Transgene driving GFP expression from the promoter of the zona pellucida gene zpc is expressed in oocytes and provides an early marker for gonad differentiation in zebrafish



Although mechanisms of sex differentiation have been studied intensely in mammals, insects, and worms, little is known about this process in lower vertebrates. To establish a marker for female gonad differentiation in zebrafish, we generated a transgenic line in which 412 bp from the promoter and 5′ mRNA leader of the female-specific zebrafish zona pellucida gene zpc are fused to the coding region of green fluorescent protein (GFP). The zpc0.5:GFP transgene is expressed exclusively in oocytes, starting from the onset of female-specific differentiation, and closely resembles the expression pattern of the wild-type zpc. Strong GFP expression persists throughout oogenesis and is visible through the body wall of females. We have also characterized a putative upstream factor of zpc, FIGalpha, and show that distribution of FIGalpha RNA is compatible with its postulated role in the regulation of zpc. The zpc0.5:GFP transgenic line described here will be useful for studying oocyte development and the mechanisms that determine sex-specific gene expression in the zebrafish. It is also the first promoter characterized to date to drive stable and efficient expression specifically in the zebrafish female germline. Development Dynamics, 2003. © 2003 Wiley-Liss, Inc.


Growing vertebrate oocytes are surrounded by an extracellular matrix membrane called the zona pellucida, which is required for follicle formation, fertilization, and early development. The mouse zona pellucida contains the three glycoproteins ZP1, ZP2, and ZP3 (also named ZPA, ZPB, and ZPC) that share a common zona pellucida protein domain (Bork and Sander, 1992). In mammals, ZP3 is involved in sperm binding to the egg. ZP3 serves as a sperm receptor, induces the acrosomal reaction, and constitutes a major barrier to interspecies fertilization (reviewed by Wassarman, 1999). The vitelline envelope surrounding eggs from fish, birds, and amphibians contains glycoproteins homologous to mammalian ZP proteins. These glycoproteins play an essential role in assembling the extracellular structural coats during oogenesis. They serve as a selectively permeable barrier during early stages of oocyte development and as a protective layer for the developing embryo. Furthermore, they participate in the sperm–egg interaction and contribute to the block of polyspermy. In most vertebrates analyzed, including mammals, Xenopus, and several fish species, genes for ZP glycoproteins are exclusively transcribed in growing oocytes (Wassarman, 1990; Kubo et al., 1997). However, some exceptions exist. For example, in chick ZP3 homologue is synthesized by follicle cells surrounding the embryo (Waclawek et al., 1998), and the ZP1 homologue is transcribed in the liver (Bausek et al., 2000). In several teleost fish species, ZP components are transcribed in the liver in response to estrogen regulation (Del Giacco et al., 1998; Sugiyama et al., 1998). In medaka, two major proteins of the egg envelope that belong to zpC (choriogenin L) and zpB (choriogenin H) classes are synthesized by the liver (Murata et al., 1997), whereas seven additional recently characterized members of the ZP gene family seem to be expressed in an ovary-specific manner (Kanamori, 2000; Kanamori et al., 2003).

In zebrafish, ZP family genes are abundantly transcribed during oogenesis: 10.3% of total transcripts expressed in the ovary code for proteins of ZPA-C families (Zeng and Gong, 2002). The ZP2 (ZPB) and ZP3 (ZPC) type genes each constitute a multigene family in the genome (Wang and Gong, 1999; Mold et al., 2001), with ovarian expression detected for at least three closely related genes of each family. In addition, expression of zebrafish ZPC, another member of ZPC family, more distantly related to mammalian ZP3 and more closely related to medaka ZPC4 and ZPC5, was found in zebrafish (Del Giacco et al., 2000). All characterized members of the zebrafish ZP family are expressed only during oogenesis (Wang and Gong, 1999; Del Giacco et al., 2000; Mold et al., 2001; Zeng and Gong, 2002).

Expression of zona pellucida ZP protein homologues in different organisms may provide a paradigm for studying mechanisms and evolution of oocyte-specific gene expression. In mouse, the expression of all three ZP genes is precisely regulated during oogenesis and restricted to a 2-week growth phase, when they represent together approximately 1.5% of the total polyA+ RNA (Epifano et al., 1995). This regulation is achieved by binding of a heterodimer consisting of the ubiquitous basic helix-loop-helix (bHLH) transcription factor E12 and the germ cell–specific bHLH factor FIG-alpha (Epifano et al., 1995; Liang et al., 1997; Soyal et al., 2000) to canonical E-box sequence (CANNTG) located approximately 200 bases upstream of the transcription start sites of ZP1, ZP2, and ZP3 (Millar et al., 1991).

Expression of oocyte-specific ZP genes in medaka is precisely synchronized, starting at early stages of oocyte development, and decreasing before onset of vitellogenesis. E-box sequences are present in relative abundance immediately upstream of oocyte-specific medaka ZP genes as well as of their fugu homologues. The expression profile of the medaka FIGalpha homologue suggests that the mechanisms controlling oocyte-specific expression may be conserved between mammals and fish (Millar et al., 1991; Kanamori, 2000; Kanamori et al., 2003). The presence of E-boxes in the putative promoters of two zebrafish ZPB homologues (Mold et al., 2001) and the identical patterns of expression of several ZP proteins in zebrafish oocytes (Del Giacco et al., 2000; Zeng and Gong, 2002) further corroborate this idea.

In the present work, we have isolated the regulatory region of the zebrafish zona pellucida gene zpc (Del Giacco et al., 2000; Zeng and Gong, 2002) and constructed a zpc:GFP transgenic fish strain in which 366 bp of the promoter, containing 3 E-box sequences, followed by 46 bp of 5′ mRNA leader are fused to the coding region of green fluorescent protein (GFP). The zpc0.5:GFP transgene is expressed exclusively in oocytes and closely resembles the expression pattern of the endogenous zebrafish zpc gene. We examined the temporal and spatial distribution of the zebrafish transcripts encoding the FIGalpha homologue. Our data suggest that FIGalpha may be one of the factors that coordinate oocyte-specific expression of zebrafish ZP genes. The onset of zpc0.5:GFP transgene expression coincides with differentiation of the bipotential zebrafish gonad into ovary or testis, providing a useful tool to study sex differentiation mechanisms in zebrafish.


Characterization of the zpc Control Region

To characterize the regulatory regions of the zebrafish zpc gene, we isolated a genomic clone and sequenced a 6-kb region upstream of the zpc start codon (submitted to GenBank accession no. AY302566). Two promoter prediction programs revealed a putative core promoter with a TATA box 30 bp upstream of the transcription initiation site (Del Giacco et al., 2000). The whole 6-kb upstream region of zpc contains 36 randomly distributed E-box sequences; seven of these E-box sequences most proximal to transcription start are at the positions −693, −637, −582, −550, −321, −135, and −69 (numbered relative to the transcription start site; Fig. 1A) and several regions homologous to repetitive sequences. At 4.8 kb from the zpc transcription start site, the last exon of another gene is located (expressed sequence tag [EST] fk03b06.x1; transcribed in the same direction as zpc, no homology to zp genes; Fig. 1A). To obtain zpGFP transgene constructs, either 412 bp (zp0.5:GFP) or 6 kb (zp6:GFP) of sequence immediately upstream of the zpc start codon were fused in frame with the mmGFP5 reporter gene (Zernicka-Goetz et al., 1996). The zp0.5:GFP construct contains E-box sequences in the positions (−69, −135, and −321; Fig. 1B). The zp6:GFP construct contains 6 kb upstream of the zpc transcription start (Fig. 1A).

Figure 1.

Structure of the zebrafish zpc upstream genomic region in constructs zpc6:GFP and zp0.5:GFP. mmGFP (Green box; Zernicka-Goetz et al., 1996) fused in-frame with the first ATG of zpc; numbers below the line indicate nucleotide positions in kilobases (kb, A) or base pairs (bp, B) relative to the transcription start site. A: Schematic representation of the 6-kb upstream region in zpc6:GFP: the black block arrow close to −4.8 kb indicates the 3′-exon of another gene (expressed sequence tag no. fk03b06.x1), arrows below the line show positions of repetitive sequences, and vertical bars with “E” indicate seven E-box sequences within 693 bp upstream of the zpc transcription start site (positions −693; −637; −582, −550; −321, −135, and −69 relative to the transcription start site; additional E-boxes further upstream and separated by repetitive DNA from the promoter are not indicated). B: Closer view of the 412-bp upstream region with the zpc promoter: the black boxes indicate three E-box sequences, TAAAA at the −30 position likely corresponds to the TATA-box, and +1 refers to the zpc transcription start site.

Establishment of Transgenic zp0.5:GFP Zebrafish

zpc0.5:GFP and zpc6:GFP constructs were microinjected into 300 and 800 one-cell stage zebrafish embryos, respectively. Approximately 90% of injected embryos survived to adulthood. We did not observe ectopic GFP expression during development of F0 fish. The injected embryos were raised to sexual maturity and F0 females (67 for zpc0.5:GFP and around 350 for zpc6:GFP) were crossed to wild-type males (AB/TL strain). Because the zpc promoter should be active in growing oocytes (Del Giacco et al., 2000), we expected to see maternally derived GFP protein in freshly laid eggs. We did not succeed in generating F2 transgenic fish for zpc6:GFP, although green eggs were found in the progeny of four zpc6:GFP-injected females (founders 1–4, see Experimental Procedures section). Among the progeny of one of the zpc0.5:GFP-injected G0 females, approximately 10% eggs contained high levels of GFP during the first day, and GFP was still detectable on the second day of development (Fig. 3). GFP-positive F1 embryos (n = 55) were raised to sexual maturity and genotyped; 52% of them (n = 29) carried the transgene in the genome. Of the F2 progeny (n = 79) obtained from a cross between a zpc0.5:GFP F1 female and a wild-type AB/TL strain male, 42% carried the transgene (n = 33). Because approximately 50% of both the F1 and the F2 progeny from a single G0 founder carry the transgene, we concluded that zpc0.5:GFP is most likely integrated at a single locus and transmitted in a Mendelian manner. We assigned the allele designation TG(zpc0.5:GFP)m1032 to this transgenic strain according to zebrafish nomenclature (

Figure 3.

Oocyte-derived green fluorescent protein (GFP) is localized to the cytoplasm during embryogenesis. A: Newly fertilized egg from zpc0.5:GFP female containing cytoplasm with GFP and interspersed yolk platelets. Cytoplasmic streaming and segregation of yolk are obvious at 5 min (B), 10 min (C), 15 min (D), and 30 min (E) postfertilization. From the two-cell stage (F) onward, GFP is evenly distributed throughout the blastodisk. The 4-cell stage (G), 8-cell stage (H), 256-cell stage (I), oblong stage (J), shield stage (K). GFP fluorescence is clearly visible at 24 hr postfertilization (L) but fades thereafter. Orientation of embryos: A–K animal pole at top; L anterior to the left, dorsal up. Scale bar = 200 μm in A (applies to A–L).

Expression Profile of the zpc0.5:GFP Transgene Is Similar to That of Endogenous zpc

Similar to endogenous zpc mRNA (Del Giacco et al., 1998; this work), zpc0.5:GFP transgene-derived GFP is most strongly expressed in stage I oocytes (Fig. 2C,D). In juvenile zpc0.5:GFP transgenic females, the earliest GFP expression in gonads is detectable through the body wall starting at 27–28 days postfertilization (dpf), and GFP expression is strongly elevated within the next several days (Fig. 2A). Oogenesis in zebrafish proceeds through five stages of maturation, identifiable on the basis of oocyte size and certain morphologic features (such as location of groups of oocytes within the “nest” [stage IA] or an individual follicle [stage IB]) and accumulation of cortical alveoli at stage II and of yolk granules at stage III (Selman et al., 1993). In an ovary isolated from 28 dpf zpc0.5:GFP fish, GFP expression is seen in some oocytes, corresponding by size to stage IB (20–140 μm), which represents the follicle phase of primary growth (Fig. 2B). At this stage, the oocyte becomes surrounded by a single layer of follicle cells. Shortly after follicle formation, the oocyte enters the diplotene stage of the first meiotic prophase, where they arrest for the remainder of oocyte development. The nucleus, or germinal vesicle, is easily seen by examination with epifluorescence; the GFP fluorescence is uniformly distributed in cytoplasm surrounding the nucleus. As the ovary grows, the majority of the zpc0.5:GFP oocytes start to express GFP. In the ovary of adult zpc0.5:GFP females, the most intense GFP fluorescence is detectable in stage IB oocytes (Fig. 2C, arrows). However, some stage IA oocytes also weakly express GFP (Fig. 2C, insert). Stage II oocytes (0.14–0.34 mm) are characterized by accumulation of cortical alveoli, occupying most of the cytoplasm by the end of this stage. At stage III, during which major growth occurs, oocytes accumulate vitellogenin and increase in size (0.34–0.69 mm). During stages II and III, most of the GFP protein fluorescence is detectable from a narrow cytoplasmic layer around the germinal vesicle (Fig. 2C,D). Interestingly, the area covered by this intense GFP staining corresponds by size to oocytes stage IB. During stage IV (oocyte maturation, 0.69–0.73 mm), the germinal vesicle migrates toward the future animal pole, breaks down, and the first meiotic division occurs. At this stage, a bright GFP spot surrounding the germinal vesicle becomes asymmetrically located and may indicate the progress of maturation (Fig. 2D, gv). After germinal vesicle breakdown, stage V mature eggs ovulate into the ovarian lumen. GFP epifluorescence at this stage is unevenly distributed (Fig. 2E), with a maximum intensity found at the animal pole (Fig. 2E, arrowhead) in the region of the future blastodisk.

Figure 2.

Expression of mmGFP in the ovaries of zpc0.5:GFP female fish. A: Green fluorescent protein (GFP) -expressing ovary (arrowhead) is seen through the body wall of a 36 days postfertilization (dpf) female fish. B: Isolated ovary of a 28 dpf female. C: Ovary of an adult female. Oocyte stages noted (Selman et al., 1993). Note the strongest GFP expression in stage IB oocytes (arrows). In stage II and III oocytes, the germinal vesicle position is visible (arrowheads). Insert: arrows point at stage IA oocytes at higher magnification. D: Oocytes at stages I–IV. Note asymmetric germinal vesicle (gv) position in stage IV oocyte (arrowhead). E: Stage V oocyte. GFP is distributed asymmetrically, most abundant in the animal pole (arrowheads). All pictures were taken with an epifluorescence microscope. Scale bars = 1 mm in A, 200 μm in B,D,E, 100 μm in C (1 mm in insert).

All of the eggs from transgenic zpc0.5:GFP females contain GFP and are bright green when visualized by using epifluorescence illumination. After fertilization, GFP fluorescence is strongest within the non-yolky cytoplasm of the eggs, and its distribution reflects the cytoplasmic streaming toward the animal pole during segregation from the yolk within the first half hour of development (Fig. 3A–E). From the two-cell stage on, GFP remains evenly distributed through the blastodisk, blastoderm, and later cellular components of the embryo during the first day of development (Fig. 3A–K) but gradually disappears during somitogenesis stages. GFP fluorescence is in general still relatively bright at 24 hours postfertilization (hpf, Fig. 3L) and is clearly detectable above background at 48 hpf (data not shown).

To test if the onset of zpc0.5:GFP expressionmaybe a reliable indicator of female gonadal differentiation, we determined whether all fish expressing GFP at early stages will become females. We sorted out GFP-expressing fish from nonexpressing ones either at 27 dpf (Experiment 1), or at 35 dpf (Experiment 2), raised them separately to sexual maturity, and counted the number of males and females among the groups of GFP-expressing and non–GFP-expressing fish (data are summarized in the Table 1). GFP expression is a reliable indicator of female development at late juvenile stages (35 dpf). Surprisingly, we occasionally observed younger juveniles at 27 dpf initially expressing zpc0.5:GFP and later differentiating into males (discussed later).

Table 1. Correlation of zpc0.5:GFP Expression at Different Juvenile Stages With Later Female Versus Male Adult Differentiationa
 MotherFatherJuvenile fish tested as GFP +Adult fish
GFP + developed as femalesGFP +/GFP− females
  • a

    Juvenile fish were tested at 27 dpf for Experiment 1, at 35 dpf for Experiment 2. In Experiment 1, 35 of 106 fish were sorted out at 27 dpf as GFP-positive and 30 of them survived to adulthood. Two of these 30 fish have lost GFP expression at 40 dpf and developed as males. All fish that developed as females and were GFP-positive at 27 or 35 days expressed GFP in the ovaries also at adult age. In Experiment 1, fish were sorted out before 29 dpf, when most of the oocytes disappear from the male gonad (Uchida et al., 2002), and GFP expression observed was most likely due to GFP activation in oocytes of juvenile gonads in these males. Fish isolated at 35 dpf based on GFP expression developed as females in 100% of the examined cases (Experiment 2). The ratio of GFP+/GFP− females observed in both experiments is close to that expected, given Mendelian segregation of the zpc0.5:GFP transgenes. GFP, green fluorescent protein; dpf, days postfertilization.

  • b

    Number of all juvenile fish.

  • c

    Number of fish sorted as GFP-positive at juvenile age and survived to adulthood.

Exp 1zp0.5GFP/−wt35 (33% of n = 106)b28 (93% of n = 30)c28/27
Exp 2zp0.5GFP/−zp0.5GFP/−23 (23% of n = 98)b23 (100% of n = 23)c23/11

Cloning of a Zebrafish FIGalpha Homologue

ZP promoter sequences critical for oocyte-specific expression in mice (E-box) are located approximately 200 bp upstream from the transcription start sites (Millar et al., 1991) and are regulated by the bHLH factor FIGalpha. A common feature of the putative promoters of medaka, fugu, and zebrafish genes with oocyte-specific expression is the presence of at least one E-box near the transcription start site (Mold et al., 2001; Kanamori et al., 2003). The zebrafish zpc promoter contains three E-boxes relatively close to the transcription start site (Fig. 1B), which raises the possibility that a zebrafish FIGalpha homolog may also be involved in the control of zpc expression. We cloned a zebrafish FIGalpha homolog by reverse transcriptase-polymerase chain reaction (RT-PCR) from ovary cDNA by using primers based on zebrafish EST sequences with highest similarity to medaka FIGalpha (accession no. BM156452 and BM156753, see Experimental Procedures section). The resulting PCR product contained a single 663-bp open reading frame 98% homologous to BM156452 at the nucleotide level (Fig. 4A; submitted to GenBank accession no. AY302567). These differences do not alter the deduced amino acid sequence and may represent single nucleotide polymorphisms between the different strains used for the EST project and our study, or it may be possible that there are several very similar copies of FIGalpha in the zebrafish genome.

Figure 4.

Zebrafish FIGalpha homolog. A: Clustal X multiple sequence alignment of deduced FIGalpha amino acid sequences from different vertebrates. Numbers in parentheses refer to GenBank protein accession numbers (medaka, human, and mouse), expressed sequence tags (Xenopus and rainbow trout), or genomic contigs (fugu and tetraodon). Black boxes indicate identity; gray boxes indicate similarities. The basic helix-loop-helix (bHLH) domain is highlighted: “basic” indicates DNA-binding region; helixI and helixII are indicated. Stars indicate conserved consensus residues of the bHLH domain (after Liang et al., 1997). Note high similarity of FIGalpha proteins outside conserved residues within the bHLH domain and in the N-terminus. Zf, zebrafish (Danio rerio), this work; medaka, Orysias latipies (AAD38902; Kanamori, 2000); human, Homo sapiens (XP_292886; Huntriss et al., 2002); mouse, Mus musculus (NP_036143; Liang et al., 1997); fugu, Takifugu rubripes (scaffold 352, Pufferfish genome assembly 2, join 50546-50725, 51262-51431, 52397-52728); tetraodon, Tetraodon nigroviridis (Tetraodon genome assembly 6, FS_CONTIG_763_1, join 14041-13862, 13238-13052, 12933-12882, 12508-12347); Xl, Xenopus laevis (AW646029, BI477650, AW643447, BI448320); Xt, Siluriana tropicalis (AL886932); rainbow trout, Oncorhynchus mykiss (BX074063, BX074064, BX078575). B: FIGalpha proteins are conserved in vertebrates. The unrelated mouse bHLH protein SCL (CAB72256), with just 37% identity to zebrafish FIGalpha, was taken as an outgroup.

Zebrafish FIGalpha protein is 45% similar to medaka FIGalpha and 42% similar to mouse FIGalpha. Analysis of publicly available genomic and EST data reveals genes closely related to FIGalpha in fugu, tetraodon, rainbow trout, and Xenopus (Fig. 4A, see legend for details). A multiple alignment of deduced FIGalpha amino acid sequences from zebrafish, medaka, mouse, human, and these organisms is shown in Figure 4A. FIGalpha homologues are more similar within the bHLH domain among each other than when compared with unrelated bHLH family proteins. There is an additional conserved region near the N-terminus with no previously identified conserved sequence motifs. Within the region between the bHLH domain and the C-terminus, however, no significant similarities are seen. FIGalpha genes form a distinct subgroup within the bHLH family of transcription factors (Fig. 4B), which may suggest a common function for these genes.

Zebrafish zpc and FIGalpha Are Expressed in the Ovary From the Onset of Sexual Differentiation

We performed whole-mount in situ hybridization on isolated ovaries with antisense RNA probes generated for FIGalpha and zpc to compare the expression patterns of these genes at different stages of oogenesis (Fig. 5A,B). The first manifestations of a developing zona pellucida appears by the end of stage I, when the oocytes are approximately 140 μm in diameter (Selman et al., 1993). zpc RNA is expressed identically to that of other previously reported zebrafish ZP genes (Del Giacco et al., 2000; Zeng and Gong, 2002). The strongest zpc expression is detected at stage I, but decreases during stage II and becomes nondetectable by stage III (Fig. 5B,D). We found that FIGalpha is also expressed in stage IA and IB oocytes (Fig. 5A) and is undetectable at time points later than stage II. In parallel, in situ hybridization for FIGalpha was performed with whole fixed testes, in which we did not detect any FIGalpha-specific signal above the low background obtained with a FIGalpha sense RNA probe (data not shown).

Figure 5.

Expression of FIGalpha and zpc during zebrafish oogenesis. In situ hybridization with antisense digoxigenin-labeled RNA probes for FIGalpha (A), zpc (B,D), or mmGFP (C) on whole ovaries of adult wild-type females (A,B) or cryosections of adult wild-type females (D) and adult zpc0.5:GFP transgenic female (C). Note the exclusive expression of FIGalpha (A), and the maximal expression of zpc (B,D) at the oocytes stage I (stages by Selman et al., 1993). Distribution of mmGFP transcripts in zpc0.5:GFP transgenic fish (C) is similar to that of zpc (D). Control hybridization of sense probes with ovaries was performed for all probes and gave no signal (data not shown). Scale bars = 100μm.

To determine the onset of zpc expression during development and to establish an expression profile for zpc and its putative regulator FIGalpha, we prepared total RNA from various developmental stages and measured the relative amount of zpc and FIGalpha messages by using semiquantitative RT-PCR. RNA was isolated from zebrafish adult ovaries and from pooled embryos and larvae from 0 to 29 dpf, which encompasses the developmental events of hatching (2.5 dpf), sex determination, and sexual differentiation. Initially, there is no detectable phenotypic difference between the two sexes in young zebrafish: all individuals develop undifferentiated ovary-like gonads. In male zebrafish, the oocytes disappear from the gonads by 32.5 dpf, when testicular differentiation takes place (Takahashi, 1977). Both zpc and FIGalpha are abundantly expressed in ovaries of the adult fish (Fig. 6A), the mRNA amounts decrease when the eggs are laid, and are absent from the developing fish during days 5–20 of development. Zygotic transcription of both genes starts at day 22 after fertilization and increases until day 29, coinciding with the progress in sex differentiation.

Figure 6.

Age-specific expression profile of FIGalpha and zpc during juvenile stages. A: Start of zpc and FIGalpha expression coincides with sexual differentiation. Total RNA was isolated from adult fish ovary, fertilized eggs, or fish of the indicated age, and analyzed by semiquantitative reverse transcriptase-polymerase chain reaction (RT-PCR) for expression of zpc and FIGalpha. The ubiquitously expressed gene EF1-alpha was used for normalization. -RT is control lane without reverse transcriptase. B:zpc and FIGalpha are expressed in a coordinated manner. Total RNA was isolated from eight single fish at 30 days postfertilization (dpf) and analyzed by RT-PCR for expression of zpc and FIGalpha. Strong expression of both zpc and FIGalpha most likely indicates developing females (fish 2, 4, 5, 7, 8).

To investigate whether expression of zpc and FIGalpha are correlated during gonadal differentiation, we isolated RNA from eight individual fish at 30 dpf and performed gene-specific RT-PCR. These fell into two distinct groups, expressing and nonexpressing, where individuals with coordinately elevated FIGalpha and zpc expression (fish 2,4,5,7,8 in Fig. 6B) are most likely corresponding to developing females.

In summary, we found that start of transcription of FIGalpha and zpc occurs simultaneously in zebrafish and both are expressed in a female-restricted manner in larvae older than 30 days: whether low levels of FIGalpha may already be expressed in females specifically at stages earlier than 30 dpf remains to be determined. As we noticed that some juvenile fish expressing zpc0.5:GFP before 30 dpf differentiated later into males, it is possible that onset of zpc expression in zebrafish may be connected to a certain growth stage of the oocytes, which is reached in gonads of all juveniles shortly before male-specific elimination of the oocytes takes place.


FIGalpha Is a Putative Regulator of the zpc Promoter

In mice, the regulation of the promoters of the ZP1, ZP2, and ZP3 genes is achieved by binding of a heterodimer of transcription factors E12 and FIGalpha to canonical E-box sequences (Epifano et al., 1995; Liang et al., 1997; Soyal et al., 2000). Females of FIGalpha knock-out mice are sterile, and oocytes lack ZP gene expression (Soyal et al., 2000), revealing a critical requirement of FIGalpha for expression of zona pellucida genes. Mouse FIGalpha together with E12 are sufficient to transactivate ZP1, ZP2, and ZP3 reporter constructs (but not endogenous genes) in mouse embryonic fibroblasts.

We have isolated a zebrafish zpc promoter fragment, which contains three putative binding sites for FIGalpha at the positions −69, −135, and −321 relative to transcription start site and demonstrate that this fragment is sufficient to drive reporter gene expression in a pattern identical to that of the endogenous zpc gene. Conservation of FIGalpha in several fish species, frogs, and mammals, together with the presence of putative binding sites for FIGalpha in the promoters of mouse, human, medaka, and zebrafish ZP genes, suggests a conserved function for FIGalpha in regulating coordinate oocyte-specific expression of ZP genes in vertebrates. We cloned and characterized the zebrafish homologue of FIGalpha, and found that FIGalpha and zpc are coexpressed during zebrafish development and in the adult ovary. Although this result does not prove a direct interaction between the two genes, it defines the place and period during which FIGalpha may act to initiate expression of zpc. However, FIGalpha alone was not sufficient to induce GFP protein or RNA expression, when injected in one-cell stage zebrafish zpc0.5:GFP transgenic embryos, or to elevate zpc RNA levels derived from the endogenous zpc gene (our unpublished observations). Like many other bHLH transcription factors, the tissue-specific protein FIGalpha in mouse forms a heterodimer with the near ubiquitous bHLH factor E12 to regulate target genes. One E12 homolog is known in zebrafish and expressed ubiquitously during early stages of development (Wulbeck et al., 1994). Alternatively, FIGalpha in zebrafish may require another binding partner, which is not expressed during embryonic stages.

zpc and FIGalpha Expression Coincides With Gonad Differentiation

zp genes and FIGalpha are among the few genes for which sex-specific expression in fish is documented. In systematic screen for genes that are expressed sex-specifically during early medaka embryogenesis (Kanamori, 2000), FIGalpha was found to be the earliest female-specific marker (1 day after hatching), while zp genes become female-specific only at later stages (5 days after hatching). In the ovary of both adult zebrafish and medaka, FIGalpha and zp expression is initiated in the smallest size oocyte population of the ovary (Kanamori, 2000; Zeng and Gong, 2002; this work). Further investigations are needed to identify upstream regulators of these genes.

We have shown that zpc and FIGalpha are coexpressed in oocytes during zebrafish development and that onset of their expression coincides with gonadal differentiation. Timing of zpc and FIGalpha expression, relative to gonad differentiation, is different in mouse and fish. Also, in contrast to mice, germ cells in fish enter meiosis before any morphologic differences between the two sexes are detectable. In mouse embryo, gonads first develop through a bipotential (indifferent) stage, during which time they have neither female nor male characteristics. Clear morphologic differences between the male and female gonad become apparent at 12.5 days postcoitum (dpc; Schmahl, 2000), at the time when germ cells, populating the gonad, still divide mitotically. In mouse female embryos, cells enter the prophase of the first meiotic division at 13.5 dpc and arrest at diplotene shortly after birth (McLaren and Southee, 1997). Onset of FIGalpha (13 dpc) or zpc (after birth; Epifano et al., 1995; Liang et al., 1997) occurs in mouse gonads already committed to female fate. In contrast, in zebrafish the process of gonadal differentiation occurs when some germ cells have entered meiosis and develop as oocytes, and overlaps with the onset of FIGalpha and zpc expression. Indeed, early diplotene oocytes were found in the ovaries of both sexes as early as 15 days posthatching (Uchida et al., 2002); zebrafish FIGalpha and zpc expression starts at day 19.5 days posthatching (22 dpf, this work), and male-specific apoptosis of the oocytes occurs at 23–30 days posthatching (Uchida et al., 2002). Therefore, it is conceivable that oocytes reaching a certain growth stage, for example diplotene or later, may intrinsically trigger expression of FIGalpha and zpc independent of the future sex of the fish. Subsequently, these oocytes become eliminated in male gonads by male-specific apoptosis. This finding may explain why in the present work we observed transient expression of zpc0.5:GFP in some males before the end of gonadal differentiation. Of interest, when removed from the gonad and cultured at ectopic positions, both male and female mouse early mitotic germ cells will develop as oocytes (McLaren and Southee, 1997; Adams and McLaren, 2002), suggesting that, without the influence of somatic tissue, a default female program of development may be intrinsic for both mammalian and zebrafish germ cells.

Practical Applications of the zpc0.5:GFP Transgenic Line

There is no established reliable procedure for morphology-based sex determination of zebrafish before they reach the adult stage. Thus, to learn the sex of juvenile fish, one must perform dissection and histologic examination. Although in the first few days after onset of expression, zpc0.5:GFP expression is not strictly female-specific, after 30–35 dpf, it can be reliably used to identify developing females, which allows determination of the sex ratio in juvenile fish populations and tracing of possible sex reversals in response to environmental or experimental insults. In addition, zpc0.5:GFP fish may be of use for studies of oocyte development and maturation (Selman et al., 1994). We envision further practical uses of zpc0.5:GFP fish, including the staging of maternal mutants with defects affecting distribution of cytoplasmic components or cytoplasmic movements, or even as a simple way to generate fluorescently labeled donor embryos for cell or tissue transplantation experiments to study cell autonomy of gene functions. Furthermore, the availability of a functional zpc promoter will make it possible to drive expression of transgenes specifically in the female zebrafish germline, as in mice, the generation of Zp3-cre transgenic animals has made specific recombination experiments possible in the female germline (Lewandoski et al., 1997).



Zebrafish were raised, maintained, and crossed as described (Westerfield, 1995). Staging was according to Kimmel and colleagues (1995). Development was carried out at 28°C. The age of embryos is indicated as hours postfertilization (hpf), the age of larvae as days postfertilization (dpf), and the age of juveniles as days posthatching. Progeny of crosses between AB and TL strain fish ( were used for injections.

Isolation of Genomic Zebrafish zpc Clones

RZPD Zebrafish genomic PAC library 706 was screened by using PCR primers for zebrafish zpc cDNA coding sequence (GenBank accession no. U55863) Zeb0 and ZebAC (Del Giacco et al., 2000). Three positive clones were isolated (BUSMP706D03169, BUSMP706D07169, BUSMP706K2086). To isolate the zpc upstream region, DNA of PAC clone BUSMP706K2086 was cut with SalI/KpnI. Southern blot-hybridization with digoxigenin (DIG)-labeled DNA probe corresponding to the first exon of zpc was carried out according to standard procedures. A 5.5-kb SalI/KpnI band was labeled with the probe and isolated from the gel, cloned into SalI/KpnI linearized pBluescript KS+ (Stratagene), and sequenced by using a Licor 4000L automated sequencer system.

Sequence analysis.

The promoter prediction programs MatInspector (Wingender et al., 2000) and Neural Network Promoter Prediction program (Reese and Eeckman, 1995) were used to identify TATA box sequence in zpc promoter.

Preparation of zpc0.5:GFP Promoter Construct and Generation of Transgenic Fish

To generate the zpc0.5:GFP promoter fusion construct, a 412-bp fragment immediately upstream of the start ATG of the ZPC ORF was amplified by PCR using forward primer DO3 (412 bp upstream of ATG sequence; 5′-AAAATCCCCATGACATGCTGC-3′) and reverse primer DO1 (5′-GCGGGATCC ATTGCCTGCTGACTAATTAAACC-3′) and ligated into the TOPO PCRII vector (Invitrogen) using the TOPO-cloning procedure. TOPO PCRII-zpGFP was cut with EcoRI/BamHI and ligated into EcoRI/BamHI linearized pG1 plasmid (personal gift of D. Gilmour and C.-B. Chien) containing the mmGFP5 coding sequence (Zernicka-Goetz et al., 1996).

Plasmid constructs zpc6:GFP and zpc0.5:GFP were prepared by using a miniprep kit (Qiagen), linearized, subjected to gel electrophoresis and further purified by using the MiniElute kit (Qiagen). Linearized plasmid was injected into one-cell stage embryos at 5 pg/embryo (solution of 50 ng/μl DNA in 100 mM KCl containing 0.02% Phenol Red). Fish were raised to adulthood and female fish were crossed to wild-type AB/TL strain fish. Germ line mosaic females were identified by GFP fluorescence detection in freshly laid eggs.

Because zpc0.5:GFP gave us the expected expression pattern, we did not repeat the experiments to generate zpc6:GFP transgenics. We think that zpc6:GFP transgenics would express GFP similar to zpc0.5:GFP, based on the following evidence. Fifteen percent of eggs laid by zp6:GFP female founder 1 were strongly expressing GFP. Unfortunately 100% of the eggs from female 1 (more than 2,000 of both GFP-positive and negative in total from 7 outcrosses) were unable to be fertilized and develop, presumably due to unrelated genetic defects. In three cases (founders 2, 3, and 4), the GFP-positive eggs were found in low abundance only after the first outcross, and were never observed in the next outcrosses of the same founder fish. None of the F1 embryos, developed from GFP-eggs of founders 2–4 (total n = 10), was genetically positive for zp6:GFP. This finding may be explained by the maternal mode of inheritance (GFP is expressed in the oocyte before meiotic cleavages and segregates to the polar body subsequently), or by mosaic GFP expression from amplified, nonintegrated zp6:GFP plasmid still present in the oocyte.

Cloning of the Zebrafish FIGalpha Gene

We performed a Blast search with a 193 amino acid protein sequence of medaka FIGalpha (GenBank accession no. AAD38902) against the zebrafish EST database (NCBI) and found two most closely matching zebrafish sequences (GenBank accession nos. BM156452 and BM156753, both ESTs were assigned by submitter as medaka FIGALPHA homologs). Translation of a 1,064 bp assembly of these overlapping zebrafish FIGalpha sequences closely matches the medaka FIGalpha gene, although there are two open reading frames, probably due to the sequencing mistake (amino acids 1–55 of medaka FIGalpha match to frame 2, and amino acids 55–189 to frame 3). We deduced that the 5′ ATG codon of the open reading frame 2 (located at base pair position 86-88 in BM156452) is the translation start of zebrafish FIGalpha, because the upstream sequence contains stop codons in all three reading frames shortly preceding this ATG. Frame 3 contains homology to the helix-loop-helix domain of medaka FIGalpha and a TAG stop codon at position 747-749 of the assembly (316-318 of BM156753). We used PCR primers BamHIFigalphaf and EcoRIFigalphar encompassing predicted start and stop codons, respectively, to PCR-amplify the putative protein-coding region of FIGalpha from a zebrafish ovary cDNA as template. The PCR product was linearized with BamHI/EcoRI and cloned into BamHI/EcoRI linearized pBluscript vector (Stratagene).

PCR primers for cloning (restriction sites in italics): Figalphaf 5′ ATG TCG TGT GAA ATG ACC GGC 3′; Figalphar 5′ CTAGGATGGGAGTGAACTTGG 3′; BamHIFigalphaf 5′ GGGGGATCCATGTCGTGTGAAATGACCGGC 3′ EcoRIFigalphar 5′GGGGAATTCCTA-GGATGGGAGTGAACTTGG 3′.

The multiple sequence alignment for FIGalpha (Fig. 4A) and the tree (Fig. 4B) were built by using the ClustalX program (Jeanmougin et al., 1998). Similarities and identities in the alignment were boxed using the Boxshade server in


RNA isolation, reverse transcription, and semiquantitative PCR was carried out as previously described (Onichtchouk et al., 1996). The mRNA for the ubiquitous translation initiation factor EF1alpha was used for normalization. Sample PCR conditions were as follows: initial denaturation step 95°C for 3 min, then 95°C for 30 sec, 58°C for 30 sec, 72°C for 45 sec; 19 cycles for EF1alpha, 27 cycles for zfZPC and 30 cycles for FIGalpha. Primers used for EF1 alpha: EFf, 5′-CCCTGGACACAGAGACTTCATC-AAG-3′; Efr, 5′-AGCATGTTGTCACCGTGCCATCCT-3′; for FIGalpha: Figalphaf, 5′-ATGTCGTGTGAAATGACCGGC-3′; Figalphar, 5′-CTAGGATGGGAGTGAACTTGG-3′; and for zfZPC as in Del Giacco et al., (2000), Zeb3 5′-CCT CTC CAG TCC AGC AGC-3′; ZebAS, 5′-TTG AGA TTT TAT AAA ACA TTT TAT TTC-3′.

In Situ Hybridization

To obtain the DNA template for zpc probe, we PCR-amplified the region 1221-1736 of zpc nucleotide sequence from zebrafish ovary cDNA by using primers Zeb3BamHI and ZebASEcoRI. The PCR product was EcoRI/BamHI cut and ligated into EcoRI/BamHI linearised pBluescript vector (Stratagene): Zeb3BamHI 5′-GGG GGA TTC CCT CTC CAG TCC AGC AGC-3′; ZebASEcoRI 5′-GGG GGA TCC CCT CTC CAG TCC AGC AGC-3′.

Ovaries were taken from females anesthetized with an overdose of Tricaine (Sigma) and either fixed for whole-mount in situ hybridization or frozen in liquid nitrogen. Whole-mount in situ hybridization of DIG-labeled antisense probes to ovaries was carried out as described in (Hauptman and Gerster, 1994) with prolonged proteinase K treatment of fixed tissues: 50 μg/ml proteinase K, 15 min at 37°C as in Kanamori (2000). Five- to 10-μm cryosections were cut from frozen tissue, dried, fixed in 4% paraformaldehyde for 5 min, treated with Proteinase K (10 μg/ml in phosphate buffered saline [PBS]) for 30 min. The following protocol was then applied: 3 min PBS; 2 min 0.1 M triethanolamine; 10 min 0.1 M triethanolamine + 0.25% acetic anhydride; 5 min PBS, 2 min each in 70%, 80%, 90%, 100%, 100% EtOH; dry the slides; add 80 μl of hybridization mix + DIG-labeled RNA (5ng/μl); apply coverslip on the slides; and hybridize overnight in a wet chamber with 50% formamide/4× standard saline citrate (SSC). After careful removal of the coverslip, wash 5 min 6× SSC at 65°C, 30 min with 50% formamide/2× SSC/10 mM ethylenediaminetetraacetic acid at 65°C, 2 × 30 min 2× SSC at 65°C, 2 × 30 min 0.2× SSC at 65°C, 10 min MAB buffer, 1 hr 2% blocking reagent/20% fetal calf serum in MAB (blocking buffer); 30-min incubation with DIG-AP conjugate in blocking buffer (1:1,000), 3 × 5 min MAB, 5 min NTMT; stain in BM-Purple, PBS overnight; 2 min each in 70%, 80%, 90%, 100%, 100% EtOH; air-dry; and embed in glycerol-gelatin.

Fluorescence Microscopy and Imaging

Images were taken by using an Axiophot 2 compound microscope (Zeiss) or a MZFLIII stereomicroscope (Leica) both equipped with epifluorescence. For Figure 2A,B, two images were taken at the same focal plane in transmitted light by using an enhanced GFP filter and then superimposed and processed by using the Adobe Photoshop program.


We thank Annette Schult for excellent technical assistance and Sabine Götter and Ralf Schlenvogt for excellent animal care. Thanks also to Andrzej Nasjadka, Soojin Ryu, Karen Lunde, and Dirk Meyer for critical comments on the manuscript and D. Gilmour and C.-B. Chien for the gift of the pG1 plasmid. W.D. received funding from BMBF-DHGP, DeveloGen AG, Göttingen, and the EU.