allele and a deficiency for the vas locus produce few eggs. These eggs are abnormally shaped, lack the dorsal appendages and frequently display a micropyle at both the anterior and the posterior ends of the chorion. Ovary preparations reveal that many of the egg chambers stop developing in early stages and, in some cases, vitellogenesis is not initiated. Transheterozygous deficiencies for the vasa gene produce a very similar phenotype; however, fewer egg chambers are formed. Two weaker alleles, vas^PD and vas^O11, are normal during oogenesis but affect embryogenesis. In addition to the pole cell and abdominal phenotype, these alleles affect cellularization of the embryo, such that only 30-50 % of all fertilized embryos develop past the blastoderm stage. These observations suggest that vas, in addition to its role in establishing the posterior center, has functions early in oogenesis and embryogenesis. The vas gene has been cloned; vas RNA is transcribed during oogenesis and is uniformly distributed in the nurse cells and the oocyte (Lasko and Ashburner, 1988; Hay et al. 1988b). The vas protein product, on the other hand, becomes concentrated to the posterior pole plasm during oogenesis and is restricted to the pole cells at blastoderm stage (Hay et al. 1988a; Lasko and Ashburner, 1990; Hay et al. 1990).

The phenotype and properties of vis have been described in detail (Schüpbach and Wieschaus, 1986a, 1989). vis embryos frequently show early abnormalities and many of them do not develop beyond the cellular blastoderm stage. The embryos stop development in early cleavage stages or form blastoderms displaying an irregular pattern of cellularization. As the frequency of such embryos is enhanced in progeny of females heterozygous for a vis allele and a deficiency for the locus, it is possible that in an amorphic condition the embryos all would stop development at an early stage. This suggests a more general function of vis during the early stages of embryogenesis.

The phenotype and properties of tud have been described in detail by Boswell and Mahowald (1985) and Schüpbach and Wieschaus (1986a). All tudor alleles described show a variable and weak abdominal phenotype. So far it is unclear whether the weak effect of tud mutations on abdominal segmentation is a locus-specific trait or whether it indicates that the alleles identified are weak, hypomorphic alleles. It is interesting to note that tud is the only gene that affects the size of polar granules in a dose-sensitive manner (Boswell and Mahowald, 1986).

The phenotypes of capu and spir alleles are quite pleiotropic and have been described in detail by Manseau and Schüpbach (1989). Strong alleles affect the egg shape, leading to dorsalization of the embryo. In addition these embryos lack abdominal segments and pole cells. Weaker alleles predominantly show the pole plasm phenotype and, to varying degrees, abdominal deletions.

Double mutants were constructed between stau^D3, and Df(2R) PC4 and nos^L7, using stocks balanced for the second and third chromosome. Double mutants with dominant Bic-D alleles were constructed with the alleles Bic-D^D71 and Bic-D^DIIIE (Mohler and Wieschaus, 1986; Wharton and Struhl, 1989) and nanos (nos^L7), pumilio (pum⁶⁸⁰), oskar (osk¹⁶⁶ and osk³⁴⁶), staufen (stau^D3. stau^HL and Df(2R)PC4), vasa (vas^PD) and tudor (tud^WC).

Cytoplasmic transplantation was carried out as described in Lehmann and Nüsslein-Volhard (1986) for pole plasm transplantation, and as described in Sander and Lehmann (1988) for nurse cell transplants. The cytoplasm of the posterior or anterior pole was transplanted into recipient embryos in a 1:1 ratio. The cytoplasm from one donor nursecell cluster or oocyte was transplanted into up to five recipient embryos. Deficiencies are as described in Vâssin and Campos-Ortega (1987), Schüpbach and Wieschaus (1986a) and Lasko and Ashbumer (1988).

Maternal genotypes of donor embryos (E) and nurse cell-oocyte complexes (O): nanos: nos^L7/nos^L7 (E,O) and nos^L7/Df(3R)Dl-A143 (E); pumilio: pum⁶⁸⁰/num⁶⁸⁰ (at 18°C); oskar: osk¹⁶⁶/osk¹⁶⁶ (E,O) and osk³⁴⁶/ osk³⁴⁶), staufen: stau^D3/stau^D3 (E) and stau^D3/Df(2R)PC4 (E,O); vasa: vas^PD/Df(2L)osp²⁹ (E) and vas¹³¹/Df(2L)A72(O)-, valois: vls^PE/vls^PG (E) and vls^PE/Df(2R)TW2 (E,O); tudor: tud^WC/tud^WC (E) and tud^WC/Df(2R)Pu^rP133 (E,O); Bic-D^D: Bic-D^D71/Bic-D^DIIIE; Bic-D^Dstau: Bic-D^D71stau^D3/stau^D3.

Maternal genotype of recipients: nanos: nos^L7/nos^L7; pumilio: pum⁶⁸⁰/pum⁶⁸⁰ (at 18°C); oskar: osk¹⁶⁶/osk¹⁶⁶-, staufen: stau^D3/stau^D3 and stau^D3/Df(2R)PC4; vasa: vas^PD/ Df(2L)osp^M-valois: vls^PE/vls^PG and vls^PE/Df(2R)TW2.

vas^O14 and stau^C8 have temperature-sensitive periods before egg deposition. Females iransheterozygous for vas^O14 and vas^PD and females transheterozygous for stau^C8 and stau^D3 were mated to sibling males and kept in large egg collection vials on apple juice agar plates supplemented with fresh yeast. At the restrictive temperature (29°C), all embryos derived from vas^O14/vas^PD females and stau^c8/stau^D3 develop one or less abdominal segments (strong phenotype). At the permissive temperature (18°C), about 20% of all embryos derived from vas^O14/vas^PD females and about 30% of embryos derived from stau^C8/stau^D3 females develop one or less abdominal segments. Flies were ‘shifted up’ from the permissive temperature (18 °C) to the restrictive temperature (29°C) or ‘shifted down’ from 29°C–18°C. After the temperature shift, eggs were collected at 1-3 h intervals (29°C) or 2–10 h (18°C) intervals, and the eggs on the agar plate were counted. After 18 h of development at 29°C or 48 h at 18°C the hatch rate was determined by counting empty egg cases. Cuticle preparations of embryos that did not hatch were examined. The percentage of embryos with one or less abdominal segments (strong mutant phenotype) among all developed eggs was determined. In addition, embryos collected at 1h (29 °C) or 2h (18 °C) intervals were shifted at various times after egg deposition and analyzed accordingly. For each time point, between 50–400 eggs were counted. It should be noted that a general increase in phenotypic strength by cold temperature after egg deposition is observed as a very minor change in phenotypic strength even in bona fide amorphic alleles and probably reflects a general cold sensitivity of the process of abdominal development rather than specific features of mutant gene products (Schüpbach and Wieschaus, 1986a; R. L. unpublished data).

Calculations about the length of the temperature-sensitive periods are based on the assumption that all stages of oogenesis are affected by temperature in a similar way. This assumption is supported by the observation that the ‘up shift’ and ‘down shift’ curves have similar shapes and span the same time interval. The conversion between the different temperatures is based on observation of living embryos kept at different temperatures. The conversion is: 1h at 29°C=1.5h at 25°C=3h at 18°C.

For cuticle preparations, differentiated embryos were dechorionated, fixed in 1:4 glycerol and acetic acid, and embedded in Hoyer’s medium (van der Meer, 1977; Wieschaus and Nüsslein-Volhard, 1986). In some cases, embryos were freed from the vitelline membrane under water and embedded and cleared in 1:1 mixture of lactic acid and Hoyer’s medium. For electron microscopy, embryos were embedded after fixation in glutaraldehyde and osmium tetroxide in Durcupan and sectioned in serial longitudinal and transverse sections. The maternal genotypes of the embryos examined were: Btc-D°^D71stau^D3/stau^D3 Five embryos were sections. In a parallel egg laying, the cuticle of 409 mutant embryos was analyzed and 94 % showed the reversed abdomen phenotype as depicted in Fig. 3B and C. Three nos^L7 embryos were sectioned all had a morphologically normal pole plasm.

Six of the seven posterior group genes studied were identified in large-scale mutagenesis experiments for the second and third chromosome (Schüpbach and Wieschaus, 1986a, 1989; Nüsslein-Volhard et al. 1987 and unpublished). The first allele of nanos was isolated fortuitously (see Materials and methods). In subsequent screens more alleles of most of the loci were isolated such that each locus is represented by a number of weak and strong alleles (Table 1 and Materials and methods). These alleles allow the construction of a phenotypic series. With increasing phenotypic strength abdominal segments are lost from the center of the abdominal region to its anterior and posterior margin, such that abdominal segments four and six are more sensitive to reduction in gene activity than abdominal segments one and eight. This series has been described in detail for the cuticle phenotype of osk (Lehmann and Nüsslein-Volhard, 1986). For all nine posterior group genes, the cuticle phenotype and expression pattern of the pair-rule segmentation gene fushi tarazu have been compared in mutant alleles of varying strengths (Carroll et al. 1986; Lehmann, 1988; Manseau and Schüpbach, 1989). This comparison revealed that the phenotypic series is principally the same for all posterior group genes, and that the defect in abdominal segmentation is caused by a posterior expansion of the thoracic primordia at the expense of abdominal primordia (Lehmann, 1988). This finding strongly supports the interpretation that all posterior group genes participate in a common process.

All mutant posterior group genes share the abdominal phenotype (Fig. 1B). However, the phenotypic analysis of each gene reveals additional phenotypic traits, which are summarized in Table 1 and in the Materials and methods. A fundamental difference in phenotype that results from mutation of different posterior group genes concerns the posterior pole plasm. Absence of posterior pole plasm, polar granules and, later in embryonic development, of pole cells, is observed in mutants of seven genes: vas, vis, stau (Schüpbach and Wieschaus, 1986a, 1989), tud (Boswell and Mahowald, 1985), osk (Lehmann and Nüsslein-Volhard, 1986), and capu and spir (Manseau and Schüpbach, 1989). Pole plasm is present and is functional in germ cell determination in mutants of pum (Lehmann and Nüsslein-Volhard, 1987) and nos (Fig. 1C). The presence of pole plasm in pum and nos mutant embryos suggests a functional hierarchy, where the function of nos and pum depends on other posterior group genes, which themselves are critical for pole plasm formation.

In order to establish the particular function of each gene within the pathway, we carried out cytoplasmic transplantation experiments. We showed previously that the abdominal phenotype of osk and pum can be rescued by transplantation of wild-type posterior pole plasm into the abdominal region of mutant embryos (Lehmann and Nüsslein-Volhard 1986, 1987). In the case of pum, the rescue was never complete, while the abdominal phenotype of osk mutant embryos could be rescued fully. As in wild-type embryos, rescuing activity is present in the pole plasm of pum mutant embryos, while it is absent from the posterior pole of osk embryos. This suggests that a signal that is required for abdominal segmentation is absent from the pole plasm of osk embryos and is present in pole plasm from pum embryos but cannot reach the abdominal region.

Transplantation of wild-type pole plasm into the abdominal region of mutant vas, vis, stau and nos embryos also restores normal abdominal development (we have not tested tud, because no allele with full penetrance for the abdominal phenotype is available). The extent of rescue is similar to that observed in osk and almost all embryos show at least some restoration of abdominal segments upon transplantation of wildtype posterior pole plasm (Table 2). A small percentage of embryos developed to hatching larvae and a few were raised to adults. Only in the case of nos were these adults fertile, suggesting that the posterior pole plasm of nos, although lacking the signal for abdominal segmentation, is normal with respect to the requirements for germ cell formation.

In order to investigate the nature of the signal(s) and to determine whether the phenotypic rescue is achieved by multiple signals or by only one signal, we carried out reciprocal cytoplasmic transplantations. Posterior pole plasm from all mutants was transplanted into the abdominal region of nos, pum, osk, vas, and stau embryos. Cytoplasm from mutants that are defective in the formation of the posterior pole plasm (osk, vas, vis, tud, stau) does not rescue any of the mutants (Table 2 and summarized in Fig. 2A). nos mutant embryos also lack rescuing activity, despite the normal morphological appearance of their pole plasm (Fig. 1C). The results show that rescue activity depends on the normal function of several genes A simple explanation for these findings is that a singular activity, which is localized in the pole plasm of wild-type embryos, is absent from the pole plasm of each of these mutant embryos. In contrast, pum embryos form a morphologically normal pole plasm and have wild-type levels of rescuing activity for all posterior group mutants including pum itself, which suggests that pum does not affect the formation or the localization of the signal (Lehmann and Nüsslein-Volhard, 1987 and discussion). These results are consistent with the posterior signal being the endproduct of a biosynthetic pathway involving all posterior group genes except pum. Alternatively, the posterior signal may be encoded by only one posterior group gene and products of other genes have accessory functions in the pathway, such as localization and transport of the signal.

Transplantation of the posterior signal to the anterior of developing embryos causes ectopic posterior differentiation and can lead to ‘bicaudal’ embryos, which lack anterior structures and have two posterior ends including two abdomens and telsons in mirror image (Frohnhöfer et al. 1986; Sander and Lehmann, 1988; Lehmann and Frohnhöfer, 1989). This ‘bicaudal’ phenotype can also be produced genetically by embryos from mutant Bic-D^D females (Mohler and Wieschaus, 1986). In transplantation experiments, it has been shown that the posterior signal is localized to the anterior and posterior poles of Bic-D^D embryos (Table 3 and Lehmann and Nüsslein-Volhard, 1986). Previously we had shown that in double mutant combination with osk, the ‘bicaudal’ phenotype is completely suppressed and results in embryos indistinguishable from single osk mutants (Lehmann and Nüsslein-Volhard, 1986; Fig. 2B). This result suggests that osk is required not only for the normal function of the posterior signal at the posterior pole, but also for its function at the anterior pole in Bic-D^D embryos.

In order to test whether the activity of all posterior group genes is necessary for the formation of a second abdomen at the anterior pole of Bic-D^D embryos, we made double mutants between different posterior group mutants and Bic-D. We found three different phenotypes, which are summarized schematically in Fig. 2B. Bic-D^Dvas and Bic-D^Dnos behave like osk in that they suppress all abdominal development as well as the formation of a second telson at the anterior. Thus, the products of the osk, vas and nos genes are critical for the normal as well as the ectopic presence of the posterior signal.

Bic-D^Dpum double mutant embryos suppress all abdominal development but display a duplication of the posterior telson at the anterior. The Bic-D^Dpum phenotype is best described as the addition of the two single phenotypes. Due to the Bic-D^D mutation, the posterior signal is localized to the anterior, suppresses bed function, and induces formation of a second telson anteriorly; due to the pum mutation the posterior signal is not distributed and abdominal segments are not formed efficiently. Thus, a similar phenotype can result from either the inefficient distribution of the mislocalized posterior signal as in Bic-D^Dpum embryos or from the absence of anterior and posterior signals as in bed nos embryos (Nüsslein-Volhard et al. 1987; Lehmann and Frohnhöfer, 1989).

A novel phenotype is observed in Bic-D^Dstau double mutants: a complete abdomen is formed in reverse orientation at the anterior of the embryo, while no abdomen is formed at the posterior (Fig. 3 B,C). At both ends, a telson is formed. Thus, stau activity, although necessary for the formation of an abdomen in the posterior half of the embryo, is dispensable for the formation of a reversed abdomen in the anterior half of the embryo, nos gene function, however, is necessary for the formation of the reversed abdomen in Bic-D® stau mutant embryos, as embryos derived from females that are mutant for Bic-D^D, stau and nos lack the abdomen and form head and thorax indistinguishable from those derived from single stau females.

The Bic-D^Dstau phenotype suggests that the posterior signal can be active at the anterior independently of stau function. Similar results have been obtained by Manseau and Schüpbach (1989) for Bic-D^Dcapu and Bic-D^Dspir double mutant combinations. To determine the spatial distribution of the ‘posterior signal’ in Bic-D^Dstau embryos, we used such embryos as donors in transplantation experiments. Table 3 shows that near normal levels of rescuing activity are present at the anterior of Bic-D^Dstau embryos while no activity can be detected at the posterior. Since the anterior cytoplasm can restore abdominal segmentation in nos, osk and stau mutant embryos, the function of the stau gene-is restricted to the localization of the posterior signal to the posterior pole. Interestingly, although normal levels of rescuing activity are found at the anterior of Bic-D^Dstau embryos, neither pole cells nor polar granules are formed at the anterior and posterior pole (Fig. 3D, Schüpbach and Wieschaus, 1986a). This result suggests that stau plays a critical role in polar granule formation and demonstrates that polar granules themselves are not necessary for localization of posterior activity to the anterior pole.

The double-mutant combinations between the posterior group mutants and Bic-D allow us to distinguish three classes of posterior group genes: (1) pum, which is not required for initial localization of the signal, (2) stau, capu, and spir, which are required specifically for the localization of the signal to the posterior and finally (3) osk, vas, and nos, which are required for localization or synthesis of the signal. We can also conclude from these experiments that the pum, stau, capu and spir genes are not involved in the synthesis of the posterior signal.

Mosaic analysis of posterior group genes demonstrates that the embryonic phenotype depends on the genotype in the female germ line (Schüpbach and Wieschaus, 1986b; Lehmann and Nüsslein-Volhard, 1986, 1987; Irish et al. 1989; Manseau and Schüpbach, 1989). Thus, the posterior group genes are likely to be transcribed in the growing egg chamber or in the nurse cells. The nurse cells are sister cells of the oocyte and are located anterior to the oocyte. Products synthesized in the nurse cells are transported into the oocyte at the end of oogenesis (reviewed in: Mahowald and Kambysellis, 1980). Indeed, a high concentration of posterior activity is detected in cytoplasm from nurse cells (Sander and Lehmann, 1988). This finding suggests that this activity is generated in the nurse cells, transported into the oocyte cytoplasm, and is localized to the posterior pole of the egg cell. To determine which posterior group gene(s) is/are required for accumulation of this activity, we transplanted cytoplasm from nurse cells or oocytes of mutant ovaries into the abdominal region of osk or nos embryos (Table 4). Rescuing activity is detected in nurse cell and oocyte cytoplasm of all posterior group mutants tested except for nos. Since nos alone is required for the activity in the egg chambers, we suggest that nos encodes the posterior activity or a component that is necessary for this activity. Furthermore, these results show that neither pum, nor five other genes that affect pole plasm (osk, vas, vis, tud, stau), are required for synthesis of the posterior activity in the nurse cells and the oocyte.

Our transplantation experiments show that nos activity is present in the nurse cell cluster which is located anterior to the oocyte. Temperature-sensitive alleles available for four loci indicate the developmental periods that are critical for abdomen formation. The temperature-sensitive periods (TSP) of stau^C8 and vas⁰¹⁴ are shown in Fig. 4A and summarized together with those for osk³⁰¹ (Lehmann and Nüsslein-Volhard, 1987 and pum⁶⁸⁰ (Lehmann and Nüsslein-Volhard,1987in Fig. 4B. The temperature sensitivity for all alleles is restricted to the abdominal phenotype, and at the restrictive temperature embryos from vas, stau and osk females lack a complete abdomen. As the pole cell defect persists at all temperatures, these adults are always sterile. Two genes, osk and vas, have TSPs for the abdominal phenotype at the end of oogenesis (stage 10-14; Mahowald and Kambysellis, 1980). This period may not only reflect a critical stage for localization of nos-dependent activity but also for pole plasm formation since vasa protein becomes concentrated at the posterior pole at stage 10 (Lasko and Ashbumer, 1988; Hay et al. 1988). The TSP of stau starts at oogenesis stage 6 and lasts until stage 10. This suggests that the role of stau in pole plasm formation precedes that of the osk and vas genes. Only pum has a TSP that is exclusively after egg deposition (Lehmann and Nüsslein-Volhard, 1987). It appears significant that the latest TSP is found for a gene that does not affect pole plasm formation and this result implies that the distribution of the signal from the posterior pole to the presumptive abdominal region occurs after egg deposition.

All nos alleles isolated so far are female-sterile mutations. We have no indication that nos is involved in the development of the male germ line. The phenotype of two alleles, nos^L7 and nos^RW, is restricted to abdomen formation. The abdominal phenotype of embryos derived from nos^L7 homozygous females is indistinguishable from that of mutant osk embryos (Fig. 1B), but, in contrast to osk, nos embryos have normal pole plasm with polar granules (Fig. 1C). However, the alleles nos^RC and nos^RD also affect oogenesis. The mutant ovaries contain a smaller number of oocyte-nurse cell complexes, and only a few eggs are produced. Mutant females produced an average of 15 eggs during a five day period (n=38 females), while control females produced on average 140 eggs during the same period (n=38 females). This phenotype suggests a role of nos in the production of oocyte-nurse cell clusters early during oogenesis. Since two alleles, nos^L7 and nos^RW, are specific for abdomen formation and no effect on oogenesis is observed even when the respective mutant alleles are tested in trans to a deficiency of the region, it seems likely that the abdominal function of nos can be mutated independently of its function for oogenesis.

Stau mutants show anterior defects in addition to the abdominal and pole plasm phenotype: the labrum and dorsal bridge are missing and the ventral arms of the cephalo-pharyngeal skeleton are fused or reduced in size (Fig. 3A). This phenotype reflects a fate map change of cells in the anterior towards a more posterior fate (Schüpbach and Wieschaus, 1986a; Driever and Nüsslein-Volhard, 1988) and could in principle be caused either by mislocalization of posterior activity to the anterior or by reduction of anterior, bed-dependent activity. In fact the anterior defect is not caused by the mislocalization of posterior activity, since this defect is also observed in stau nos double mutants which lack posterior activity (data not shown). This experiment shows that the anterior defect is due to a direct effect of stau on bed as suggested by the altered pattern of bed RNA localization (St. Johnston et al. 1989) and the reduced levels of bed protein in stau embryos (Driever and Nüsslein-Volhard, 1988). Transplantations of wildtype anterior cytoplasm into stau mutant embryos confirm this interpretation, since the head defects of stau embryos are rescued (22/22 developed and injected embryos). Embryos from mutant bed females lack this rescuing activity. Stau gene function is thus unique among the maternal genes, since it is required for the accumulation of both the anterior, bed-dependent signal (Frohnhöfer 1987) and the posterior, nos- dependent signal.

Among the genes that we have studied, the nos gene is the only candidate gene that could encode the posterior signal necessary for abdomen formation. The strongest evidence in support of this conclusion is that rescuing activity for the abdominal phenotype requires normal nos gene function during oogenesis when the posterior signal is presumably synthesized, and during embryogenesis when its function is required for abdominal development. Consistent with this view, nos is required for formation of a normal as well as a duplicated abdomen in ‘bicaudal’ embryos. We conclude that the other posterior group genes affect the posterior signal indirectly, by interfering with its localization, distribution, or stability, but not with its synthesis. In the following discussion, we will use the terms posterior signal, posterior activity and nos gene product interchangeably. However, we cannot rule out that additional, unidentified components are required for posterior activity, and that these additional factors are transplanted together with nos.

During early embryogenesis, nos activity is required in the prospective abdominal region. However, our transplantation experiments demonstrate that nos- dependent rescuing activity is concentrated at the posterior pole of wild-type embryos. Recent experiments show that nos RNA is localized to the posterior pole (Wang and Lehmann, 1991) and suggest that the pole plasm serves as an anchor for nos RNA. Thus, nos RNA at the posterior pole may represent the nos- dependent rescuing activity, and nos protein may spread from this site of synthesis to the prospective abdomen.

Because the nos mutant phenotype and the results of transplantation experiments indicate that nos function is not required at the posterior pole but rather more anteriorly in the abdominal region, it is not clear why nos RNA is localized. Since nos activity is not detectable in mutant embryos that lack pole plasm, we suggest that a critical threshold of nos protein in the abdominal region is facilitated by nos RNA localization at the posterior pole. In addition, it may be important that nos activity is absent from the anterior of the embryo where it could suppress anterior development. Restriction of nos RNA to the posterior pole could be ensured further by destabilizing any nos RNA that has not been localized to the posterior pole. This hypothesis would account for the loss of rescuing activity in embryos lacking a morphologically distinct pole plasm, e.g. in osk mutants, and would suggest that the specialized structures associated with the pole plasm of the developing embryo may facilitate this critical placement of nos within the posterior region.

We assume that nos RNA is translated at the posterior pole and that nos protein emanates from there to the abdominal region. The pum phenotype and the TSP of pum suggest that the defect in pum embryos occurs after nos RNA has been localized to the posterior pole. Consistent with this idea are two observations: First, pum does not affect the localization of mw-dependent rescuing activity at the posterior pole (this study and Lehmann and Nüsslein-Volhard, 1987). Second, embryos defective in both nos and pum products are rescued by transplantation of wild-type posterior pole plasm in the same way as single pum mutants: a small number of abdominal segments are formed at the injection site in pum or pum nos double mutant embryos injected with posterior pole plasm (Lehmann and Nüsslein-Volhard, 1987; R.L. unpubl. observ.). This rescuing response is different from that of the other posterior group mutants including nos, in which a complete abdomen can be formed after injection of posterior pole plasm. We thus assume that pum function is necessary for the distribution or stability of nos protein.

The structure of the posterior pole plasm is controlled by a group of genes whose mutant phenotype can be best described as a grandchildless phenotype since the progeny of mutant females are sterile. Previously, in our analysis of the posterior group gene osk, we proposed that the pole plasm fulfills a dual function, and that germ cell formation and abdomen formation are controlled by the same set of genes (Lehmann and Nüsslein-Volhard, 1986). We now know that the two functions can be separated genetically, since nos and pum affect abdomen formation, but not germ cell formation. However, seven of the posterior group genes (osk, vas, val, stau, tud, capu, spir) do have both an abdominal and grandchildless phenotype (Boswell and Mahowald, 1985; Schüpbach and Wieschaus, 1986a; Lehmann and Nüsslein-Volhard, 1986; Manseau and Schüpbach, 1989).

The products of the ‘grandchildless genes’ appear to be involved in nos stability and/or localization at the posterior pole because mutations in all of these genes result in loss of nos activity in the embryo. The products of the ‘grandchildless genes’ may indeed provide the physical structure which serves as an anchor for nos RNA. However, weak, conditional alleles of some of the ‘grandchildless genes’ allow normal abdomen formation and presumably normal nos localization while polar granules are apparently absent and germ cells do not form. Although fully formed polar granules are absent in these mutants, it is possible that components of polar granules are present and perhaps partially assembled. Indeed, vasa protein, a component of the polar granules (Hay et al. 1988a) is concentrated at the posterior pole in mutants which have no polar granules (Lasko and Ashbumer, 1990; Hay et al. 1990).

It is neither clear how the ‘grandchildless gene’ products interact to form the posterior pole plasm nor which of the ‘grandchildless gene’ products may be directly involved in anchoring nos. Stau may be required for transport of nos-dependent activity from the nurse cells to the posterior pole of the oocyte, since stau is required for localization of nos activity to the posterior, but not anterior, pole of Bic-D^D embryos. Since nos-dependent activity is found at the anterior of Bic-D^Dstau but not stau mutant embryos, the altered Bic-D^D product is required to trap posterior activity at the anterior. Spir and capu mutants have a phenotype similar to that of stau in double mutant combination with Bic-D^D, which suggests that these genes are also involved in the transport of nos-dependent activity to the posterior pole (Manseau and Schüpbach, 1989).

The effect of Bic-D^D on nos localization to the anterior pole might be mediated through pole plasm components like osk and vas, since nos activity is absent from the anterior pole of Bic-D^Dosk and Bic-D^Dvas embryos. Similarly, the effect of stau on nos localization to the posterior pole may also be mediated through osk and vas since the TSP for the abdominal phenotype of stau precedes that of vas and osk and stau is required for vasa protein localization (Lasko and Ashburner, 1990; Hay et al. 1990). Thus, although it is possible that vas binds nos at the posterior pole, it is, however, unlikely that vas anchors nos at the anterior of Bic-D^D mutant embryos since vasa protein is not enriched at the anterior of Bic-D^D mutant embryos (Wharton and Struhl, 1989; Hay et al. 1990; Lasko and Ashburner, 1990). Therefore, further studies are required to determine the role of these genes in localizing nos.

Determination and polarity of the oocyte may be functionally linked to determination of anterior-posterior polarity within the egg as suggested by the phenotype of Bic-D. Mutations in Bic-D can result in defects in abdomen formation and/or defects in the determination of the oocyte (Mohler and Wieschaus, 1986). Once the oocyte has become determined and its position established, the staufen, cappuccino, and spire gene products are required for assembly of pole plasm components at the posterior of the growing oocyte. These gene products may transport pole plasm components to the posterior pole or provide the architecture required for assembly of other pole plasm components such as oskar, vasa, valois and tudor at the posterior pole. Because a morphologically normal pole plasm is required for the formation of functional germ cell precursors, these genes are necessary for the production of gametes.

Two types of signals may be components of the pole plasm, yet not critical for its structural integrity. One type of signal is represented by nos, which is critical for abdominal development, but not germ cell development. The other type of signal might be the products of genes, yet to be identified, which would affect solely germ cell formation, determination, or function. However, it is also conceivable that the polar granule as an organelle is the germ line determinant.

We are especially grateful to Dr J. Campos-Ortega and K. Tietze for isolation and initial characterization of the first nanos allele, the mutant ‘L7’. We would like to thank Drs M. Ashbumer, J. Campos-Ortega, M. Muskavitch and T. Schüpbach, for mutant stocks. R. L would like to acknowledge the help of N. Thompson at the MRC in Cambridge, England with electron microscopy. We thank our colleagues in Cambridge, USA and Tübingen, Germany for stimulating discussion and helpful comments on the manuscript. Aji Kron helped with word-processing. Part of these studies were conducted at the MRC, Cambridge England and at the Whitehead Institute, MIT Cambridge USA. R.L. received support from an Otto Hahn fellowship of the Max-Planck Society and from NIH grant RO1-GM40704.