Identification of a component of Drosophila polar granules

Asymmetric segregation of cytoplasmic components is observed during the early cleavages of many vertebrate and invertebrate embryos (Davidson,1986). Although such cytoplasmic localization is often assumed to be developmentally important, with few exceptions (Weeks & Melton, 1987), little is known about the nature and function of the segregated components or about the mechanism and control of partitioning. One well-known example is the localization of a distinctive germ cytoplasm in early embryos of many organisms; its segregation into germ line cells is associated with their commitment to the germ cell lineage (Beams & Kessel, 1974; Eddy, 1975). In insects and anuran amphibians, experimental procedures have demonstrated that germ cytoplasm is both necessary (Bounoure, 1934; Buehr & Blacklar, 1970; Frohnhöfer et al. 1986; Geigy, 1933; Jazdowska-Zagrodzinska, 1966; Okada et al. 1974; Smith, 1966; Warn, 1975) and sufficient (Illmensee & Mahowald, 1974; Niki, 1986; Ikenishi et al. 1986) for the initiation of germ line development.

The germ cytoplasm is ultrastructurally distinctive, containing large numbers of electron-dense cytoplasmic organelles without delimiting membranes (polar granules in Drosophila and germinal granules in amphibians) (Balinsky, 1966; Mahowald, 1962,1968).

These granules are obvious candidates for elements important in germ cell determination because of their localization exclusively in germ cytoplasm. In Drosophila, the recent isolation of maternal-effect gap mutants which lack polar granules and fail to form germ cells (pole cells) lends further support to this hypothesis (Boswell & Mahowald, 1985; Lehmann & Niisslein-Volhard, 1986; Schiipbach & Wieschaus,1986) . However, the molecular requirements for germ cell determination are unknown. Studies of the molecular nature of polar granules and the mechanisms of their segregation are therefore important for understanding how the fate of germ cells is specified during early development. A fortuitously discovered monoclonal antibody against polar granules, Mab46Fll, may be useful for this purpose, as described below.

Drosophila melanogaster were raised on standard cornmeal-yeast-agar medium at 25 °C. Normal flies were of Oregon-R (OR) wild-type strain. Balanced stocks of vasa (vas^PD23) and pumillio⁶⁸⁰ were kindly provided by Drs T. Schiipbach and E. Wieschaus, Princeton University, and Drs C. Nüsslein-Volhard and R. Lehmann, Max-Planck Institute, Tübingen, respectively.

Polyclonal antisera directed against horseradish peroxidase (HRP) stain the nervous system of Drosophila as well as that of other insects (Jan & Jan, 1982). In an attempt to generate antibodies directed against subpopulations of neurones, a hybridoma fusion and screen for monoclonal antibodies was carried out, in which HRP (Sigma type VI) was used as the immunogen. In the course of screening the clone supernatants on tissue sections, an IgM-secreting clone (designated Mab46Fll) was isolated which recognizes cells of the Drosophila germ line. This antibody does not bind HRP and thus the relationship of the antigen recognized to the immunogen is unknown.

Embryo collections were done according to Allis et al. (1977). Embryos were kept at 25 °C and time points were taken at 1 h intervals. Embryos were processed for fixation, devitellinization and antibody staining as described by Bodmer & Jan (1987). Mab46Fll culture supernatant was diluted fivefold in blocking buffer (0·1 m-phosphate, pH 75, 0·3% Triton X-100, 0·3% sodium deoxycholate, 2 % bovine serum albumin (Miles, fraction V)) prior to use.

For cryostat sections, embryos or ovaries were fixed as above for 5 min, covered with a drop of OCT compound (Lab Tek products), frozen onto a chuck at −20°C and sectioned at 12 pm with a Reichert-Jung 2800 Frigocut. Sections were placed on slides coated with 0·1 % gelatin and fixed for 10 min in 2 % formaldehyde, 0·1 m-phosphate buffer, pH 7·6. Slides were processed for antibody staining as above.

For whole-mount staining of larval and adult gonads, the following procedure was adopted. A number of gonads were dissected in Schneider’s Drosophila medium (Gibco). Gonads were then placed on a glass slide coated with 0-1 % gelatin and the bulk of the excess solution withdrawn. A second slide similarly coated was then gently lowered down onto the first slide, pressing the tissue between them. The two slides were then placed against a block of dry ice for 5min. The slides were then split apart with a razor blade and transferred to prechilled acetone (− −70°C) for 5 min, to prechilled methanol (−70°C) for 3 min, and then through an ethanol series (at room temperature) into 0·1 m-phosphate buffer, pH 7·6. Slides were processed for antibody staining as above.

Immunoelectron microscopy was carried out as follows. Embryos were dechorionated in bleach: water (1:1), rinsed extensively with water and placed onto double-stick tape. They were then covered with fixative (2 % glutaraldehyde, 4% paraformaldehyde, 15% picric acid, 0·lm-sodium cacodylate, pH7·2–7·4). The vitelline membrane was ruptured with a glass microelectrode and embryos allowed to fix for 10min. Embryos were then nudged out of their vitelline membranes with a pair of blunt forceps and fixation allowed to continue for another 30 min. Tissues were dissected under the fixative and incubated similarly. Samples were then rinsed three times with 0·lm-sodium cacodylate pH 7·2·1 A. En bloc staining was performed with 2 % aqueous uranyl acetate for 2 h. A brief water rinse was followed by dehydration in 35%, 75% and 95% ethanol. The tissue was infiltrated with Lowicryl K4M (Polyscience): ethanol in proportions 1:2, 1:1 and undiluted K4M for 2h each, at room temperature. The initiator was benzoin ethyl ether. Polymerization was effected with a u.v. lamp (model UVG-11, Ultraviolet Products) at 356 nm for 24–48 h. Immunostaining was carried out as follows. Sections were incubated in water for lmin, in blocking buffer (3 % BSA, 2 % normal goat serum, 0-1 m-phosphate buffer, pH 7-2,0-3 % Triton X-100,0·3 % deoxycholic acid) for 5 min, and with Mab46Fll, diluted 1:4 in blocking buffer, for 15 min. Sections were then rinsed in blocking buffer with three changes of 2 min each, and subsequently in the same buffer with secondary antibody (goat anti-mouse IgM conjugated with 5nm colloidal gold (Jansen)), for 10 min. Grids were then rinsed as above, then three times, for 1 min each, with water. Sections were then poststained with 2 % aqueous uranyl acetate for 5 min and rinsed thoroughly with water. Sections were examined and photographed with a Philips 400 transmission electron microscope.

Protein extracts were prepared from various sources by homogenizing the tissue in a glass/glass homogenizer in approximately 10vol. of extraction buffer (2·5% SDS, 0·1 m-Tris-HCl, pH6·8, 10mm-ethylenediaminetetraacetic acid (EDTA)) to which the protease inhibitors (100 pu-benzamidine, 10mm-sodium bisulphite, 2 mm-phenylmethylsulphonyl fluoride (PMSF), 1 mm-ethyleneglycol-bis-(B-aminoethyl ether) N,N,N′,N′-tetraacetic acid (EGTA), 10/2M-leupeptin, 10gM-pepstatin, 100 Ki.u. ml⁻¹ aprotinin) were added just before use. Samples were then boiled for 5 min and spun at full speed in a microfuge (Eppendorf) for 5 min to pellet insoluble material. The supernatants were aliquoted and stored at −80°C. For SDS-PAGE, equal amounts of sample protein were precipitated with 15 % trichloroacetic acid (TCA) at 4°C for 20 min, and then pelleted in a microfuge at full speed at 4°C. The supernatants were aspirated and the pellets resuspended with acetone at 4°C. The samples were repelleted as above and the supernatants aspirated. The acetone pellets were then processed for SDS-PAGE under reducing conditions according to Laemmli (1970), and run on 10% polyacrylamide gels with 4 % stacking gels. Protein was transferred to nitrocellulose (Schleicher & Schuell, BA85) according to Towbin et al. (1979), using a Biorad transblot cell at 0·1A for 8-12 h. Nonspecific binding sites were blocked by incubation of filters for 1 h in blocking buffer (10 mm-Tris-HCI pH 8 0, 150mm-NaCl, 0·05 % Tween-20 and 2 % BSA), at 37°C. Filters were subsequently incubated at room temperature for 1 h with blocking buffer and Mab46Fll (1:10 dilution of culture supernatant), 1 h with affinity-purified rabbit anti-mouse IgM antibody (Zymed 1:1000 dilution), and 1h with alkaline phosphatase-conjugated goat anti-rabbit antibody (Promega Biotec, 1:7500). Three 15 min washes in blocking buffer followed each antibody incubation. Following a final wash, in blocking buffer without BSA, the alkaline phosphatase reaction product was developed according to the Promega Biotec protocol.

To obtain a protein sample enriched for pole cells, a variation of the protocol of Allis et al. (1977) for the isolation of pole cells was adopted, in which Percoll (Pharmacia) was substituted for Renografin-60 as a centrifugation medium. A final crude cell pellet (Allis et al. 1977) was resuspended into 3·5 ml of Chan and Gehring buffer (Chan & Gehring, 1971) (55mm-NaCl, 40mm-KCl, 15 mm-MgSO₄, 5mm-CaCl₂, 20mm-glucose, 50mm-sucrose, 10mm-Tricine, pH 7·0, l mg ml⁻¹ BSA) minus calcium (C&G-Ca²⁺), and loaded on a Percoll gradient constructed as described below. A 90 % Percoll stock was created by diluting 10 x C&G — Ca²⁺ 1:9 with 100% Percoll. This was diluted appropriately with lx C&G — Ca²⁺ to generate 15 %, 70 % and 80 % working Percoll solutions. A 35 ml 15–70 % linear Percoll gradient was created over a 5 ml 80 % bottom layer in 50 ml Falcon tubes. The gradient was centrifuged at 400 g for 20 min in a Sorvall SS34 rotor with the brake off. 2 ml gradient fractions were collected from the top. These were diluted 1:1 with C&G minus BSA (C&G -BSA) and spun at 120g for 10 min to pellet the cells. Resuspension and pelleting were repeated twice to remove traces of Percoll. Cells were then resuspended in 500 of C&G -BSA and counted using a hemocytometer. Cell-rich fractions were saved. A 100μl sample of cells from each fraction was processed as described below for immunocytochemistry, and the rest of the sample processed for Western blotting.

For immunocytochemistry, cells isolated from Percoll gradients were allowed to attach to glass coverslips coated with 100 μg ml⁻¹ concanavalin A (Sigma type IV) and 1 mg ml⁻¹ of poly-D-lysine hydrobromide (Sigma, approximate M_r345 000) for 5 min, fixed in 4% formaldehyde, 0-1 m-phosphate buffer, pH 7-6, for 15 min, and then processed for antibody staining as above.

Pole cell number and position in whole-mount embryos were recorded at different developmental stages using a Nikon Optiphot microscope equipped with differential interference contrast optics. A Nikon camera lucida was used to outline each labelled cell. Counts were made from these drawings.

Embryogenesis in Drosophila begins with a period of internal nuclear divisions; this is followed by migration of most of these nuclei to the egg periphery where blastoderm formation occurs. In the unfertilized egg and cleavage-stage embryo, cortical cytoplasm at the posterior pole contains large numbers of polar granules, 0·2–0·5 gm, electron-dense structures (Mahowald, 1962). At the end of the eighth nuclear division (beginning of stage 3; staging is that of Campos-Ortega & Hartenstein, 1985), several nuclei migrate into this cortex and become incorporated into three to five cytoplasmic buds that protrude from the posterior egg pole. Most polar granules become incorporated into these buds (Counce, 1963; Mahowald, 1968). Subsequently, during stages 3 and 4, these buds divide twice (Foe & Alberts, 1983), before 12–32 pole cells pinch off. These cells divide from 0 to 2 times prior to blastoderm formation (Technau & Campos-Ortega, 1986). Polar granules are segregated into these newly formed pole cells, which also contain large, electron-dense, nuclear structures that are distinct from nucleoli and known as nuclear bodies (Mahowald et al. 1976). Later during embryogenesis, polar granules and nuclear bodies fragment and small bits of electron-dense material (presumably derived from these structures) are found dispersed throughout the cytoplasm. Some of this electron-dense material is associated with the cytoplasmic face of the nuclear envelope and is called nuage (Mahowald, 1971a; Allis et al. 1979). Nuage is also seen in germ line cells in larval and adult stages, and appears to be a germ-line-specific structure.

Staining with Mab46Fll parallels the localization of polar granules, nuclear bodies and nuage. In the unfertilized egg and cleavage-stage embryo, immuno-reactivity is confined to a thin cortical cap at the embryonic posterior pole (Fig. 1A). Immunoreactivity subsequently becomes associated with the nuclei that migrate into the posterior pole cortex (Fig. 1B). This localization may involve active sequestration by the nuclei or degradation in the surrounding regions, since cortical internuclear regions are relatively devoid of immunoreactivity. By the time pole cells form, immunoreactivity is almost always localized only to cells with a typical pole cell morphology (Fig. 1C,D).

At the EM level immunocytochemical label is associated with polar granules and nuclear bodies of newly formed pole cells (Fig. 2). Later in embryogenesis immunoreactivity in the pole cell becomes scattered throughout the cytoplasm and associated with nuage (data not shown; nuage-associated label is shown in Fig. 6A, in the adult ovary).

Although material similar to Drosophila polar granules and nuage has been observed in the germ line of a number of species (Beams & Kessel, 1974; Eddy, 1975), the antibody does not recognize these structures in Xenopus laevis, Caenorhabditis elegans or various other Drosophila species, including the closely related species, D. simulans and D. mauritiana.

In Drosophila roughly twice as many pole cells are initially produced as ultimately become associated with the gonads (Rabinowitz, 1941; Sonnenblick,1941; Turner & Mahowald, 1976; Zalokar & Erk, 1976; Underwood et al. 1980; Technau & Campos-Ortega, 1986). From experiments in which exogenously labelled pole cells are transplanted to unlabelled embryos, it appears that this reduction involves both pole cells becoming lost and pole cell death during migration to the site of the gonad. There is little or no evidence suggesting pole cells contribute to tissues other than the germ line (Underwood et al. 1980; Technau & Campos-Ortega, 1986). In order to determine the utility of Mab46Fll as a pole cell lineage marker, we have used this antibody to follow the fate of pole cells and their migrations during embryogenesis. For this purpose, the number of Mab46Fll immunoreactive cells and their distribution between ‘lost’ and appropriately migrating cell populations, through stage 14 of embryogenesis (by which time the presumptive gonads have formed), was tabulated and plotted as frequency histograms in Fig. 3.

Most of the Mab46Fll immunoreactive cells present at the posterior pole of the embryo by the beginning of gastrulation (stages 6·7) (x̄ = 41 ± 0·7; ±S.E.M.; n = 100 for each sample; Fig. 3A) have a typical pole cell morphology (Fig. 1E,F) and are found outside the blastoderm. A variable number (0 · 8) is found lying interdigitated between neighbouring, nonstaining, blastoderm cells or underneath the blastoderm layer, apposed to its inner face but clearly separated from the yolk by the yolk membrane (Fig. 1C, arrowhead; 1D,E, arrows). At the beginning of gastrulation, pole cells are carried dorsally and anteriorly above a plate of about 150 cells (Fig. 1E,F) which forms a sack that encloses the pole cells and carries them inside the embryo during germ band elongation (Fig. 1G,H). At the end of germ band elongation (stage 10), the pole cells leave this sack by passing through the walls of the posterior midgut primordium (Fig. 1I,J). Occasionally a cell appears trapped in the midgut, or between cells of the midgut epithelia; however, the vast majority migrate through without apparent mishap. The average number of immunoreactive cells present during late stage 10 and early stage 11 is 43 ± 0·5 (Fig. 3B).

During early germ band shortening (early stage12), pole cells become displaced laterally (Fig. 1K,L), and a number (5 ± 0 · 4, Fig. 3C) of apparently lost cells is seen (Fig. 1L). The total number of pole cells has also decreased, to 37 ± 0 · 7 (Fig. 3C). By the end of germ band shortening (early stage 13), most pole cells are closely associated with each other and are contacting mesoderm at the level of the fifth to eighth abdominal segments (A5-A8) (Fig. 1M,N). The total number of labelled cells has decreased further (from 37 ± 0-7 to 29 ± 0 · 4, Fig. 3C,D). This is reflected in a decrease in the number of appropriately migrating cells (from 32 ± 0 · 5 to 26 ± 0 · 4, Fig. 3C,D) as well as in the number of apparently lost pole cells (from 5 ± 0 · 4 to 3 ± 0 · 2, Fig. 3C,D). No further reduction is seen through the middle of stage 14, when the appropriately migrating pole cells have condensed at the level of A5 and become surrounded by mesoderm (Fig. 1O,P; Fig. 3E,E’).

The following general observations are relevant concerning pole cell fate. First, immunoreactive cells that appear lost are found scattered throughout the embryo up to and through hatching. Therefore it is not the case that pole cells that do not reach the gonad invariably die or become incorporated in other tissues, resulting in a loss of expression of the Mab46Fll antigen. Second, weakly staining cells which have a distorted, blebbing morphology and are in the process of degeneration are found in both lost and appropriately migrating cell populations from the end of germ band elongation onward (data not shown). These observations are consistent with the hypothesis that the loss of labelled cells seen during germ band shortening is due to pole cell death, occurring in both lost and appropriately migrating cell populations. By stage 14, we find an average of 14 pole cells per gonad. This number is higher than that reported by others during roughly stages 15 – 16, with 5 – 7 pole cells in gonads of some embryos and 9 – 13 in gonads of others (Sonnenblick, 1941; Technau & Campos-Ortega, 1986). This difference in pole cell number suggests that pole cell death continues to occur following initial gonad formation, during stage 14.

Pole cell fate, as revealed by Mab46Fll staining, generally agrees with conclusions obtained from previous studies (Underwood et al. 1980; Technau & Campos-Ortega, 1986). However, compared to studies using cell transplantation, immunocytochemical staining with Mab46Fll provides a much simpler and more quantitative method of following pole cells in wild-type and mutant embryos. To illustrate this point, we have carried out Mab46Fll staining of embryos from homozygouspumillio mothers (Fig. 4). The maternally supplied pumillio gene product (located at the embryonic posterior pole) is required for proper abdomen specification; mutant embryos lack variable numbers of abdominal segments. In these embryos, the formation of pole cells and their subsequent migrations during gastrulation are normal (Fig. 4A,B). The behaviour of these cells during germ band shortening is severely disturbed, however. Lost cells are found scattered throughout the stage-13 mutant embryo (Fig. 4C,D), whereas in the wild-type condition they would be aligning themselves along the body wall at the level of A5-A7 (see Fig. 1N). This defect in migration is likely due to an abnormal environment, since functional pole cells can be recovered when pumillio pole cells are transplanted into otherwise sterile host embryos (Lehmann & Niisslein-Volhard, 1987).

Synthesis of polar granule components and assembly and transport of these elements to the future embryonic posterior pole occurs during oogenesis (Mahowald, 1962, 1971a). As a step towards understanding these processes, we have used light and EM immunocytochemistry to examine the cellular and subcellular localization of the Mab46Fll antigen throughout oogenesis.

Each of the two adult ovaries contains 10 – 20 parallel egg tubes or ovarioles within which developmental stages of oogenesis are arranged linearly. (See King, 1970, and Mahowald & Kambysellis, 1978 for reviews.) In Fig. 5A, a whole-mount preparation of part of an ovariole stained with Mab46Fll is shown. At the most anterior end, in the germarium, perinuclear label is present in several large cells, probably stem cells. These cells divide unequally to produce, after each division, another stem cell and a daughter cystoblast. The cystoblast subsequently undergoes four mitotic divisions with incomplete cytokinesis to produce a cluster of 16 cells interconnected by intercellular bridges. These are the smaller labelled cells, located in the posterior germarium (Fig. 5A). One of these 16 cells enters prophase of meiosis 1 and will become the oocyte; the other fifteen become nurse cells. During vitellarium stages of oogenesis, in which the bulk of oocyte and nurse cell growth occurs, perinuclear and cytoplasmic label is found in the 15 nurse cells (Fig. 5A,B). Staining is not seen in the oocyte until stage 10 (Fig. 5B, see below). Label is also not apparent in the follicle cells or any other somatically derived tissues of the gonad.

At the EM level, perinuclear label in germ line cells of the germarium (data not shown), and in nurse cells of the vitellarium (Fig. 6A), is found associated with electron-dense clumps of material on the cytoplasmic face of the nuclear envelope. These are the nuage or fibrous bodies previously described in the Drosophila germ line (Counce, 1963; Mahowald, 1971a; Mahowald & Strassheim, 1970). Label in the nurse cell cytoplasm may reflect the site of synthesis of the antigen.

During stages 10 – 12 of vitellogenesis, the nurse cells rapidly decrease in size as they inject their contents into the growing oocyte through intercellular channels at the anterior end of the oocyte. Cytoplasmic mixing rapidly distributes this material throughout the ooplasm (Gutzeit & Koppa, 1982). Localization of polar granules to the posterior pole of the oocyte takes place during this period, beginning during stage 10 and continuing through stage 13 (Mahowald, 1962, 1971a). At the earliest stage when Mab46Fll immunoreactivity is detectable in the oocyte (around stage 10; Fig. 5B), it is located at the posterior tip of the oocyte and is associated with polar granules (Fig. 6B). By stage 14 (the mature oocyte, Fig. 5C), the number and size of these granules has increased, and they are often found associated with mitochondria (Fig. 6C). Because both polar granules and nurse cell nuage label with Mab46Fll, our data support the hypothesis that polar granules may derive from the nuage (Mahowald, 1971a). Since materials from nurse cells enter the oocyte at its anterior end and are mixed rather thoroughly by the swirling movements in the oocyte, the rapid localization of the Mab46Fll antigen to the posterior tip of the oocyte suggests that specific targeting mechanisms are required for this localization.

Synthesis of the Mab46Fll antigen by nurse cells during oogenesis is consistent with a postulated role of the antigen as a component of polar granules, structures important in early germ line determination. Synthesis in the adult male germ line, however, would suggest that this antigen plays additional roles in germ line function. We have carried out immunocytochemical staining with Mab46Fll to address this possibility.

Fig. 5D shows the apical half of a whole-mount testis stained with Mab46Fll. Small cells near the apex are germ line stem cells and their offspring the cyst progenitor cells. Cyst progenitors divide mitotically four times to produce a cluster of 16 primary spermatocytes. (See Lindley & Tokuyasu, 1980 for a detailed review of spermatogenesis.) Perinuclear label is present in each of these cell types (Fig. 5D). One or more large dots of label are also seen in the cytoplasm or associated with the nuclear envelope (Fig. 5D, arrow). These are probably the dense cytoplasmic masses described by Kessel (1981) (see Discussion). More distal to these stages, larger maturing spermatocytes are seen. Perinuclear and punctate staining are detectable until about the time spermatocytes enter meiosis. At the EM level, Mab46Fll label in germ line cells of the testis is primarily found over perinuclear, electron-dense structures (data not shown), similar to those labelled in the nurse cells of the ovary (Fig. 6A). The significance of antigen localization to these structures will be returned to in the Discussion.

During late embryogenesis (stage 16 onwards) and during larval life, a general nuclear staining is apparent in essentially all cell types (Fig. 7D), in addition to the germ line staining seen with Mab46Fll. Similar staining is not seen with second antibody alone or with several other mouse IgM monoclonals used (data not shown). Although it persists throughout larval stages, this staining becomes undetectable in adult somatic tissues (cf. somatic cells of ovary in Fig. 5).

We have carried out Western blotting with Mab46Fll to identify the antigen recognized by this antibody. Each of the tissues that label immunocytochemically with Mab46Fll contains an immunoreactive 72 – 74 × lO³M_r protein (Fig. 7A). The intensity of this band on Western blot is roughly correlated with the abundance of the Mab46Fll antigen seen immunocytochemically. Thus, stage-2 to -4 embryos show a weak immunoreactive band. This band is much more prominent in the lanes containing total protein from adult ovaries or a fraction enriched (roughly 10-fold) for pole cells from stage-2 to -5 embryos. A similar band is also present in protein from adult testis.

If the 72 – 74 × 10³M_T protein is in fact a component of polar granules, it should be absent in mutant embryos which are missing polar granules. One mutant which displays such a phenotype is the maternal-effect mutant vasa (Schiipbach & Wieschaus, 1986). Embryos from homozygous mutant mothers lack detectable polar granules and fail to make pole cells (Schiipbach & Wieschaus, 1986). They also lack detectable Mab46Fll antigen immunocytochemically (Hay et al. unpublished data). Fig. 7C shows that these embryos contain no detectable immunoreactive protein, even following prolonged development of the alkaline phosphatase reaction product. The 45xlO³Af_r band in wild-type as well as mutant embryos is one of the yolk polypeptides, major constituents of the early embryo. This band only appears following prolonged exposure and is also present in control experiments using other mouse IgM monoclonal antibodies (data not shown). Thus it is likely the 72-74 xlO³M_r protein from early embryos is associated with polar granules.

Previous studies by Waring et al. (1978) revealed a single major protein species of 95xlO³Af_r enriched for in pole-cell-enriched cell populations. A band of approximately 70 × 10³M_r also appears as a fairly major component of the various fractions (Waring et al. 1978). The relation between this band and the Mab46Fll antigen is not clear. However, of the various tissues and fractions we have examined, the 72 – 74 × 10³M_T antigen does not comigrate with any major protein species detectable with staining by Coomassie blue. Neither have we detected any immunoreactivity associated with proteins of higher molcular weight, even when special attention was paid to avoid possible degradation.

Total protein from late (stage 16 and 17) wild-type embryos (Fig. 7A) (and from homozygous vasa embryos derived from homozygous vasa mothers), has a large amount of an immunoreactive protein. This cannot be accounted for by an increase in the germline-associated antigen, due to gonad growth, but may be due to the nuclear staining of somatic tissues that is well developed by this time (Fig. 7D). Furthermore, the level of the nuclear staining at different stages of the life cycle correlates well with that of the immunoreactive 72 − 74x10³M_r band (Fig. 7B). This band is found at low levels in early embryos. A large increase is apparent by stage 16, when nuclear staining appears. During the second and third larval instars, the intensity of the 72 – 74 × lO³m_r immuno-reactive band decreases. In adults with gonads removed, there is even less immunoreactive protein, which corresponds to the lack of detectable nuclear staining in adult somatic tissues. The 72 − 74 × lO³Af_r protein identified as a component of polar granules comigrates with the protein associated with somatic nuclei. Whether these are the same protein remains to be determined.

Polar granules are localized to the posterior pole of mature oocytes and early embryos of Drosophila and may be associated with determinants of germ line cells (Illmensee & Mahowald, 1974). We have isolated a monoclonal antibody, Mab46Fll, that recognizes a component of polar granules. This antibody labels germ cells throughout the lifecycle of the fly. Using this antibody, we have made the following observations. (1) The rapid rate at which the Mab46Fll antigen is transported from anteriorly located nurse cells to a thin cortical cap at the posterior end of the oocyte suggests the presence of a specific and effective targeting machinery. (2) The association of this antigen with nuclei in the pole buds of cleavage-stage embryos and its clearing from the internuclear cortical regions raise the possibility that it may be actively sequestered. Likely candidates for this sequestration may include cytoskeletal components that show distinctive patterns of distribution at these early stages (Warn et al. 1985; Karr & Alberts, 1986). (3) Germ-line-specific structures, including polar granules, nuclear bodies, nuage and dense cytoplasmic masses, are related biochemically because they all contain the Mab46Fll antigen. (4) The Mab46Fll antigen appears as a 72 – 74 × 10³M_r protein on Western blots. This study suggests that the Mab46Fll antibody may be used for further biohemical characterizations of the components of polar granules. In addition, this antibody serves as a faithful marker for pole cells throughout embryogenesis. Discussions of some of these points are presented below.

In Drosophila and other Dipterans, many more pole cells are formed than are found in the gonads (Rabinowitz, 1941; Sonnenblick, 1941). This has suggested that pole cells may contribute to tissues other than the germ line (Poulson & Waterhouse, 1960). However, recent transplantation experiments using [³H]thymidine or HRP-labelled pole cells have found no evidence for pole cell incorporation into other tissues (Underwood et al. 1980; Technau & Campos-Ortega, 1986) and suggest instead that unaccounted cells may have got lost and/or died. Consistent with these observations, we find that two major causes of the low number of pole cells found in the embryonic gonad are (1) pole cells becoming lost, primarily during germ band shortening and (2) pole cell death, occurring in both the lost and appropriately migrating populations, also primarily during germ band shortening.

In pole cell transplantation experiments, in which large numbers of labelled pole cells are transplanted into the posterior pole of an unlabelled blastoderm embryo, the final number of pole cells in the gonads is similar to that of an uninjected embryo (X = 9). This has suggested that there is regulation of the final number of pole cells in the gonad, with an upper limit of about 13 pole cells per gonad at about stage 16 (Technau & Campos-Ortega, 1986). Our finding of an average of 14 labelled cells per gonad during stage 14 suggests that some of this regulation may occur following gonad formation, during stages 15 – 16.

This antibody should prove useful in examining aspects of postembryonic germ cell behaviour since it recognizes premeiotic germ cells throughout the fly lifecycle. It should also greatly facilitate study of the fate of pole cells in embryos in which maintenance of the germ line, following cell formation, is affected (Engstrom et al. 1982 ; Oliver et al. 1987 ; Togashi et al. 1986).

Polar granules, nuclear bodies, nuage and dense cytoplasmic masses are germline-specific structures. The observations that these structures are all recognized by Mab46Fll and that tissues in which they are found show a common immunoreactive 72 – 74 × 10³M_r protein on Western blotting indicate they are biochemically related. The fact that similar structures are found in the definitive germ line of phylogenetically distant groups, including other dipterous insects (Mahowald & Boswell, 1983), nematodes (Strome & Wood, 1982), amphibians (Smith et al. 1983) and mammals (Eddy & Hahnel, 1983), suggests that they are important for some fundamental aspect of germ cell life. The function of any of these structures is unknown. It is perhaps relevant, however, to note that in Drosophila polar granules (in oocytes and early embryos) and dense cytoplasmic masses (in spermatocytes) have been demonstrated to contain RNA, and are present at stages when mRNA storage for later use is likely to be important (Mahowald, 1971b; Kessel, 1981). Thus, it seems plausible, though there is no direct evidence, that these structures store RNA necessary for the determination of germ line cells in the early embryo and for spermatid formation in the testis, respectively.

If polar granules are directly involved in pole cell formation and germ line determination, it should be possible to isolate mutants that lack polar granules and which fail to form pole cells. Such mutants, members of the grandchildless-knirps class of maternal-effect mutations, have been isolated in Drosophila (tudor, Boswell & Mahowald, 1985; vasa, valois, staufen, Schiipbach & Wieschaus, 1986; oskar, Lehmann & Niisslein-Volhard, 1986). Embryos derived from females homozygous for mutations at these loci lack a germ cytoplasm that contains polar granules and they fail to form pole cells. In addition, however, embryos show deletions of a variable number of abdominal segments, and may exhibit head defects or incomplete blastoderm cellularization, depending on the allele. The absence of this germ cytoplasm appears closely associated with the loss of pole cells and not with deletions of abdominal segments since, in the case of tudor and oskar, the abdominal segmentation and pole cell phenotypes are separable (Boswell & Mahowald, 1985; Lehmann & Niisslein-Volhard, 1986). Similarly, in mutants showing a bicaudal phenotype (a mirror-image duplication of the posterior abdomen at the anterior end) the transformed anterior end lacks polar plasm and does not form pole cells (Bull, 1966; Niisslein-Volhard, 1977; Mohler & Wieschaus, 1985). The above grand-childless mutations thus seem most likely to be due to defects in shared elements necessary for the synthesis, packaging, transport or localization of abdomen and pole cell determinants. Indeed, Mab46Fll staining has revealed defects in these processes in the above mutants (Hay et al. unpublished data).

The most direct way of addressing whether polar granules are necessary and sufficient elements for pole cell formation is to isolate them and their constituents from oocytes or preblastoderm embryos (when polar plasm is competent for nuclear induction), for biochemical and genetic analysis. A step in this direction is the identification of a 72-74 xlO³M_r protein as a likely component of polar granules, nuclear bodies and nuage. If, based on this association, Mab46Fll can be used to purify polar granules for further molecular characterization of their protein and RNA components, it may be possible to test critically polar granule function in the early embryo.

We would like to thank T. Schiipbach and E. Wieschaus for sending us the vasa^PD23 stock, and R. Lehmann and C. Nüsslein-Volhard for sending us the pumillio⁶⁰ stock. We also thank T. Kornberg for use of his fly cage facility. We especially thank K. Logan for help analysing the data concerning pole cell number, E. Bier, K. Logan, K. Neugebauer, and T. Schwarz for their valuable comments on the manuscript and Ms L. Schulte for help in preparing the manuscript. This work was supported by NTH and HHMI grants to L.Y.J. and Y.N.J. B.H. was supported by an NLH predoctoral training grant. L.Y.J. and Y.N.J. are HHMI investigators.