parva germina, a gene involved in germ cell maintenance during male and female Drosophila gametogenesis



We report the initial characterization of a gene, parva germina (pag), required for germ cell maintenance in both males and females. pag gonads contain a small number of germline stem cells at the onset of gametogenesis. In contrast, adult mutant gonads are either empty or have a very small number of germ cells that never develop in 16-cell cysts. Ovarioles and testes, therefore, are rudimentary, and the very few germ cells they contain are unable to differentiate into eggs or sperm. Germline stem cells are progressively depleted over time. The average number of germ cells, therefore, decreases in pag mutant ovarioles with the age of the mother, whereas the proportion of agametic germaria goes up. These observations suggest that the pag gene product is involved in germ cell maintenance in both male and female gametogenesis. Developmental Dynamics 232:835–844, 2005. © 2005 Wiley-Liss, Inc.


The source of germ cells for the continuous production of gametes is ensured in Drosophila by a specialized cell type, the germline stem cells (de Cuevas et al., 1997). Like stem cells found in several tissue lineages of multicellular organisms ranging from Hydra to mammals (Lin, 1998), Drosophila germline stem cells undergo asymmetric division and the two daughters adopt separate fates: one remains a stem cell, the other differentiates. Because tissue maintenance in multicellular eukaryotes depends on a balance between cell proliferation and cell differentiation, the developmentally asymmetric division of the germline stem cells represents a special tool to investigate differentiation mechanisms (Lin, 1997). Analysis of mutants that affect early germ cell differentiation and stem cell survival can identify the processes governing the asymmetric cytoplasmic distribution. The first visible sign of germline establishment in Drosophila occurs very early during embryogenesis when a small number of migrating nuclei reach the posterior pole of the syncytial embryo and give rise to the pole cells (Foe et al., 1993). During gastrulation, the pole cells migrate through the midgut epithelium to the gonadal region where they become surrounded by somatic mesodermal cells and become the embryonic gonads that remain undifferentiated (Campos-Ortega and Hartenstein, 1997) in both males and females until the transition from larvae to pupae. In female gonads, somatic cells begin to differentiate at the larval–pupal transition, partitioning germline cells into 17–19 ovarioles (King, 1970; Godt and Laski, 1995). Female germ cells have divided two or three times by symmetric mitosis in the undifferentiated larval ovary, suggesting that these cells have not yet acquired a stem cell identity. However, when the ovarioles form, continuous production of egg chambers starts, suggesting germ cells have acquired a stem cell identity. Oogenesis begins when a stem cell divides to produce two daughters. Two–four germline stem cells reside in a specialized region of the ovariole, called the germarium, that is subdivided into four regions corresponding to the various stages of cyst development (Mahowald and Kambysellis, 1980). Each stem cell divides to produce a self-renewing stem cell that remains at the tip of the germarium and a cystoblast that moves away and undergoes four rounds of mitosis with incomplete cytokinesis to form a cyst of 16 interconnected cells, the cystocytes. The 16-cell cyst is then surrounded by a monolayer of somatically derived follicle cells in regions 2–3 to form the egg chamber that then moves from the germarium to the vitellarium (see Spradling, 1993).

Mutant analysis has identified some of the genes involved in the control of the asymmetric division of the germline stem cells, including piwi (Lin and Spradling, 1997; Cox et al., 1998), fs(1)Yb (Yb) (King and Lin, 1999), and decapentaplegic (dpp; Xie and Spradling, 1998), which define a somatic signaling pathway. By contrast, the pumilio (pum; Lin and Spradling, 1997; Forbes and Lehmann, 1998; Parisi and Lin, 1999), nanos (nos; Forbes and Lehmann, 1998; Wang and Lin, 2004), and bag-of-marbles (bam; McKearin and Spradling, 1990; McKearin and Ohlstein, 1995; Ohlstein and McKearin, 1997; Chen and McKearin, 2003a) gene transcripts are differentially expressed between stem cell and cystoblasts, suggesting that these genes could be involved in the regulation of intrinsic mechanisms. The fusome provides morphological evidence for an intrinsic mechanism mediating asymmetric stem cell division. This specialized cytoplasmic structure rich in membrane skeletal proteins (Lin and Spradling, 1995) associates with the anterior pole of the mitotic spindle during stem and cyst cell divisions (Deng and Lin, 1997) and remains asymmetrically distributed between the cystocytes, generating polarity in the developing cyst (de Cuevas and Spradling, 1998).

Candidate genes involved in germline stem cell development and maintenance might be identified by mutations that affect germ cell differentiation at very early stages. In this study, we provide initial characterization of the gene parva germina (pag) that is required for germ cell maintenance in both male and female. Mutations in pag result in rudimentary ovaries and testes where very few germ cells are unable to differentiate into eggs or sperm.


The pag Phenotype

The new locus parva germina (pag) was identified from the insertion line 0263/32 from a collection of Drosophila strains carrying lethal or semilethal P-element insertions on the third chromosome (Deak et al., 1997). Homozygous pag males and females have normal viability but have very small gonads and are sterile. To ask whether the P-lacW element associated with the third chromosome of this line could be responsible for the sterility phenotype, both were mapped by recombination. w/+; ru h th st cu sr e ca/pag females were mated to +/Y; ru h th st cu sr e Pr ca/TM6C males, and their progeny were scored for third-chromosome crossovers; individual crossover/ru Pri ca males were crossed back to w/w; pag/TM6C females, and several crossover/pag females (and usually also males) were tested for fertility from each crossover. Along the way, the presence or absence of the P-lacW was also noted for each crossover. At least several of both crossover products of all seven intervals were tested. Female and male sterility co-map to the middle of 3L; the eight informative crossovers place pag midway between h (66D10) and th (72D1). The P-lacW maps to the middle of 3R, probably between sr and e, more or less consistent with its in situ location at 94A5-10. The pag sterility phenotype is not due, therefore, to the P insert but to a different genetic lesion. A set of deficiencies deleting bits of the interval between 68 and 72 were tested for complementation with pag. Of those tested, only Df(3L)vin7 (68C8-11;69B4-5) does uncover pag, both genetically by male and female sterility and morphologically by very rudimentary ovaries and testes. Moreover, because Df(3L)F10 (69A2-proximal) does not uncover it, pag lies somewhere between 68C8 and 69A1. Furthermore, the next deficiency distal (lxd15, 67E;68C10-15) also gives adults/pag with visible phenotypes like those /vin7 but milder; however, these flies are fertile, and their gonads look roughly normal, although females seem have rather fewer mature eggs than usual.

To investigate whether the pag mutation might disrupt early processes of gamete development, we examined ovaries from females either homozygous or hemizygous (pag/Df(3L)vin7) for pag. Ovaries from females of known age (from 0 to 20 days old) were dissected and stained with antibodies against the germ cell-specific protein Vasa (Lasko and Ashburner, 1990) to mark the germline and antibodies against β-tubulin to outline somatic cells and ovary morphology. The ovarian phenotypes of the two allelic combinations examined were quite similar, although the ovarioles from pag/pag females presented slightly milder defects.

Newly eclosed heterozygous pag/TM6C sibling females were phenotypically wild-type and had tubular ovarioles containing obvious germaria (Fig. 1A), in which germline stem cells divided regularly producing cystoblasts, which then divided four more times to produce the 16-cell cyst; these cysts moved down the germarium, which contains seven to eight egg chambers at different stages of oogenesis (Fig. 1B,C). In contrast, ovarioles from newly eclosed mutant flies were very thin and showed severe defects in gamete development (Fig. 1D). The most obvious defect was that no ovarioles contained any 16-cell cysts anywhere along their length. Moreover, even in newly eclosed pag/Df(3L)vin7 females 60% of ovarioles completely lacked any germ cells whatsoever (57% for pag/pag); the other 40% had a few (Table 1). Approximately 23% of germaria contained one or two stem cells only (Fig. 1E,F); the remaining 17% had only three to six germ cells. On the basis of their position 46% of these cells could be discriminated as cystoblasts, 51% as germline stem cells, and 3% as two-cell stage dividing cystocytes. Very occasionally (0.4%) in the milder homozygous pag mutants a germarium also contained one cluster of six to eight small undifferentiated cells that stain positively with the anti-Vasa antibody, in addition to germline cells at the tip of the ovariole. The germ cells in young pag females appeared to be compromised in successful division. The variability within germaria from young females is consistent with germ-cell proliferation being successful earlier but declining later. This finding predicts that the germ-cell number should be further reduced in older females and, conversely, that embryos and pupae should have more germ cells than young adults.

Figure 1.

Ovary phenotype of pag/Df(3L)vin7 newly eclosed females. A–F: Ovaries from heterozygous balancer control (A–C) and pag/Df(3L)vin7 (D–F) females were simultaneously stained with antibodies against β-tubulin (green) and Vasa (red) or Vasa alone. A–C:pag/TM6C ovaries have ovarioles with egg chambers and well-developed germaria (A) that are usually composed of four regions (B,C): region 1 contains stem cells at its apical tip, cystoblasts and 2- to 8-cell clusters; region 2a contains 16-cell cysts that have completed dividing; in region 2b, the germline cysts start to be surrounded by follicle cells; region 3 contains a distinct egg chamber that is also cyst 1 of the vitellarium. D:pag/Df(3L)vin7 ovarioles are very thin with few if any germ cells. E,F: Detail of a mutant ovariole showing terminal filament cells (arrow in E) and only two germ cells in the reduced germarium. Scale bar in F = 20 μm in A,D, 10 μm in B,C,E,F.

Table 1. Germ Cells in pag/pag and pag/Df(3L)vin7 Ovariolesa
GenotypeAge (days)Ovarioles examinedAverage number of gcNumber (%) of germaria with
0 gc1 gc2 gc3 gc4 gc5 gc6 gc7 gc8 gc> 8 gc
  • a

    Ovarioles from pupae and adult mutant flies were stained with the anti-Vasa antibody to recognize germ cells (gc).

pag/pagLate pupa2442.1293 (38.1)22 (9.0)31 (12.7)18 (7.4)51 (20.9)9 (3.7)15 (6.2)2 (0.8)2 (0.8)1 (0.4)
pag/Df(3L)vin7 2621.68107 (40.8)28 (10.7)51 (19.5)23 (8.8)38 (14.5)3 (1.1)10 (3.8)1 (0.4)1 (0.4) 
pag/pag02211.29126 (57.0)16 (7.2)27 (12.2)19 (8.6)20 (9.0)3 (1.4)8 (3.6)1 (0.5)1 (0.5) 
pag/Df(3L)vin7 3470.96209 (60.2)42 (12.1)39 (11.2)33 (9.5)17 (4.9)2 (0.6)5 (1.4)   
pag/pag53010.67196 (65.1)39 (13.0)46 (15.3)10 (3.3)9 (3.0)1 (0.1)    
pag/Df(3L)vin7 2530.49175 (69.2)37 (14.6)36 (14.2)5 (2.0)      
pag/pag102540.30206 (81.1)26 (10.2)19 (7.5)1 (0.4)2 (0.8)     
pag/Df(3L)vin7 4110.18353 (85.9)4.0 (9.7)18 (4.4)       
pag/pag203160.08298 (94.3)12 (3.8)6 (1.9)       
pag/Df(3L)vin7 4520.04437 (96.7)12 (2.7)3 (0.7)       

Germ Cells Are Progressively Depleted With Age

Ovaries from 5-, 10-, and 20-day-old mutant females were examined to ask whether germ cells were being lost with time. The proportion of agametic germaria increased with age of mother, whereas the proportion of germaria with more than two cells decreased (Table 1). Strikingly, all germaria that had any germ cells at all had one or two at the very tip, presumably stem cells; no germaria were observed that seemed to have only cystoblasts or cystocytes but no stem cells. Consequently, as females aged, first cystocytes disappeared from this population of germaria, then cystoblasts, then finally stem cells themselves. Occasionally, pupal ovarioles of homozygous pag mutants displayed rare clusters of six to eight small undifferentiated germline cells, posterior to the large germ cells localized at the tip of the germarium.

To assay whether pag is required for germline development before oogenesis, we stained gonads from early third instar mutant larvae with the anti-Vasa antibody. Abnormalities in the number or morphology of germ cells in third instar larval ovaries, when stem cells are formed, could reflect early defects in primordial germ cell proliferation and development. We found that homozygous (n = 45) and hemizygous (n = 20) mutant ovaries were similar to wild-type heterozygotes (n = 30). Ovaries from wild-type sibling females had 52.8 ± 8.9 germ cells located in the medial region of the ovary (Fig. 2A), whereas ovaries from pag mutants contained 47.9 ± 9.8 germ cells in the same position (Fig. 2B).

Figure 2.

pag is not required for germ cell proliferation during early developmental stages. A,B: Ovaries from pag/TM6C (A) and pag/Df(3L)vin7 (B) early third instar larvae stained for actin (red) and Vasa (green) to reveal cell boundaries and germ cells. Note that the number and position of germ cells in hemizygous mutant ovaries are very similar to those of the heterozygous control larvae. Scale bar = 15 μm in B (applies to A,B).

To examine the onset of oogenesis itself, we examined the development of mutant pupal ovaries at various intervals: immediately after puparium formation and 12 hr, 24 hr, and 48 hr later. Due to the time required for germline stem cell divisions and cystoblast development, these intervals should be sufficient to monitor every major event in germline proliferation, differentiation, or cell death. Under these conditions, successful stem cell divisions would be detected as an increased number of germline cysts. In phenotypically wild-type ovaries from pag/TM6C heterozygotes examined 1–2 hr after puparium formation, germline stem cell had initiated asymmetric divisions and germ cells had increased in number (Fig. 3A). By contrast, in mutant ovaries, the number of germ cells did not significantly increase with respect to the third larval stage (Fig. 3B). Twelve hours after pupation, the ovaries from the wild-type heterozygous sibling females had partitioned into individual ovarioles, all of which contained stem cells and early germline cysts, mostly at the four-cell stage (Fig. 3C). Actin and Vasa staining showed that the germ cells in mutant pupal ovaries had partitioned into the normal number of distinct ovarioles (Fig. 3D; 15.3 ± 5.1), although 15% of the mutant ovarioles already completely lacked germ cells. The remaining mutant ovarioles contained several germline cells ranging from one to two to six to eight, but we never found more than two-cell cysts at this stage. Approximately 24 hr after puparium formation, the germaria from pag/TM6C pupae were filled with germ cells (Fig. 3E). By contrast, in pag/Df(3L)vin7 ovaries the percentage of empty ovarioles had increased to 24%, and, when present, the germ cell number within each germarium did not exceed six to eight (Fig. 3F). By 48 hr after pupation, the control ovaries had well-developed germaria and stage 1 egg chambers had budded off the germarium (Fig. 3G). pag mutant ovaries, however, showed no increase in germline cell number as the pupae increased in age; rather, the percentage of empty ovarioles increased to 33% (Fig. 3H). The observation that germline cells in the germarium do not increase in number with time suggests that they are unable to undertake the cell division cycle as they do in the wild-type ovary. This suggestion might be consistent with the failure to form normal 16-cell cysts in the germarium. As a milder manifestation of this defect, ovaries aged 48 hr or more occasionally (0.3%) contain next to the stem cells a cluster of small undifferentiated cells that failed to become encysted (not shown). The decrease in germline cells as the female ages, might be also consistent with an increased cell death. In support of this possibility, we observed that pag mutant germ cells often had irregular size and uneven Vasa staining. By contrast, wild-type germ cells were of consistently large size, with smooth surface and evenly stained with the anti-Vasa antibody. Frequently, we observed in pag mutant ovaries Vasa-positive cell debris, indicating that germ cells underwent degeneration (not shown).

Figure 3.

pag is required for germline maintenance during pupal development. Ovaries from pag/TM6C control (A,C,E,G) and pag/Df(3L)vin7 (B,D,F,H) pupae were stained for actin (red) and Vasa (green). A: In control ovaries fixed 1–2 hr after puparium formation, the stem cells have proliferated to form a large cluster. B: Mutant germ cells have not increased significantly in number and form a smaller subequatorial cluster. C,D: Twelve hours after pupation, the ovaries in pag/TM6C pupae (C) have partitioned into individual ovarioles that contain stem cells (arrowheads) and two- and four-cell cysts (arrows), whereas the pag/Df(3L)vin7 ovarioles (D) contain fewer germ cells and only two-cell cysts (arrows). E,F: Twenty-four hours after puparium formation the germ cells have proliferated and filled the germarium in heterozygous control ovaries (E), whereas the germ cell number has decreased further in hemizygous mutant ovaries (F). G,H: By 48 hr after pupation, the control ovaries (G) have differentiated germaria and stage 1 egg chambers (arrows), whereas the number of germ cells in the mutant ovaries (H) is reduced yet further and many of the ovarioles, although well differentiated, are devoid of germ cells. Note that the pattern of proliferation of the somatic cells that form the terminal filament does not differ between pag/TM6C and pag/Df(3L)vin7 pupal ovarioles (tf, terminal filament cells). Scale bar = 15 μm in H (applies to A–H).

The proper execution of cytokinesis in germ cells can be unambiguously detected by the presence of ring canals. Because ring canals are formed by stabilization of the contractile rings after cytokinesis, they can be readily observed by antibodies against the kinesin-like protein encoded by pavarotti (Pav-KLP) that is associated with the central region of the spindle at telophase and is required for the correct organization of the spindle during cytokinesis (Adams et al., 1998). Pav-KLP is, indeed, a component of the ring canals in both testes (Carmena et al., 1998) and ovaries (Minestrini et al., 2002). Strong Pav-KLP staining was found in wild-type control ovarioles at the midbody of the just dividing germline stem cells, whereas a feeble signal was observed at the ring canals connecting cystoblasts and cystocytes (Fig. 4A). The antibody against Pav-KLP also recognize ring canals in mutant germaria (Fig. 4B), indicating that cystoblasts could be formed and undergo at least one incomplete cytokinesis.

Figure 4.

Mutant germ cells can form ring canals. A,B: Germaria from pag/TM6C controls (A) and pag/Df(3L)vin7 mutants (B) were stained with anti–Pav-KLP antibody. A: The Pav-KLP gene product accumulates at the midbody regions of dividing stem cells (arrow) and cystoblasts (large arrowheads) and at the ring canals that interconnect dividing cystocytes (small arrowheads) in the pag/TM6C germarium. B: Ring canals in mutant germaria (arrowhead) are also stained by the anti–Pav-KLP antibody. Scale bar = 10 μm in B (applies to A,B).

Fusome Organization in pag Ovarioles

The failure of stem cell maintenance in pag mutants is not due to secondary defects of abnormal ovary development, because the mutant ovarioles show normal morphology during larval, pupal, and adult stages, although they are smaller than wild-type ovarioles. The normal number of terminal filaments are formed in the mutants, so that during pupal development, the ovary differentiates normally, partitioning the stem cells into individual ovarioles as in wild-type pupae.

A characteristic feature of germ cells is the spectrosome/fusome, a large organelle containing several cytoskeletal components, including actin. This organelle might reflect the underlying intracellular machinery for the asymmetric division. We analyzed, therefore, by serial confocal sectioning the apical germarium region in both pupal and adult mutant ovarioles, stained with Rh-phalloidin to outline cell boundaries and to mark spectrosomes in stem cells and cystoblasts and to label branched fusomes in germline cysts. pag stem cell spectrosomes were often displaced from the stereotypical position they occupy in control ovarioles (Fig. 5A,B); we frequently (67%; n = 351) observed round spectrosomes in abnormal positions, far from the apical germ cell membrane (Fig. 5C,D). However, two-cell ovoid fusomes connect dividing cystoblasts in both control heterozygous pag/TM6C (Fig. 5A) and pag/Df(3L)vin7 (Fig. 5C) ovarioles. We also observed elongated spectrosomes that were asymmetrically shared between daughter germ cells in mutant ovarioles (Fig. 5E,F). Although these spectrosomes were often irregularly shaped, most of their body lags in the apical cell as usually occurs in wild-type germarium during late telophase of the asymmetric divisions of germ stem cells. A total of 43% (n = 127) of the elongated spectrosomes found in mutant germ cells had their proximal extremity closely associated to the terminal filament cells (Fig. 5F), as usually occurs in control ovaries, whereas the remaining 57% lacked such connections (Fig. 5E).

Figure 5.

Fusome organization in pag ovaries. A–F: Detail of the apical region of pupal ovarioles both from pag/TM6C controls (A,B) and pag/Df(3L)vin7 mutants (C–F), stained for actin (A,C,E,F) or Vasa (B,D), fixed 24–30 hr after puparium formation. A,B: Detail of the apical region of a control ovariole containing a round fusome, or spectrosome, close to the terminal filament cells (arrow) and two ovoid fusomes. The more apical connects one stem cell and its daughter cystoblast (small arrowhead), the other is between two just-divided cystocytes (large arrowhead). C,D: Detail of the apical region of a mutant ovariole showing one stem cell with a round spectrosome that is displaced from the basal terminal filament cells (arrow), one cystoblast (large arrowhead), and two two-cell cysts connected by ovoid fusomes (small arrowheads). E,F: Detail of mutant germaria showing stem cells with round fusomes in abnormal position (arrow) or elongated wild-type–like fusomes between dividing stem cells and cystoblasts (arrowheads). Note that the proximal end of the fusome in F is attached to the terminal filament cells, whereas the fusome in E is shorter and not connected to the filament cells. Scale bar = 10 μm in F (applies to A–F).

pag Is Also Required in the Male Germline

Heterozygous control third instar larval testes were wild-type in both size and appearance and contained many cysts with 16 primary spermatocytes and some older cysts that had progressed into meiosis (Fig. 6A). The branched fusomes connecting dividing germ cells indicated that spermatogonia underwent amplification (Fig. 6B). pag/Df(3L)vin7 testes at this stage were invariably smaller than controls and devoid of large germline cysts (Fig. 6C). A total of 50–60 germ cells were present in a cortical monolayer in the dorsal region of the mutant testes, and only two-cell fusomes were seen (Fig. 6D). During pupal development, the germ cells in the mutant testes did not change their position inside the testis but their number gradually decreased. We found 25–36 germ cells in testes dissected 24 hr after puparium formation (n = 91, Fig. 6E). This finding contrasts with the development of pupal testes in pag/TM6C heterozygotes, where germline stem cells divide repeatedly to form more cysts that undergo spermatogenesis. Testes from newly emerged heterozygous control flies contained germline cells at all stages of spermatogenesis, including germline stem cells at the apical tip as well as numerous bundles of mature sperm (Fig. 6F). In contrast, newly emerged hemizygous mutant males (Fig. 6G) had shorter and thinner testes with very few germ cells randomly disposed inside the gonad (we find from 2 to 15 germ cells in each testis). We occasionally found unbranched fusomes connecting three to four cells in testes from newly emerged pag/pag males (not shown). This finding suggests that the male stem cells are occasionally able to divide to generate a few cells that appear to take on a spermatogonial fate but fail to complete the normal four rounds of division, resulting in abnormal germline cysts consisting of a few or even individual germ cells, that further degenerate. The spermatogonial nature of these cells is further supported by the position they occupied far from the tip of the testis, the usual location of germline stem cells. The number of germ cells dramatically decreased in aged mutant males (Fig. 6H), and all testes examined from 10-day-old pag/pag (n = 178) and pag/Df(3L)vin7 (n = 212) males lacked germ cells when they were stained with anti-Vasa.

Figure 6.

A–H: Testes from pag/TM6C control (A,B,F) or pag/Df(3L)vin7 (C,D,E,G,H) males were dissected and prepared to visualize actin (red) and germ cells (green) in A,C,E; actin alone in B,D; or microtubules (green) and germ cells (red) in F,G,H. A,B: Testes from third instar control larvae (A) are filled with germ cells at various stages of gametogenesis that are groped in distinct cysts, as revealed by the branched fusomes (B, arrows). C,D: Germ cells form a continuous monolayer at one hemisphere of mutant testes in third instar larvae (C); fusomes are short and connect only germ cell pairs (D, arrow). E: The number of germ cells is reduced in testes from mutant pupae. F: Testes from newly eclosed control males contain many germ cells at various stages of differentiation, including mature sperm cells (arrow), and the germline stem cells are located at the tip of the testis. G,H: In contrast, testes from newly eclosed mutant males have few germ cells (G) that decrease in number in 5-day-old males (H). Scale bar in H = 30 μm in A,C, 80 μm in B,D, 15 μm in E–H.


In this study, we have shown that mutations in the pag locus lead to female and male sterility by disrupting gametogenesis at very early stages. Germ cell viability and proliferation seem to progress normally to the time at which the ovary or testis form. Once gonad formation begins, pag germ cells begin to disappear.

The loss of germline over time and the formation of empty ovarioles is a phenotype shared with several other female sterile mutations and could reflect defects in stem cell maintenance after their differentiation without self-renewing or mitotic defects that lead to delay or arrest in division and subsequent degeneration (for review, see Deng and Lin, 2001). These mutations include lesions in the cell-autonomously required gene pumilio (pum; Lin and Spradling, 1997; Forbes and Lehmann, 1998) or defects in genes involved in extrinsic signal pathways such as fs(1)Yb (Yb) (King and Lin, 1999), piwi (Cox et al., 1998), and decapentaplegic (Xie and Spradling, 1998; Chen and McKearin, 2003b). All the above mutations are not essential for the further differentiation of the cystoblasts and give an ovary phenotype in which germline stem cells fail to be maintained and the remaining germ cells in the germarium take a cystoblast fate and, ultimately, develop in egg chambers. By contrast, function of bam (McKearin and Spradling, 1990) and benign-gonial-cell-neoplasm (bgcn; Gateff, 1982) is required to regulate stem-cell-to-cystoblast differentiation and mutant ovarioles lack cystoblasts and cystocytes. However, the germ cells lacking bam or bgcn become trapped in a stem-like state and proliferate without the formation of syncytial clusters to produce distinctive tumorous germ cell phenotypes. We reasoned that, if pag stem cells differentiate without self renewing, we should have found ovarioles devoid of stem cells at the tip of the germarium, but with a few dividing cystocytes and egg chambers. Because ovarioles with intervening agametic regions are never observed in pag females, we conclude that the elimination of the stem cells that leads to empty ovarioles cannot be attributed to a differentiation process without self-renewing. We favor the hypothesis that the pag gene product could be primarily required for germ cell division. We cannot rule out, however, that pag could be also required for germ cell survival. An increased cell death, therefore, could cancel out the effect of germ cell proliferation in the pupal ovaries.

The average number of germ cells in mutant ovarioles from late pupae and adult females (Table 1) suggests that, in both pag/pag and pag/Df(3L)vin7 genotypes, the half-life of the germ cells once in the ovary is approximately 5 days. Moreover, the pag/pag genotype shows a slightly higher germ cell average than the pag/Df(3L)vin7, in agreement with the milder effect of the mutation in homozygotes. One difference between stem cells and their differentiating daughters is that the stem cells divide less often (on average, in wild-type, the cystoblast daughter will have undergone four mitotic divisions before its stem siblings will divide again). This finding suggests that division per se is the risk factor here; the more times a lineage has divided since some prior event, the more likely the products of that division will die (or at least cease to be recognized by the antibody used for this study). That there are any germ cells at all, and some successful divisions, suggests that the germ cell population is initially competent but loses that competence with subsequent cell divisions. The presence of only stem cells in many ovarioles of newly eclosed mutant females argues that stem cells either fail to divide or divide only a few times early in their life span to generate a few cystoblasts that degenerate. On the other hand, the finding that two-cell stage cystoblasts usually do not persist beyond the first division indicates that the pag gene product is also required during and after the cystocyte division. This is consistent with the failure to detect branching fusomes that in wild-type ovaries mark multicellular syncytia. In pag/pag females, in which there is a milder effect of the mutation, a few clusters of small Vasa-positive cells are occasionally observed in the germaria of some ovarioles, in addition to apically located stem cells. These clusters, however, never form true egg chambers, surrounded by follicle cells.

In addition to asymmetric divisions, germline stem cells can divide symmetrically in wild-type to replace any stem cells that are lost (Margolis and Spradling, 1995). While asymmetric divisions are required to self-renew stem cells and generate cystoblasts, symmetric divisions are essential for maintaining the proper number of stem cells. Replacement of stem cells must function efficiently in wild-type ovaries because the rate of stem cell loss does not increase with age (Xie and Spradling, 2000). The observation that many pag ovarioles have only one stem cell instead of the usual two or three suggests that replacement of lost stem cells by symmetric division does not occur efficiently in pag mutants.

Germline stem cells execute self-renewing and asymmetric divisions in close association with somatic cells that form a niche (Spradling et al., 2001). Stem cells contact the cap cell membrane by adherens junctions (Song et al., 2002), so that the daughter that inherits those junctions stays at the tip as a stem cell, maintaining high levels of Dpp signal (Kai and Spradling, 2003), the other daughter floats loose and differentiates into a cystoblast. It could be argued that defects in maintaining the close association between stem cells and cap cells could modify the anatomical asymmetry of the germarium tip that ensures that equivalent stem cell daughters receive different fate-determining signals. Perhaps, cells slightly displaced from the terminal filament do not reach sufficiently high levels of extrinsic signals to activate the pathways required to specify self-renewal and continuous maintenance of the germline stem cell population (Lin, 1998). Therefore, stem cells would gradually lose their identity and function and stop dividing, remaining quiescent at the tip of the germarium. This alteration in stem cell behavior could have repercussions on their daughters cystoblasts, that also display severe defects of proliferation and differentiation. Like pag, adult zeropopulation growth (zpg) -null gonads contain a few germ cells and lack later stages of germ cell differentiation. The zpg locus encodes a germline-specific gap–junction protein that has been proposed to mediate the passage of signals between somatic and germ cells required for survival of early germ cells during gametogenesis of both sexes (Tazuke et al., 2002).

pag is unusual among genes involved in germline stem cells maintenance, because both oogenesis and spermatogenesis are similarly affected. Three genes, bam, bgcn, and piwi, have been described to be required for germline differentiation in both male and female (McKearin and Spradling, 1990; Gonczy et al., 1997; Cox et al., 1998). The last one, the piwi gene, is also involved in germline stem cell maintenance. Lesions in these genes either lead to aberrant spermatocyte differentiation or to the formation of an abnormal number of sperm cysts. By contrast, the young pag testes contain only a few scattered spermatogonial germ cells, and aged males contain none. So, a premature loss of stem cells in pag accounts for the lack of germ cell lineage in adult testes and suggests that this gene is involved in cell-cycle progression and/or maintenance of stem cells and their daughters. That the pag male phenotype parallels the female counterpart suggests that the pag gene could be required at earlier steps of germline maintenance in both sexes. However, the defects of germ cell proliferation are not identical. There is a loss of germ cells in early pupal male gonads, although little reduction in the number of germ cells is seen in the similar stage of female gonads. This finding may indicate that the pag gene product is required at different times during male and female gonad development. Consistently, both male and female wild-type gonads initiate their differentiation during the transition of larvae to pupae, but stem cell specification and maintenance seem to follow different pathways (Xie and Spradling, 2000; Kauffman et al., 2003). The apparent stochastic nature of the germ cell loss in both pag males and females might be explained by random events during germ cell division or residual activity from a partial-loss-of-function allele.


Drosophila Strains and Cytological Mapping

pag locus was identified in a screen of a subset of a collection of mutants that showed female sterility (Deak et al., 1997). We named this mutant parva germina (pag), a Latin term that means “few germ cells,” after the cytological phenotype described here. The pag mutation was kept over the TM6C balancer that carries the body-shape marker Tubby (Tb), allowing identification of homozygous pag larvae and pupae. To map pag, we used a deficiency set on the third chromosome obtained from the Umeà Stock center. pag/TM6C flies were crossed to flies from each pertinent deficiency stock, and both the mutant/Df males and females from each cross were tested for fertility. Fly culture and crosses were performed at 25°C on standard medium according to standard procedures; dissections were performed at room temperature. Oregon R and pag/TM6C were used as controls.


A mouse monoclonal anti–β-tubulin (Boehringer, Mannheim, UK) was used at 1:200 dilution; a rabbit polyclonal anti–Pav-KLP polyclonal Rb3301 (Adams et al., 1998) at 1:100; a rabbit polyclonal anti-Vasa antibody (Lasko and Ashburner, 1990) at 1:400. Goat anti-mouse or anti-rabbit secondary antibodies coupled to fluorescein or rhodamine (Cappel, West Chester, PA) were used at 1:600 dilution. DNA was visualized with Hoechst 33258 (Sigma, St. Louis, MO). The actin cytoskeleton was visualized with 10 units/ml of rhodamine-phalloidin (Molecular Probes). Bovine serum albumin (BSA) was obtained from Sigma.

Fluorescence Preparations

Ovaries and testes from Oregon R, pag/TM6C, pag/pag, and pag/Df(3L)vin7 flies were dissected into Drosophila Ringer's or phosphate buffered saline (PBS). Female larvae were recognized by the size of their gonads, and the age of female pupal ovaries was determined according to King et al. (1968). We used two different fixation protocols. To visualize microfilaments and germ cells simultaneously, we fixed gonads in 4% formaldehyde (methanol-free, Pella Scientific) for 30 min. The samples were then transferred to 1% Triton X-100 in PBS for 10 min, rinsed 3× in PBS, and incubated for 60 min in 0.1% PBS+BSA (PBS-BSA). Gonads were then incubated for 5–6 hr at room temperature in the anti-Vasa antibody. After washing in PBS-BSA, the gonads were incubated for 1 hr with a fluorescein-coupled goat anti-rabbit to which rhodamine-phalloidin was added.

Alternatively, after dissection, gonads were transferred to cold methanol for 10 min and then into cold acetone for 5 min. After a PBS wash, the gonads where incubated for 1 hr in PBS containing 0.1% BSA. For double staining of microtubules and germ cells gonads were incubated overnight at 4°C with the anti-Vasa antibody, then the anti–β-tubulin antibody was added and the incubation proceeded for 4–5 hr at room temperature. After washing in PBS-BSA, the samples were incubated for 1 hr with the appropriate secondary antibodies. Controls of the secondary antibodies alone were done for all preparations. For localization of Pav-KLP, ovaries were incubated overnight with the polyclonal Rb3301 serum. For simultaneous DNA staining, the samples were incubated 5 min in 1 μg/ml Hoechst 33258. Samples were then rinsed 3× in PBS and mounted in small drops of 90% glycerol containing 2.5% n-propyl-gallate.

Confocal Microscopy

Digital optical sections of whole-mount ovaries and testes were examined by using a Leica TCS 4D laser scanning confocal microscope equipped with an argon-krypton Laser and coupled to a Leica DMRBE microscope equipped with 63× PL Apo 1.4 objectives (Leica Lasertechnik, Heidelberg). For double staining, the images of the two fluorochrome distributions were recorded separately by averaging 8–16 scans of a single optical section to improve the signal/noise ratio. Images collected at several focal planes were superimposed, merged into a single file and imported into Adobe Photoshop to adjust size and contrast.


We thank A.T. Carpenter for cytological mapping of the pag locus, critical reading of the manuscript, and helpful comments. We also thank P. Lasko for kindly providing the anti-Vasa antibody. D.M.G. was funded by a Medical Research Council grant and M.G.R. received a grant from PAR (University of Siena).