In this study, primordial germ cells (PGCs) in zebrafish were described histologically using eosinophilic granules as a marker. PGC-like cells (PL-cells) with eosinophilic granules were identified initially at the sphere stage (4-hr postfertilization), and were observed until the bud stage, the earliest stage to which PGCs with proper morphology could be traced. The morphology and distribution of eosinophilic granules in PL-cells changed during epiboly. Mitoses of PL-cells were observed only from the shield to bud stage, after eosinophilic granules aggregated to the perinuclear region in PGCs. These shifts of eosinophilic granules corresponded histologically to those of germ plasm described in Xenopus. These results suggest that eosinophilic granules represent germ plasm in fish and that PL-cells with these granules correspond to the PGCs or presumptive PGCs (pPGCs).
The bodies of multicellular organisms consist of various kinds of cells which are broadly classified into somatic and germ lines. In many animals, the germ line can be identified as cells which are morphologically different from somatic cells in the early developmental period. Generally, the germ line is formed through the mechanisms of predetermination or epi-genesis (Nieuwkoop and Sutasurya, 1979; Timmermans, 1996; Wakahara, 1996). In worms, flies and frogs, a special cytoplasm called germ plasm or pole plasm (in flies) is located asymmetrically within eggs and is inherited into some cells, which differentiate into germ cells (Nieuwkoop and Sutasurya, 1979). The cytoplasm is believed to predetermine the fate of germ cell differentiation, because transplantation and UV irradiation experiments have led to perturbation of germ-cell formation in Drosophila (Okada et al., 1974; Illemensee and Mahowald, 1974; Kobayashi and Okada, 1989) and in anurans (Smith, 1966; Ikenishi et al., 1986). The cytoplasm contains specific RNA and protein, such as vasa (Hay et al., 1988; Bardsley et al., 1993; Komiya et al., 1994; Ikenishi et al., 1996; Ikenishi and Tanaka, 1997). In Drosophila, females homozygous for a mutation of the vasa gene give rise to progeny that are sterile due to their lack of pole cells, the future germ cells (Williamson and Lehmann, 1996).
On the other hand, primordial germ cells (PGCs), which differentiate to germ cells, may arise epigenetically in urodeles (Kotani, 1957, 1958; Nieuwkoop and Sutasurya, 1979; Michael, 1984), mammals (Lawson and Hage, 1994), and birds (Ginsburg, 1994). In these animals, germ line cells can be identified only after the expression of morphological features of germ cells. In teleost, PGCs have been investigated exclusively by both light and electron microscopy (Johnston, 1951; Gamo, 1961; Hamaguchi, 1982; Timmermans and Taverne, 1989; Gevers et al., 1992b; Kazama-Wakabayashi et al., 1999; Braat et al., 1999). In these studies, PGCs have been identified by their morphological characteristics, namely roundish shape, relatively large size, large nucleus, and a clear nuclear membrane. Eosinophilic granules have been reported as histological characteristics common to PGCs in some species, such as medaka (Gamo, 1961), rosy barb (Timmermans and Taverne, 1989), and goldfish (Kazama-Wakabayashi et al., 1999). However, fish PGCs have not been traced back by these histological markers to before the gastrula period.
In zebrafish, germ line lineage can be traced throughout embryogenesis by means of vasa transcripts. Vasa mRNA distinguishes germ cell precursors from other somatic cells in cleaving embryos, and is inherited by the future germ cells (Yoon et al., 1997; Braat et al., 1999). Electron microscopic observation has also revealed germ-plasm-like structures containing vasa mRNA (Knaut et al., 2000). These results suggest that the formation of PGCs is predetermined by maternal factors. Like as the PGCs of other teleost species, however, zebrafish PGCs are not distinguished from the somatic cells before the gastrula stage under light microscope (Braat et al., 1999). In the present study, to describe the histological characteristics of PGCs, we made detailed observations of PGCs in zebrafish, especially before the gastrula period.
MATERIALS AND METHODS
Zebrafish, Danio rerio, were kept in the Aquarium Center of the Faculty of Fisheries, Hokkaido University. Parent fish were maintained at 29.0°C. Embryos were obtained through natural matings, and dechorionated with 0.1% trypsin (DIFCO) and 0.4% urea in Ringer's culture solution containing actinase E (KAKEN). Dechorionated embryos were rinsed with Ringer's culture solution (128 mM NaCl, 2.8 mM KCl, 1.8 mM CaCl2) containing 0.01% penicillin, 0.01% streptomycin and 1.6% albumen. Dechorionated embryos were cultured at 29±1°C in separate wells (Greiner) filled with Ringer's culture solution for 24 hr, and afterward in another separate well filled with culture solution (1.8 mM CaCl2, 1.8 mM MgCl2) containing the same antibiotics as above. The stage of development was identified according to Kimmel et al. (1995).
For histological observation of PGCs, embryos were fixed at appropriate stages from the 2-cell stage to 10-days postfertilization (dpf) with Bouin's fixative for 2 hr and transferred to 80% ethanol. Fixed samples were dehydrated using a butyl alcohol series, and subsequently embedded in a paraffin block. Serial transverse sections of 8 μm thickness were stained with hematoxylin and eosin according to the standard procedures.
Histological changes of zebrafish PGCs
We observed the morphology of PGCs in zebrafish embryos after fertilization (Figs. 1, 2, 3). PGCs were found at 6 dpf around the upper part of the body cavity, the future gonadal region (Fig. 1F). These cells were confirmed as PGCs based on characteristics similar to those observed in other teleosts: i.e., roundish shape, relatively large size, large nucleus, and clear nuclear membrane. Their sizes varied between 10 and 20 μm in major diameter, and the PGCs were apparently larger than other surrounding cells. The PGC cytoplasm, nuclear membrane and granular structures in the nucleus were all stained faintly with eosin.
PGCs were detectable at 10-hr postfertilization (10 hpf; bud stage), based on the histological features described at 6 dpf (Fig. 1A). Almost all PGCs were located among hypoblast cells around the para-axial part of an embryonic body. Compared with surrounding cells, PGCs contained large eosinophilic granules adjacent to their nuclear membranes. It was difficult to distinguish PGCs from other somatic cells before the bud stage using morphological features, especially their size and shape.
When PGCs were investigated at earlier stages, cells that exhibited eosinophilic granules, but not other characteristics of PGCs, were noticeable at 4 hpf (sphere stage) (Fig. 2A). We named these cells PGC-like cells (PL-cells). The eosinophilic granules in PL-cells, as well as those in PGCs, differed from yolk in their ability to be stained with eosin. From the sphere stage to the beginning of gastrulation, eosinophilic granules in PL-cells were scattered in the cytoplasm rather than around the nuclear membrane (Fig. 2A, B). From gastrulation to early segmentation, these scattered granules came to attach to a discrete section or the entire surface of the nuclear membrane (Fig. 1A, B, C, Fig. 2C, D), forming a capor ring-like structure similar to those described in Xenopus (Whitington and Dixon, 1975). The nuclei of PL-cells differed from those of other somatic cells in their ability to be stained with hematoxylin; the former was stained more faintly than the latter. From the late segmentation period to 6 dpf, the morphology of eosinophilic granules changed to disperse as small particles attached to the nuclear membrane, and was not conspicuous in PGCs entering the gonadal primordium (Fig. 1D, E, F). In this period, PGC nuclei exhibited fainter and more homogeneous eosin staining than those at previous stages. Cell divisions of PL-cells and PGCs were observed from the shield stage to the 2-somite stage period (Fig. 3). Eosinophilic granules were distributed almost equally to daughter cells as a few large granules, which were located between two chromosome clusters in the dividing cells (Fig. 3 A2, B2). The number of PL-cells increased nearly two or three times from the 30% epiboly stage to the bud stage period (Table 1).
Number of PL-cells and PGCs in embryos during normal development
The sizes of PL-cells and PGCs changed during development. PL-cells were about 30 μm in major diameter at the sphere stage. They became smaller, but were still larger than surrounding cells after the epiboly movement. The size of PGCs did not change following gastrulation, and was constant until at least 6 dpf.
The distribution of PL-cells and PGCs
The distribution of PL-cells and PGCs shifted with the advance of development (Fig. 4). In embryos at the sphere stage, PL-cells were found mainly in the lower part of the blastoderm, but did not cluster obviously in this region. From the beginning of epiboly to the shield stage, PL-cells were located in the deep cell layer close to the yolk syncytial layer (YSL). From the shield stage to the beginning of somitogenesis, all PGCs were observed in the hypoblast, and were often located side by side (Fig. 2C). The PGCs at the 1-to 6-somite stage (10.3-12 hpf) were located between the lateral mesoderm and the YSL, and were gradually approaching both sides of the embryonic body at the same distance from the notochord (Fig. 1B). At the 8-to 14-somite stage (13–16 hpf), when the somites were elevated from the yolk with morphogenesis of the neural primordium, PGCs were still located on the YSL. At the 21-somite stage (19.5 hpf), the location of PGCs shifted medi-ally, near the extended pronephric duct (Fig. 1D). At the hatching stage (48 hpf), they shifted toward a more medial position around the gut (Fig. 1E). Thereafter, up to 6 dpf, they were located on the peritoneum, dorsal to the intestine during the expansion of the body cavity (Fig. 1F).
The distribution of PGCs also shifted along the antero-posterior axis with embryonic development after epiboly (Fig. 4). PGCs were located in various positions in embryos throughout the epiboly movement, gradually moved to the trunk region, and formed clusters at the level of the third to sixth somite along the embryonic body at 12 hpf (the 6-somite stage). At 15 hpf (the 12-somite stage), these clusters were located at the level of the first somite to the third somite. Subsequently, PGC clusters scattered to the posterior along the embryonic body.
PGCs could be recognized in zebrafish embryos from 10 hpf onward under light microscope, being morphologically similar to those described in other teleost species, such as blackbass (Johnston, 1951), medaka (Gamo, 1961; Hamaguchi, 1982), rosy barb (Timmermans and Taverne, 1989; Gevers et al., 1992b) and goldfish (Kazama-Wakabayashi et al., 1999). Fish PGCs have not been traced back histologically to before the gastrula stage. In the present study in zebrafish, some blastomeres, referred to as PL-cells, contained eosinophilic granules and were identified after the sphere stage (4 hpf; late-blastula). The morphological identification of PGCs at this stage has not been reported in any teleost species.
Recent studies on vasa mRNA, a germ line-specific marker, indicated that germ cell precursors were segregated early from somatic cell lineage and that the germ cell lineage could be traced throughout embryogenesis in zebrafish (Yoon et al., 1997; Knaut et al., 2000). Maternal vasa mRNA segregation changes from asymmetric to symmetric at the sphere stage (4 hpf), and zygotic vasa transcription is initiated shortly after the loss of unequal vasa mRNA segregation (Knaut et al., 2000). Thereafter, proliferation of PGCs and perinuclear localization of Vasa protein are observed. The PGCs described by Knaut et al., (2000) had a location and proliferation corresponding to those of the presently observed PL-cells containing eosinophilic granules. This suggests that these PL-cells corresponded to PGCs or presumptive PGCs (pPGCs).
We observed granules that stained deeply with eosin in zebrafish PGCs. In medaka, rosy barb and goldfish, a few eosinophilic granules have been observed in the cytoplasm adjacent to nuclei of PGCs (Gamo, 1961; Timmermans and Taverne, 1989; Kazama-Wakabayashi et al., 1999). Especially in rosy barb, perinuclear dense bodies in PGCs corresponded with “nuage”, germ line-specific materials that were detected ultrastructually, and were clearly visible after staining with eosin and gallocyanin, indicating the presence of protein and RNA (Timmermans and Taverne, 1989; Gevers et al., 1992b). Therefore, eosinophilic granules should be one of the characteristics used to identify PGCs in teleosts.
Eosinophilic granules changed their morphology and location in PL-cells and PGCs during development, which was roughly classified into three phases (Fig. 5) as follows: (1) large and small patches were distributed in the cytoplasm of PL-cells at the late-blastula stage; (2) smaller particles formed a cap or ring in the cytoplasm or perinuclear region of PGCs from the gastrula stage to the early segmentation period; and (3) small particles were observed in a juxtanuclear and inner part of the nuclei position in PGCs from the late segmentation period to the hatching period. These morphological profiles of eosinophilic granules were similar to the confocal profiles of Vasa protein that were determined by Knaut et al., (2000) using whole-mount staining. In particular, Vasa protein was detected at the sphere stage when eosinophilic granules were identified for the first time. The perinuclear distribution of eosinophilic granules corresponded to that of Vasa protein (Knaut et al., 2000). In Xenopus, germ plasm also shows similar morphological profiles: moving to a perinuclear region, forming a cap or ring in the late blastula or early gastrula, and inheritance by both daughter cells in 2–3 mitotic divisions (Whitington and Dixon, 1975; Dziadek and Dixon, 1975, 1977). It is tempting to speculate that eosinophilic granules might be fish germ plasm and contain Vasa protein. However, Braat et al. (2000) reported that Vasa protein localizes in many blastomeres at the early cleavage stage. In the future, it will be necessary to analyze the histochemical characters of eosinophilic granules.
During gastrulation, three kinds of alterations occurred in PL-cells; namely, the shift of eosinophilic granules to a peri-nuclear location, affinity for hemetoxylin in the nucleus, and clonal divisions, in which eosinophilic granules were inherited by each daughter cell. Knaut et al., (2000) reported that zygotic vasa RNA transcription is initiated after the sphere stage and that proliferation of PGCs occurs during gastrulation. In rosy barb, proliferation of pPGCs was observed at a similar stage (Timmermans and Taverne, 1989). These results suggest that segregation of PGCs in zebrafish occurs during gastrulation and that the shift of eosinophilic granules plays an important role in differentiation from pPGCs to PGCs. In Xenopus, Whitington and Dixon, (1975) suggested that the shift of germ plasm to the perinuclear region triggered both an equal distribution of germ plasm at mitosis and an increase in the number of pPGCs. Dixon (1994) proposed that germ plasm play a role in protecting germ cells during the processes specializing the adult body form. In zebrafish, specific cell lineages arise at the gastrula stage (Kimmel and Warga, 1986; Kimmel et al., 1990; Warga and Kimmel, 1990). In our histo-logical observations, the hematoxylin-stainability of the nuclei of PL-cells and other somatic cells was equivalent at the sphere stage, but the nuclei of PL-cells exhibited fainter staining during gastrulation. This would indicated the PGCs would be segregated from somatic cells during gastrulation. According to Dixon's proposal, the eosinophilic granules in zebrafish might play a role in protecting the totipotency or pluripotency of PGCs. In Xenopus, on the other hand, without a proper environment, cells containing germ plasm differentiate to somatic cells even if they are germ line cells (Wylie et al., 1985; Ikenishi and Tsuzaki, 1988). However, the role or function of eosinophilic granules in zebrafish remains to be determined.
The present results suggest that the mode of germ cell formation in zebrafish is close to those in Drosophila and Xenopus; that is, it occurs predeterminally. Gevers et al. (1992a) used cell lineage analysis to demonstrate that PGCs in rosy barb originate from lower tiers of blastomeres at the 64-cell stage. In goldfish, microsurgical analysis showed that PGC-bearing cells were predetermined by cytoplasmic factors and already distributed in the lower part of the blastoderm at the mid-blastula stage, after which they could not be regenerated (Kazama-Wakabayashi et al., 1999). These results would also agree with the hypothesized mechanism in zebrafish, and would imply that fish species might share a common mechanism of germ cell formation. However, conflicting data have been reported in medaka: vasa-positive cells could not be traced to before the late-gastrula stage (Shinomiya et al., 2000). Therefore, it is not clear whether the germ line is predeterminant. With respect to germ plasm, the question might be resolved by means of transplantation techniques and so forth.
In conclusion, zebrafish PGCs were light-microscopically identified by the presence of eosinophilic granules as in other teleost species, and then were traced back to the sphere stage. It seemed that blastomeres containing eosinophilic granules could be set aside from specification of somatic cell lineage, because of their faint staining with hematoxylin. The germ plasm of Drosophila must contain a combination of factors controlling both the cellular behavior of the early germ cells as well as the activation of germ cell-specific gene expression. Eosinophilic granules are one of the candidates for controlling germ cell specification in fish. Detailed biochemical analysis of the granules in relation to vasa RNA and Vasa protein is needed to confirm our conclusions.
We wish to thank Emeritus Professor Dr. Fumio Yamazaki, Faculty of Fisheries, Hokkaido Univ., and Dr. Akira Goto, Graduate School of Fisheries Science, Hokkaido Univ., for their kind guidance in this study. We would like to thank Dr. Norio Nakatsuji, Institute for Frontier Medical Sciences, Kyoto Univ., and Dr. Takako Nakatsuji, Tokai Univ. School of Marine Science and Technology, for their critical reading of the manuscript and valuable input. We wish to thank Dr. Yuji Yamazaki, Faculty of Science, Toyama Univ., for his kind advice and collection of papers. Thanks are also due to the members of the Laboratory of Embryology and Genetics, Faculty of Fisheries, Hokkaido University. This study was supported in part by a Grant-in-Aid from the Ministry of Education, Science and Culture (07660230) to E.Y.