Isolation of Zebrafish zic Genes
Isolation of zebrafish odd paired-like (opl), which is most closely related to mammalian zic1, has been reported previously (Grinblat et al., 1998; Rohr et al., 1999). opl/zic1 belongs to the zic gene family whose members are characterized by patterned expression in the gastrula ectoderm (Kuo et al., 1998; Nagai et al., 1997; Nakata et al., 1997, 1998). Modern members of the zic family are likely to have originated from a single common ancestor gene, because intron positions are precisely conserved in Drosophila opa (Benedyk et al., 1994), mouse zic genes (Aruga et al., 1996b), and zebrafish opl (Grinblat et al., 1998). There are at least three, and as many as five, distinct zic genes found in each of the vertebrate species examined (Furushima et al., 2000; Nakata et al., 2000).
The Zic family of proteins is characterized by the presence of five zinc fingers of the C2H2 class, which are highly conserved both within vertebrates and with the Drosophila opa protein (Benedyk et al., 1994; Aruga et al., 1996a). In addition, Zic proteins contain a large, poorly conserved N-terminal region of unknown function. A short, variable length C-terminal domain follows the zinc finger region, and ends in a highly conserved short peptide NFNEWYV.
To find new members of the zebrafish zic gene family, partial cDNA clones were isolated from late gastrula stage embryonic cDNA by using degenerate polymerase chain reaction (PCR). Complete cDNA clones were then isolated by high-stringency hybridization to a mid-gastrula cDNA library (see Experimental Procedures section for a detailed description). Predicted amino acid sequences are shown in Figure 1 aligned to the known vertebrate zic genes. This alignment and the corresponding evolutionary tree (not shown) suggest that the newly identified genes encode the zebrafish Zic2 and Zic3 proteins. Overall, the predicted zebrafish Zic proteins are more similar to their respective mouse and Xenopus orthologs (80–90% sequence similarity) than to each other (approximately 60% sequence similarity).
Figure 1. Sequence alignment between predicted open reading frames of zebrafish opl/zic1 (Grinblat et al., 1998), zebrafish zic2, zebrafish zic3, and representative members of the zic gene family found in mouse (Aruga et al., 1996a) and Xenopus laevis (Nakata et al., 1997, 1998; Kuo et al., 1998). Zinc finger region is shown with a bar over the sequence. Positions of sequence identity are indicated with black background. Dr, Danio rerio; Xl, Xenopus laevis; Mm, Mus musculus.
Download figure to PowerPoint
Patterned Expression of zebrafish zic Genes in the Presumptive Neurectoderm During Gastrulation and Somitogenesis
Spatial and temporal distributions of zic2 and zic3 transcripts were examined by whole mount in situ hybridization and compared with opl. In the early gastrula (shield stage) embryo opl/zic1 is not expressed (Fig. 2A) and zic2 is expressed in the blastoderm margin only (Fig. 2B). zic3, in contrast, is expressed in the blastoderm margin (presumptive mesendoderm) and in the posterior dorsal quadrant of the prospective ectoderm (Fig. 2C). This early restriction of zic3 expression, which is not observed in either Xenopus or mouse embryos (Nagai et al., 1997; Nakata et al., 1997), shows that the early gastrula ectoderm is asymmetric along its A/P axis.
Figure 2. zic gene expression is patterned in the neurectoderm during gastrulation. Staged embryos were stained for opl/zic1, zic2, zic3 (purple), and otx2 RNA (orange in G–I) by using whole-mount in situ hybridization. A–C: Early gastrula (shield) stage embryos, dorsal view, anterior at the top. A:opl/zic1 is not expressed. B:zic2 is expressed in the blastoderm margin, the presumptive mesendoderm, only. C:zic3 expression in the posterior portion of the dorsal ectoderm and in the blastoderm margin. D–F: Mid-gastrula (80% epiboly) stage embryos, dorsal view, anterior at the top. D:opl/zic1 expression restricted to an anterior domain of the neural plate. E:zic2 expression domain includes the anterior opl domain and extends more posteriorly. F:zic3 is expressed in the posterior portion of the neural plate and at the edge of the neural plate anteriorly. G–I: Late gastrula (90% epiboly) stage embryos, dorsal view, anterior at the top. G,H: Expression of opl (G) and zic2 (H), shown in purple, overlap the otx2 (orange) expression domain in the presumptive forebrain. I: anterior portion of zic3 expression overlaps the posterior portion of the otx2 domain. Dots indicate outline of the blastoderm margin. White arrowheads point to the posterior extent of otx2 staining, which corresponds to the future posterior midbrain.
Download figure to PowerPoint
By mid-gastrulation all three genes are expressed in the dorsal ectoderm of the zebrafish embryo, in overlapping but distinct domains: opl/zic1 is expressed primarily in the anterior-most domain of the neural plate (Fig. 2D), zic2 expression includes the anterior opl/zic1 domain and extends more posteriorly (Fig. 2E), and zic3 is expressed most strongly in the posterior portion of the neural plate (Fig. 2F). All three genes overlap the domain of otx2 expression that marks the presumptive forebrain and midbrain (Li et al., 1994): opl/zic1 and zic2 anteriorly (Fig. 2G and H, respectively) and zic3 posteriorly (Fig. 2I). By mid-somitogenesis, the three zic genes are expressed in the dorsal brain (Fig. 3A–C) and, by the end of somitogenesis, are expressed in overlapping but distinct patterns in the dorsal brain (Fig. 3D–F), as are their orthologs in Xenopus and mouse (Nagai et al., 1997; Kuo et al., 1998; Nakata et al., 1998).
Figure 3. Patterned zic gene expression in the dorsal brain during somitogenesis. Staged embryos were stained for opl/zic1, zic2, or zic3 RNA (purple) by using whole-mount in situ hybridization. A–C: Mid-somitogenesis (15 hpf) embryos, lateral view, anterior to the left, dorsal at the top. A:opl/zic1 is expressed in the dorsal brain and in the dorsal half of each somite (Rohr et al., 1999). B:zic2 is expressed in the dorsal brain, the forming optic stalk, and weakly in the developing tailbud. C:zic3 is expressed in the dorsal brain and strongly in the tailbud; also note transient expression in the ventral diencephalon. D–I: Late somitogenesis (prim-5) embryos, side view, anterior to the left, dorsal at the top. D:opl/zic1 is expressed throughout the dorsal brain and in the optic stalk. E:zic2 expression is restricted to the dorsal diencephalon, dorsal portion of the cerebellum, and the optic stalk (confirmed in sections, data not shown). F:zic3 is expressed in the dorsal diencephalon, tectum, and the optic stalk. a, anterior; p, posterior; t, telencephalon; di, diencephalon; m, midbrain; h, hindbrain; asterisks indicate optic stalk.
Download figure to PowerPoint
Expression of zic3 Marks an Early Posterior Domain in the Dorsal Ectoderm
To examine the early asymmetric pattern of zic3 expression in more detail we examined its expression relative to the ventral marker bmp2b/swr (Kishimoto et al., 1997; Nguyen et al., 1998). zic3 expression is first detectable in late blastula embryos and is restricted to the dorsal half of the blastoderm (data not shown). Double in situ hybridization for zic3 and bmp2b/swr, which is expressed through most of the blastoderm at this stage (Nikaido et al., 1997; Koos and Ho, 1999), shows that the two domains overlap significantly. In early gastrula embryos, zic3 continues to be strongly expressed in the dorsal ectoderm but is now restricted to its posterior portion (Fig. 4A). The expression domain of bmp2b/swr is now ventrally restricted in the ectoderm (Nikaido et al., 1997; Koos and Ho, 1999) and is complementary to the zic3 expression domain (Fig. 4A).
Figure 4. Early pattern in the ectoderm, marked by zic3 expression, is regulated by bone morphogenetic protein (BMP) signaling. Embryos stained by whole-mount in situ hybridization (ISH) for expression of zic3 (orange in A, purple in C–F) and bmp2b/swr (purple in A). All embryos are shown in animal pole view, dorsal to the right. A:zic3 (orange) and bmp2b/swr (purple) are expressed in complementary domains in wild-type embryos. B: Diagram of a specification map, derived from in vitro explant culture assays, of the early gastrula ectoderm (Grinblat et al., 1998). V, ventral non-neural ectoderm; A, anterior forebrain; P, posterior forebrain. C: Embryo derived from a din/+x din/+ cross, showing a wild-type pattern of zic3 expression. D: Sibling embryo from the same cross, showing a strong reduction in dorsal zic3 expression. Dorsal marginal staining remained unaffected (not shown). E: Embryo derived from a snh/+ x snh/+ cross, showing a wild-type pattern of zic3 expression. F: Sibling embryo from the same cross, showing up-regulation of zic3 expression in ventral posterior ectoderm (white asterisk). Note that the animal pole is still devoid of zic3 expression. d, dorsal; white asterisk marks ventral side.
Download figure to PowerPoint
Three domains of specification in the early gastrula ectoderm of zebrafish have been defined by using explant assays (Grinblat et al., 1998). Although this method is a powerful way to examine lineage commitment, it is an indirect assay. Tissue corresponding to the three specification domains, when explanted and cultured in isolation from the embryos, goes on to express markers characteristic of ventral non-neural (V), of anterior forebrain (A), and of posterior forebrain (P) development, respectively (Fig. 4B). The V domain of the specification map coincides with the high expression domain of bmp2b/swr and does not express significant levels of zic3. The A domain corresponds to the area that expresses low levels of both zic3 and bmp2b. Strong expression of zic3 is limited to the area corresponding to the P domain of the specification map. Together, these two genes represent direct markers of the asymmetry in the early gastrula ectoderm previously shown in indirect explant culture assays.
Dorsal, but not Posterior, Restriction of Early Gastrula zic3 Expression Is Dependent on BMP Function
Several signaling pathways are known to regulate pattern formation in the gastrula ectoderm, among them the BMP signaling pathway (Solnica-Krezel, 1999; Gamse and Sive, 2000). At mid- to late gastrula dorsal expression of opl/zic1 is dependent on BMP signaling in zebrafish (Grinblat et al., 1998) and in Xenopus (Gamse and Sive, 2001). However, the A/P pattern in the zebrafish ectoderm is regulated independently of BMP signaling starting at mid-gastrula stages (Barth et al., 1999; Nikaido et al., 1999; Nguyen et al., 2000). The complementary expression patterns of zic3 and bmp2b shown above raise the possibility that, in early gastrula ectoderm, BMP signaling may regulate expression of zic3. We tested this possibility by using mutants in the BMP pathway.
In zebrafish, two members of the BMP family are known to be expressed at high levels during early gastrula stages of development: BMP2b, encoded by the swr locus (Kishimoto et al., 1997; Nguyen et al., 1998), and BMP7, encoded by the snh locus (Dick et al., 2000; Schmid et al., 2000). A secreted antagonist of BMP signaling, chordin, is encoded by the din locus (Fisher et al., 1997; Schulte-Merker et al., 1997). Mutant alleles for all three loci have been isolated and characterized molecularly (see references above, and Fisher and Halpern, 1999). This analysis showed that loss of din function leads to expansion of ventral fates (Hammerschmidt et al., 1996), whereas loss of bmp2b/swr or bmp7/snh function causes loss of ventral fates and expansion of dorsal and lateral tissues (Mullins et al., 1996). However, the effect of these mutations on A/P pattern in the ectoderm at early gastrula stages has not been described.
To test whether BMP antagonism is required for correct expression of zic3 in the early gastrula ectoderm, we used an allele of din, which encodes a truncated chordin variant and is thought to be a genetic null (Fisher and Halpern, 1999). Embryos derived from a cross between din/+ parents were stained by whole-mount in situ hybridization for expression of zic3. Wild-type patterns were observed in 75% of the embryos (28 of 37 embryos; Fig. 4C), whereas 25% of the embryos (9 of 37) failed to express zic3 (Fig. 4D). This finding is consistent with a requirement for din function to activate zic3.
Embryos homozygous for mutant alleles in bmp2b/swr or bmp7/snh were similarly obtained from crosses between heterozygous parents. The swrtc300 allele used encodes an altered BMP2b protein with a mis-sense mutation in the mature domain and is thought to function as a dominant negative (Kishimoto et al., 1997). The snhty68a allele has been shown to act as a temperature-sensitive null allele at the restrictive temperature used during all experiments (Dick et al., 2000). Identical defects in zic3 expression were observed for both mutants and involved a ventral up-regulation of zic3 expression in the ectoderm. For swr, wild-type expression of zic3 was observed in 75% of the embryos and expanded expression in 25% of the embryos (3 of 12, data not shown). For snh, 16% of the embryos showed the defect (36 of 227, Fig. 4F). This rate is less than the expected Mendelian ratio of 25%, suggesting either that the mutant protein is partially functional at the restrictive temperature or that zic3 expression can be restricted dorsally even in the absence of function at the bmp7/snh locus. Importantly, in both mutants zic3 expression continued to be excluded from the anterior region of the ectoderm. These data are consistent with the anterior exclusion of zic3 being regulated independently of BMP signaling.
An alternative explanation is that bmp2b/swr and bmp7/snh play redundant roles in restricting anterior expression of zic3. If that is the case, simultaneous removal of both is expected to allow ectopic zic3 expression anteriorly. swr; snh double mutant embryos were generated from a cross between swrtc300/+;snhaub/+ parents. The allele we used for this analysis, snhaub, has been shown to be a genetic null (Schmid et al., 2000). Of the 117 embryos tested, 37.5% (44 embryos) were expected to be homozygous for either swrtc300 or snhaub and 6.25% (7 embryos) were expected to be homozygous for both. We observed 34% (40 embryos) with zic3 expression expanded ventrally, and none with zic3 expression expanded into the anterior ectodermal region. Therefore, swrtc300;snhaub mutant embryos are phenotypically identical to single mutant embryos, and do not exhibit anterior de-repression of zic3 transcription. The difference between the expected (43.75%) and observed (34%) numbers of embryos showing the mutant phenotype may be due to zic3 expression being patterned correctly in a proportion of the snhaub embryos, as predicted by the results of single mutant analysis above. We conclude that BMP signaling is required for dorsal restriction of zic3 expression in the early gastrula ectoderm but is not required for its posterior restriction.