nanos1 is required to maintain oocyte production in adult zebrafish

Abstract

Development of the germline requires the specification and survival of primordial germ cells (PGCs) in the embryo as well as the maintenance of gamete production during the reproductive life of the adult. These processes appear to be fundamental to all Metazoans, and some components of the genetic pathway regulating germ cell development and function are evolutionarily conserved. In both vertebrates and invertebrates, nanos-related genes, which encode RNA-binding zinc finger proteins, have been shown to play essential and conserved roles during germ cell formation. In Drosophila, maternally supplied nanos is required for survival of PGCs in the embryo, while in adults, nanos is required for the continued production of oocytes by maintaining germline stem cells self-renewal. In mice and zebrafish, nanos orthologs are required for PGC survival during embryogenesis, but a role in adults has not been explored. We show here that nanos1 in zebrafish is expressed in early stage oocytes in the adult female germline. We have identified a mutation in nanos using a reverse genetics method and show that young female nanos mutants contain oocytes, but fail to maintain oocyte production. This progressive loss of fertility in homozygous females is not a phenotype that has been described previously in the zebrafish, and underlines the value of a reverse genetics approach in this model system.

Introduction

Eggs and sperm are derivatives of embryonic germ cells called primordial germ cells (PGC). In most, if not all animals, PGCs are specified at extragonadal locations and proliferate en route to and within the embryonic somatic gonad. In males of most species, PGCs give to germline stem cells that are the source of continual sperm production throughout the reproductive life of the adult (reviewed in Lin, 2002). In females the fate of embryonic germ cells appears to be species dependent. In mammals it is generally accepted that a limited expansion of germ cell numbers occurs during embryogenesis and prior to birth all germ cells enter meiosis and arrest in the diplotene stage of the first meiotic prophase. As such, the ovaries of juvenile and adult mammals contain only post-mitotic germ cells and therefore have a determinate number of mature eggs they can produce (Franchi et al., 1962).

In contrast to mammals, there is histological evidence that the adult ovaries of many non-mammalian vertebrates, such as bony-fish (teleost) and amphibians, contain both mitotic germ cells, called oogonia, as well as maturing oocytes (reviewed in Wallace and Selman, 1990; Chaves-Pozo et al., 2005). Early germ cells can readily be identified from other cell types in electron micrographs based on their size, localization within a germinal epithelium, and because they uniquely contain germ cell-specific cytoplasmic granules, called nuage (Grier, 2000). It is hypothesized, though not directly proven, that in these species oocytes are either cyclically or continually produced in adults by oogonial stem cells. A stem cell-based mechanism for oocyte production could readily explain the extraordinary fecundity of fish such as the halibut, which can produce several hundred thousand eggs per spawning season (Norberg et al., 1991), or for other species, such as the zebrafish, that can produce eggs year-round (Selman et al., 1993). Thus, in contrast to mammals, the ovaries of some species of fish and amphibians appear to have an indeterminate number of eggs they can produce. However little is known about the genetic regulation of oocyte production in any vertebrate.

In Drosophila, oocyte production is known to be a stem cell driven process and mutational analysis has identified several germ cell autonomous factors that are required for continued oocyte production in adults (reviewed in Gilboa and Lehmann, 2004b). Among these, mutations in the nanos locus result in females that produce a limited number of functional oocytes, but fail to maintain oocyte production (Forbes and Lehmann, 1998; Bhat, 1999). nanos encodes an RNA binding protein that, together with Pumilio, negatively regulates the translation of hunchback mRNA in the posterior of the early embryo (Barker et al., 1992; Tautz and Pfeifle, 1989). In addition, nanos is also required cell autonomously in PGCs for their maintenance in early embryos (Kobayashi, 1996; Forbes and Lehmann, 1998; Deshpande et al., 1999), and in adult ovaries is expressed in the germline stem cells (Forbes and Lehmann, 1998). It has therefore been proposed that in adult ovaries, nanos functions to maintain GSCs, and thus oocyte production, by negatively regulating genes that promote GSC to differentiate into oocytes (Gilboa and Lehmann, 2004a; Wang and Lin, 2004). The role of nanos in regulating PGC survival appears to be evolutionarily conserved as nanos orthologs in both mice (Tsuda et al., 2003) and zebrafish (Köprunner et al., 2001), are required for survival of PGCs during early embryogenesis. By contrast, a role for nanos orthologs in adult gonads in an organism other than Drosophila has not been described.

We show here that in zebrafish, nanos1 (nos1) is expressed in early stage germ cells in larval and adult ovaries. To address the function of nos1 in germline development and function, we used the TILLING reverse genetics technique to identify an ENU-induced point mutation that results in a probable loss-of-function allele (Draper et al., 2004). We determined that nos1 homozygous mutants derived from heterozygous parents are viable and have normal PGC survival in early embryos, indicating that maternally provided nos1 function is sufficient for this process. In contrast the ovaries of 2.5 month old nos1 mutant females contain late stage oocytes, but are devoid of early stage oocytes, and the gonads of 6 month old adult females are devoid of all oocytes. Thus, similar to Drosophila, nos1 in zebrafish is required to maintain oocyte production in adult ovaries.

Materials and Methods:

Fish strains.

The wild-type strain used is *AB. Prior to characterizing the mutant phenotype, nanos1(fh49) carrier fish were outcrossed four times to wild-type fish, before crossing heterozygous parents to generate homozygous mutants.

Genotyping.

The fh49 allele eliminates an MseI site in the wild-type sequence nanos1 sequence. This polymorphism was used to genotype all fish used in this study prior to analysis as follows: DNA was extracted from caudal fin tissue amputated from anesthetized fish by incubating tissue in 50μl DNA extraction buffer (10mM Tris pH 8.4, 50mM KCl, 1.5 mM MgCl₂, 0.3% Tween-20, 0.3% NP-40) for 20 min at 95°C, cooled to 55°C prior to adding 5μl of 10 mg/ml Proteinase K, incubated for 1 hr at 55°C followed by 20 min at 95°C. A 157 base pair fragment of the nanos1 gene containing the MseI polymorphisms was amplified by polymerase chain reaction (PCR) from individuals using the following primer pair (forward/reverse):AGACTGAGGCCGTGTACACCTCTCACTACT/GAGCAGTAGTTCTTGTCCACCATCG, in a 10μl PCR reaction using the following amplification parameters: 95°C 1 min, followed by 35 cycles of 95°C, 10 sec.; 60°C, 10 sec; 72°C, 15 sec. Following PCR, 20μl of MseI digestion buffer (1X MseI buffer supplied by manufacture, 0.5u MseI/μl, 1 mg/ml bovine serum albumin), and incubated at 37°C for a minimum of 4 hrs. MseI digest the wild-type fragment into 125 bp and 32 bp fragments. Fragments were then separated by gel electrophoresis through 2% agarose. Post-in situ embryos were genotyped as above, except DNA extraction was preformed with 1/5 the volumes.

Whole-mount RNA in situ hybridization and immunostaining.

Gonads for RNA in situ and immunostaining were prepared as follows: Adult fish were euthanized with an overdose of Tricaine (Sigma) followed by submersion in ice water for 10 min. Prior to overnight fixation at 4°C in 4% paraformaldehyde/1X Phosphate buffered saline at 4°C, the heads were removed and the body cavities were opened along the ventral midline. After fixation, intact fixed gonads were dissected from the fish. RNA in situ hybridization was preformed as described previously (see http://zfin.org/zf_info/zfbook/chapt9/9.82.html) except that gonads were treated with 50 μg/ml Proteinase K, 15 min. at 37°C prior to hybridization as described in Onichtchouk et al. (2003). Riboprobes for in situ hybridization were generated as previously described: nanos1 (Köprunner et al., 2001); vasa (Yoon et al., 1997); erg2b (Oxtoby and Jowett, 1993); ziwi (Tan et al., 2002) myod (Weinburg et al., 1996). Embryonic staging was according to Kimmel et al. (1995). For double-labeling for nos1 RNA and Vasa protein, antibody staining was carried out after RNA in situ hybridization essentially as described previously (Amacher et al., 2002). Anti-Vasa antibody (Knaut et al., 2000) was used at 1:5,000.

Reverse-transcription Polymerase chain reaction (RT-PCR).

Tissue was dissected from adult zebrafish and RNA was isolated as previously described (37). RT-PCR was as previously described (Draper et al., 2001) using the following primer pair for nos1 (forward/reverse): AGACTGAGGCCGTGTACACCTCTCACTACT/GAGCAGTAGTTCTTGTCCACCATCG, and odc (forward/reverse): ACACTATGACGGCTTGCACCG/CCCACTGACTGCACGATCTGG.

Microinjection into zebrafish embryos.

Templates for synthesis of nos1 (wt) and nos1(fh49) mRNA were generated by PCR and cloned into the CS2+ plasmid. Synthetic nos1 (wt) and nos1(fh49) mRNA was then prepared using the mMessage mMachine kit (Ambion) and was diluted in 100mM KCl prior to injection into 1-cell stage embryos (quantity specified in Fig. 3 legend). For experiment shown in in Fig. 3 panels G and H, embryos were obtained from a female with the genotype nos1(fh49); Tg[vas:EGFP], which expresses EGFP in germ cells (Krovel and Olsen, 1997).

Fig. 3.

Maternal nos1 is required for primordial germ cell survival. (A) Numbers of germ cells in embryos at the stages identified on the X axis as assayed by in situ hybridization using a vasa probe. Shield and 14-somite stage embryos were derived from a cross between nos1^−/− females and nos1^+/+ males, while 17-somite stage embryos were derived from a cross between nos1^−/− parents. Standard deviations and numbers of embryos analyzed for each class are indicated. In legend, Mnos1⁻ indicates embryos derived from nos1^−/− females. PGC localization in representative shield-stage embryos derived from nos1^+/+ females (B) or nos1^−/− females (C) as reveled by vasa in situ hybridization (blue). PGC localization in representative 17-somite stage embryos stained for myod and egr2b in red, and vasa in blue, derived from nos1^+/+ females (D),nos1^−/− females (E) or nos1^−/− females injected with 200 pg synthetic nos1 mRNA (F). Arrows point to correctly localized vasa expressing PGCs, while arrowheads indicate ectopically localized PGC’s. Germ cell development in Mnos1 embryos can be rescued by injection of nos1(wt) RNA(G; n=12/12) but not nos1(fh49) mutant RNA (H; n=0/23) at the one cell-stage, as assayed in a 24 hpf embryo. Arrow in G points to EGFP expressing germ cells (see materials and methods). Scale bars: 100μm.

Histology.

Adult zebrafish were sacrificed with an overdose of Tricaine (Sigma), their heads and tails were removed and their torsos fixed for two days at room temperature in either Bouin’s Solution (Sigma) or Dietrich’s Fixative (30% EtOH, 10% formalin, 2% glacial acetic acid), as previously described (Moore et al., 2002). Fish were then paraffin embedded following standard protocols. 7 μm thin sections were cut, deparaffinized in Histo-Clear (National Diagnostics) and stained with hematoxylin and eosin. For in situ hybridization on paraffin sections (Fig. 1A), fish were fixed for three days at room temperature in Larison’s Fixative (30% EtOH, 10% formalin, 0.05 M sodium phosphate, pH 7.2); K. Larison and J. Mathews, personal communication). Prior to paraffin embedding, fish were incubated for 2 days in Decalcification Solution (30% EtOH, 3% formalin, 0.2 M EDTA, pH 7.5), which was changed twice daily (K. Larison and J. Mathews, personal communication). 7-μm sections were cut, depariffinized in Histo-Clear (National Diagnostics) and re-hydrated into phosphate buffered saline a graded EtOH series. Slides were process essentially as described above for whole-mount in situ hybridization.

Fig. 1.

nanos1 is expressed in adult female germ cells. RNA in situ hybridization reveals that at 21 dpf (A and B) and 35 dpf (C and D). nos1 expression (A and C) appears to be restricted to perinuclear stage oocytes (arrows) as compared to vasa (B and D) that appears to be expressed in both perinuclear stage oocytes (p.o.) as well as germ cells that are less than 20μm in diameter (arrow in D). (E) RNA in situ hybridization of a transverse section through a 2 month old female zebrafish reveals that nos1 is expressed at high levels in early stage oocytes (oriented dorsal up). In stage Ib oocytes, nos1 transcripts are enriched in a spherical cytoplasmic structure that resembles the Balbiani body (arrow). (F) In adult ovaries, vasa expression reveals that early stage oocytes localize to a ventral zone on the ovarian surface (arrow; ovary is oriented with anterior to the left and dorsal up). (H) Higher magnification of region boxed in (F) reveals an overall organization of early stage oocytes within this ventral zone, with stage Ia oocytes located dorsal to stage Ib oocytes. Similar to nos1, vasa RNA is also enriched in spherical cytoplasmic structures within stage Ib oocytes (arrow). (G) nos1 is highly expressed in stage Ib oocytes, but is either not expressed in clusters of stage Ia oocytes that localize to the ventral zone, or is expressed at levels below our ability to detect (Compare G and H). nos1 expression can not be detected in testes by in situ hybridization (I) while vasa expression can be readily detected in developing spermatocytes (J). (K) RT-PCR analysis confirms that nos1 expression in adults is female specific. ornithine decarboxylase (odc) expression analysis is included as a positive control. (L) A confocal image of an ovary double-labeled for Vasa protein (green) and nos1 RNA (red) show that nos1 and Vasa co-localize in early stage oocytes (Ib) but not in the smallest (<20μm) Vasa-positive germ cells (arrow). Scale bars: 50 μm in A-D, L; 200 μm in E, I and J; 1 mm in F; 100 μm in E and F.

Results

Expression of nanos1 in the Adult Female Germline

Zebrafish nanos1 (nos1) has previously been shown to encode a maternally expressed mRNA that localizes to primordial germ cells (PGC), and their precursors, during the early cleavage stages of embryogenesis where it functions to promote PGC survival (Köprunner et al., 2001). We analyzed the expression of nos1 in larval and adult zebrafish by in situ hybridization and compared it to the expression of vasa, a gene that appears to be expressed in germ cells at all stages (Yoon et al., 1997). We found that prior to 21 day post-fertilization (dpf), we could not detect nos1 by in situ hybridization (data not shown). However, beginning at 21 dpf, we could detect nos1 expression in early stage oocytes (Fig. 1A). In contrast to nos1, vasa is expressed in hundreds of germ cells in 21 dpf gonads, which include early stage oocytes as well as smaller, presumptive undifferentiated germ cells (Fig. 1B).

In zebrafish, sex is determined by an unknown, non-chromosomal mechanism. Prior to 25 dpf, the gonad is bipotential, yet beginning around 10 dpf, early stage oocytes can be found in all gonads (Uchida et al., 2002). While the oocytes produced by presumptive females continue to mature, the few oocytes formed in presumptive males arrest development, undergo apoptosis and are cleared from the gonads by 30 dpf (Uchida et al., 2002). By 35 dpf, ovaries can be readily distinguished from testes owing to their larger size, characteristic asymmetric morphology and the presence of numerous perinucleolar stage oocytes (Uchida et al., 2002; stage Ib according to Selman et al., 1993). At this stage, nos1 expression continues to be oocyte specific (Fig. 1C) in contrast to vasa that appears to label both oocytes and numerous smaller (<20μm) germ cells. Finally, by sexual maturity (∼3 months of age), nos1 expression is female germline specific as assayed by both in situ hybridization (Figs. 1E, G, I) and RT-PCR (Fig. 1K). nos1 appears to be expressed throughout most stages of oocyte development, with the highest levels restricted to stage Ib oocytes (20−140 μm in diameter; Selman et al., 1993) as compared to later stage oocytes (Figs. 1E, G). However, it is possible that the apparent differences in expression levels may simply be due to the large volumetric differences between early and late stage oocytes.

Using vasa expression to identify germ cells, we found that stage I oocytes are not randomly distributed in ovaries, but instead cluster in a narrow zone that is located on the lateral surface of the ovary in a strip that runs roughly parallel to the anterioposterior axis (Fig. 1F). When viewed at higher magnification (Fig. 1H), an overall organization within this region becomes apparent: a band of <20μm germ cells (stage Ia oocytes and putative oogonia; Selman et al., 1993) localize to the dorsal-most region, whereas stage Ib oocytes localize ventrally. This overall organization of stage I oocytes is not apparent in all ovaries we have examined and may reflect a specific stage during the oocyte production cycle. We can, however, identify a band of <20μm germ cells in most ovaries, suggesting the possibility that this organization may have functional significance. In contrast to vasa expression, nos1 does not appear to be expressed in <20μm germ cells, but is instead expressed in stage Ib and more mature oocytes (compare Figs. 1G and H).

We further investigated the co-localization of nos1 and vasa in early stage germ cells by co-labeling ovaries for nos1 RNA and Vasa protein and analyzed their patterns of co-localization using confocal microscopy. We found that while Vasa protein was present in both the <20μm germ cells and stage IB oocytes, we could only detect nos1 RNA in stage IB and later stage oocytes (Fig. 1L). We therefore conclude that either nos1 is not expressed in <20μm germ cells, or is expressed at levels that are below our ability to detect.

Identification of a Null Mutation in nanos1

To determine if nos1 function is required for normal germline development in zebrafish, we identified an N-ethyl-N-nitrosourea (ENU) induced mutation, called nos1(fh49), using the TILLING reverse genetic methodology (Draper et al., 2004; McCallum et al., 2000; Colbert et al., 2001). The nos1(fh49) allele contains a premature stop codon that is predicted to truncate the Nos1 protein midway through the first of two zinc finger RNA binding domains (Figs. 2A and B). Because the zinc binding domains are essential for nanos function in Drosophila (Curtis et al., 1997), we conclude that nos1(fh49) is likely a null allele (see below for functional evidence). The nos1(fh49) mutation disrupts an MseI restriction enzyme cleavage site, enabling the design of a polymerase chain reaction-based genotyping assay (Fig. 2C). Using this assay we identified fish heterozygous for the nos1(fh49) mutation and crossed them to generate homozygous mutants lacking zygotic nos1 function (refereed to below as Znos1). We determined that nos1(fh49) homozygous mutants are viable and survive to adulthood in the expected Mendelian ratios (+/+:+/−:−/− = 1.03:1.88:1.09, n=315; x²=1.3, P> 0.5).

Fig. 2.

The nos1(fh49) mutant allele encodes a truncated form of the protein. (A) Sequence traces of nos1 from wild-type (top) and nos1(fh49) heterozygotes revels that nos1(fh49) is a point mutation that converts the Leu(110) codon into a stop codon. The mutation disrupts an MseI restriction enzyme site present in wild-type. (B) Schematic of the predicted wild-type and mutant Nos1 protein structures. The CCHC zinc finger RNA binding domains are shown in yellow. (C) PCR-based genotyping assay utilizing the MseI restriction site polymorphism. nos1(fh49) heterozygotes have both wild-type (lower) and mutant (upper) bands.

Maternal nanos1 is required for PGC survival during early embryogenesis

In zebrafish, nos1 was originally identified as a maternal mRNA that localized to primordial germ cells (PGC) during the early stages (Köprunner et al., 2001). Using morpholino oligos to deplete embryos of maternal nos1 gene product, Köprunner et al., (2001) showed that nos1 was required for survival of PGCs. However, while most PGCs in nos1 morphants fail to migrate and apoptose, sufficient numbers of PGCs survive and localize correctly to the future gonad (Köprunner et al., 2001) such that most morphants develop into fertile adults (C.B.M. personal observation). Because it is possible that the incomplete penetrance of the morphant phenotype is due to either incomplete translational block of nos1 mRNA by the morpholino, or due to the presence of maternally translated Nos1 protein in the egg, we re-evaluated the requirement of maternal nos1 for proper PGC development using our genetic mutation. We mated zygotic (Z)nos1 females to wild-type males and determined the numbers and locations of PGCs in their offspring. We found that embryos derived from Znos1 females had normal numbers of PGCs at the shield stage (13±3, Figs. 3A-C), but at the 14-somite stage, had only 1±1 PGC compared to an average of 20±4 PGCs present in wild-type controls (Fig. 3A). Importantly, the occasional PGCs that we observed in 14-somite stage embryos derived from Znos1 mothers were always in extra-gonadal locations (data not shown). As anticipated from these results, embryos derived from Znos1 mothers developed into agametic adults that were phenotypically male (n=39/39), consistent with a recent report showing that in zebrafish, where the sex determining mechanism is non-chromosomal, germ cells are required for female, but not male, development (Slanchev et al., 2005). In contrast to embryos derived from Znos1 females, embryos derived from Znos1 males mated to wild-type females had normal numbers of PGCs at all stages examined (data not shown). Thus, nos1(fh49) mutation results in a strict maternal-effect sterile phenotype.

To confirm that the maternal effect-sterile phenotype we observed in nos1(fh49) mutants is due specifically to loss of wild-type nos1 gene product, we attempted to rescue this defect by microinjecting wild-type nos1 mRNA into 1-cell stage embryos lacking maternal nos1 (refereed to below as Mnos1). During normal development, maternally supplied nos1 mRNA localizes to presumptive PGCs in a process that is dependent on sequences in its 3’UTR (Köprunner et al., 2001). We therefore injected full length in vitro transcribed nos1 mRNA into 1 cell-stage Mnos1 zygotes and compared the numbers and locations of PGCs produced to those of uninjected siblings or embryos derived from wild-type parents. At the 17-somite stage, wild-type embryos contained on average 30±9 PGCs that were localized in two bilateral clusters in the lateral mesoderm adjacent to somites 3−5 (Fig. 3A, D), while Mnos1 embryos had on average 3±3 PGCs that were randomly distributed along the body axis (Fig. 3A, E). In contrast, Mnos1 embryos injected with wild-type nos1 mRNA at the 1 cell-stage had on average 26±8 PGCs that were correctly localized (Fig. 3A, F). Finally, injection of Mnos1 embryos with RNA derived from the nos1(fh49) allele did not rescue PGC survival (compare Figs. 3H and 3G). These data argue strongly that the maternal-effect sterile defect described above is due solely to the loss of wild-type nos1 gene product and that the nos1(fh49) allele is likely a null allele. Thus, eliminating maternal Nos1 protein genetically demonstrates that maternal nos1 is required for PGC survival, but not specification, in the zebrafish embryo.

nanos1 is Required for Maintaining Oocyte Production in Adult Females

While investigating the maternal function of nos1, we noticed that the numbers of eggs produced by Znos1 females declined dramatically in the first months after reaching sexual maturity. While normally female zebrafish are fertile for over two years, Znos1 females were completely sterile by six months of age. To understand the basis of the apparent Znos1 female infertility, we compared histological sections of six month old wild-type and Znos1 fish. Six month old wild-type female ovaries contain all stages of developing oocytes as well as mature eggs (Fig. 4A). In contrast, we could find no oocytes or eggs in the body cavities of six month old Znos1 females (Fig. 4B). Consistent with the female-specific expression of nos1, Znos1 males (Fig. 4D) had histologically normal testes when compared to those of wild-type males (Fig. 4C), and remained fertile throughout their lives.

Fig. 4.

Zygotic nos1 is required for adult germline maintenance. Histology of 6 month old females (A and B) and male (C and D) gonads. (A) Wild-type ovaries contain oocytes (stages indicated). (B) nos1 mutant females do not contain oocytes. The testes of wild-type (C) and nos1 mutants (D) are indistinguishable. 3 month-old wild-type (E, G and I), and nos1 mutant (F, H and J) ovaries freshly isolated (E and F), or stained for vasa (G and H) or ziwi (I and J) RNA. Wild-type ovaries contain numerous stage I oocytes, visualized as small (7−140μm diameter) clear cells on the surface of the ovary (E) or as small vasa RNA⁺ (G) and ziwi RNA⁺ (I) cells (arrows). In contrast, nos1 mutant ovaries contain very few stage I oocytes (F, H, J; arrows in H and J). RNA in situ hybridization to detect vasa RNA reveals that 35 dpf nos1 mutant ovaries (L) contain fewer <20 μm germ cells (arrow) than wild-type ovaries (K). sg, spermatogonia; sp, sperm; l, liver. Scale bars: 100 μm in A-D, K, L; 200μm in E-J.

To determine at what point defects in germline development could be detected in Znos1 females, we compared the development of wild-type and Znos1 gonads at various developmental stages. We found that while ovaries isolated from 2.5 month old wild-type females contain diverse stages of developing oocytes (Fig. 4E), ovaries isolated from nos1 mutant sibling females contained mostly late-stage oocytes (stage II-III) and few if any early stage oocytes (stage I; Fig. 4F).

We investigated this phenotype in more detail by comparing the expression patterns of both vasa and ziwi RNA in wild-type and Znos1 ovaries by in situ hybridization. ziwi encodes a zebrafish member of the Piwi/Argonaute family of RNA binding proteins (Tan et al., 2002), members of which have been shown to regulate the maintenance of stem cells in various organisms (Cox et al., 1998; Carmell et al., 2002). ziwi has previously been shown to be expressed in PGCs during early larval development (Tan et al., 2002), and like vasa, we found that ziwi is expressed at high levels in apparently all early stage germ cells in adults (compare Figs. 4G, I, and data not shown). Using these genes to identify germ cells, we compared the stages of oocytes present in wild-type and Znos1 ovaries isolated from three month old females. We found that in contrast to wild-type ovaries, Znos1 ovaries contained few if any stage Ia oocytes and were nearly devoid of stage Ib oocytes (Figs. 4G-J). This reduction of early stage oocyte numbers relative to late stage oocytes, as reveled by vasa expression, can be observed as early as 35 dpf (compare Figs. 4K and L), suggesting that nos1 is required early during gonad development to maintain oocyte production.

Discussion

Our results show that zygotic function of nos1 in zebrafish is required for the continued production of oocytes in adult females, but is not required for initial oocyte production or for oocytes to mature into fertile eggs. The early onset of sterility in females due to a failure to maintain oocyte production is strikingly similar to that observed for Drosophila nanos mutant females (Gavis and Lehmann, 1992), and suggest that a genetic program regulating both germ cell development during embryogenesis and germline maintenance in adults has been evolutionarily conserved from Drosophila to vertebrates. In Drosophila, nanos function is required at two points during germline development. First, maternally provided nanos is required for survival of PGCs during early embryogenesis (Kobayashi et al., 1996; Forbes and Lehmann, 1998). While previous work using antisense morpholinos demonstrated that nos1 is similarly required for normal PGC migration and survival in zebrafish (Köprunner et al, 2001), our genetic demonstration that this PGC phenotype results in complete infertility in fish lacking maternal nos1 confirms that the maternal function for nanos is clearly conserved in fish. In contrast to zebrafish and Drosophila, zygotically expressed nanos orthologs have been shown to be required for survival of migrating PGCs in mouse embryos (Tsuda et al., 2003). Second, nanos is expressed in the adult ovary were it is required to maintain a self-renewing population of germline stem cells by preventing their precocious differentiation into oocytes (Gilboa and Lehmann, 2004; Wang and Lin, 2004). Because nanos is not required for oocyte maturation, nanos mutant mothers produce a limited number of functional oocytes before the onset of sterility due to stem cell depletion. In the mouse, it has so far not been possible to address the function of nanos genes in adult germline maintenance because nanos function is required zygotically for PGC survival in the embryo (Tsuda et al., 2003). In zebrafish, however, the early PGC survival function of nanos is provided by maternal nos1, as it is in Drosophila, thus allowing us to directly address the role of nos1 in adult germline maintenance.

We have reported here that nos1 is required for maintaining a continuous supply of mature eggs in adult zebrafish. In striking contrast to young adult wild-type females, whose ovaries contain all stages of maturing oocytes, the ovaries of young adult nos1 mutant females contain only late-stage oocytes. Based on our analysis of this phenotype, we conclude that nos1 functions at an early stage in oocyte production but is not required for oocyte maturation: nos1 mutant females produce apparently normal numbers of germ cells during larval development and these cells are capable of differentiation into oocytes which can further mature into fertile eggs. Any model to explain the role of nos1 in maintaining oocyte production must take into account two observations. First, the expression of nos1 appears to be initiated in follicle stage oocytes (stage Ib), and is not expressed in pre-follicle stage oocytes or in any <20μm germ cell. Second, nanos1 is not required for germ cells produced during larval development to differentiate into mature fertile eggs. Below we present three models for the role of nos1 in maintaining egg production in zebrafish.

Model I:

nos1 is required to maintain a pool of germline stem cells in the adult ovary that provide a continual supply of new oocytes, analogous to the function of nanos in the Drosophila ovary. As such, zebrafish would have an indeterminate number of eggs they could produce. In Drosophila, oocyte production is known to be a stem cell-driven process and nanos is a key factor required for maintaining the GSC fate, but not for oocyte maturation (Gilboa and Lehmann, 2004a; Wang and Lin, 2004). Nanos is clearly expressed in GSCs and is thought to negatively regulate genes that drive oocyte differentiation (Gilboa and Lehmann, 2004a; Wang and Lin, 2004). Because nanos is not required for oocyte maturation, nanos mutant mothers produce a limited number of functional oocytes before the onset of sterility. If germline stem cells are present in the zebrafish ovary, they most likely reside among the <20μm population of germ cells- a population of cells that is lost in nos1 mutant ovaries. Since expression of nos1 has not been detected in these cells, it must be postulated that nos1 functions via a cell non-autonomous mechanism to maintain GSCs. Alternatively, it is possible that nos1 is expressed in GSCs, but their numbers are small, making them difficult to detect. In this regard, it should be noted that in Drosophila, Nanos is expressed dynamically such that its levels are intermediate in GSCs, low during the mitotic divisions of the cystoblast, and then high in the 16-cell cyst that forms the ovariole (Forbes and Lehmann, 1998;Gilboa and Lehmann, 2004a; Wang and Lin, 2004). If this is also true in zebrafish, then the <20μm population of germ cells that do not express detectable nos1 may be a transiently mitotic population of cells, similar to the Drosophila cystoblast, that lies developmentally between the GSC and the early differentiating oocyte. Transient amplification of progenitor cells is a common theme among many, if not all, stem cell populations and can explain how a small number of stem cells can produce a large number of differentiated progeny (Spradling et al., 2001).

Model II:

nos1 is required to maintain the germline properties of early stage oocytes, analogous to the function of nanos in maintaining PGC survival during embryogenesis in Drosophila (Kobayashi, 1996; Forbes and Lehmann, 1998; Deshpande et al., 1999), zebrafish (Köprunner et al., 2001) and mouse (Tsuda et al., 2004). In this model, female germline development in zebrafish is more similar to that of mammals (Franchi et al., 1962), in that all oocytes are produced during early development (perhaps during embryonic and larval stages in zebrafish) and arrest in diplotene of the first meiotic prophase until they are activated to mature at a later date. As such, zebrafish would have a determinate number of eggs they could produce. In the absence of nos1, these arrested oocytes do not maintain their germ cell properties and are lost, perhaps by apoptosis as occurs to zebrafish PGCs lacking maternal nos1 (Köprunner et al., 2001). In this model it is likely that nos1 functions cell-autonomously in maintaining early diplotene-stage oocytes as nos1 is expressed at high levels in these cells.

Model III:

nos1 is required to maintain the germline properties of a population of pre-meiotic germ cells, analogous to the function of nanos in maintaining PGC survival. In this model, no new germ cells are produced in the adult ovary and new oocytes are periodically (or continuously) recruited from a preexisting pool of quiescent, pre-meiotic germ cells. In the absence of nos1, these cells do not maintain their germ cell properties and are lost, perhaps by apoptosis as occurs to zebrafish PGCs lacking maternal nos1 (Köprunner et al., 2001). This pre-existing pool is likely to be the <20μm germ cells, so as with the GSC model discussed above (Model I), the absence of detectable nos1 in the <20μm germ cell population necessitates that we propose nos1 to function non-cell autonomously. Because “nests” of synchronously developing pre-follicle stage oocytes are observed in the ovaries of many fish species, including zebrafish (Wallace and Selman, 1990; Selman et al., 1993), it is possible that the quiescent pre-meiotic germ cell in this model have the capacity upon activation to undergo a limited number of mitotic divisions (i.e. transiently amplify) to produce a clone of cells that will all enter meiosis synchronously.

Because the early-onset sterility of Znos1 female zebrafish is strikingly similar to the phenotype of Drosophila nanos mutants, we favor the GSC model (model I). It is clear that mitotic germ cells can been detected in the adult ovaries of other teleost species (e.g. Chaves-Pozo et al., 2005; Braekevelt and McMillan, 1967) and germline stem cells are widely believed to underlie the tremendous fecundity of many species of non-mammalian vertebrates, including teleost, amphibians and reptiles (Tokarz, 1978). Additionally, recent results in mice suggest that GSCs may also be present in ovaries of some mammals (Johnson et al., 2004; 2005). However, to our knowledge there is no definitive evidence (e.g. from transplantation experiments) that proves beyond a doubt the existence of germline stem cells in the ovaries of any vertebrate, including zebrafish. Nevertheless, we are confident that model II can be ruled out as it is clear that adult zebrafish ovaries contain pre-meiotic germ cells, as all stages from pre-meiotic interphase to the diplotene stage of the meiosis I can be observed (Selman et al., 1993; BWD unpublished; N. Kochakpour and P. B. Moens, pers. com.), and genes expressed at the earliest stages of meiosis, such as dmc1 and spo11, can readily be detected (H. Kassabian and BWD, unpublished). It will be possible to distinguish between models I and III by performing cell transplantation experiments with genetically marked adult germ cells (e.g. expressing GFP; Krøvel and Olsen, 1997). For example, a transplanted stem cell (model I) will give rise to many oocytes over a long period of time, whereas a transplanted quiescent pre-meiotic germ cell (model III) will produce only one or a few oocytes.

Regardless of the mechanism by which oocytes are produced in zebrafish, our finding that the >20μm germ cells localize to a discrete zone on the surface of the adult ovary suggest that this region may have functional significance in regulating oocyte production. For example, this zone could serve as a niche that regulates the behavior of pre-meiotic oocytes, similar to the germline stem cell niche in Drosophila (reviewed in Spradling et al., 2001). In addition, it is likely that this organization is evolutionarily conserved in teleost as a similar arrangement has been observed in the ovaries of fish in the distantly related Syngnathida family, which include pipefishes and seahorses (Begovac and Wallace, 1987; Wallace et al., 1991).

Our finding that nos1 plays a role in the maintenance of oocyte production in zebrafish sets the stage for identifying the genetic pathway that regulates oocytes production in vertebrates. The genetic accessibility and reproductive fecundity of the zebrafish, in which animals can generate hundreds of eggs every week, make this an ideal model system in which to study both conserved and vertebrate-specific mechanisms of egg production.