Expression of the zic1, zic2, zic3, and zic4 genes in early chick embryos

Background The zic genes encode a family of transcription factors with important roles during early development. Since little is known about zic gene expression in chick embryos, we have characterized the expression patterns of the zic1, zic2, zic3, and zic4 (zic1-4) genes during neurulation and somitogenesis. Findings We used in situ hybridization to analyze the expression patterns of the zic1-4 genes during early chick development (HH stages 7-19). The zic1-3 genes showed both overlapping and gene-specific expression patterns along the length of the dorsal neural tube and in the dorsal parts of the somites. In addition, unique expression domains of zic genes included: zic2 in the neural plate, periotic mesoderm and limb buds; zic3 in the paraxial mesoderm surrounding the neural plate, in presomitic mesoderm and in the most recently formed epithelial somites; zic2 and zic3 in developing eyes. zic4 expression was limited to dorsal fore- and midbrain regions and, unlike the expression of the zic1-3 genes, zic4 expression was not detected in the hindbrain and trunk. This was in contrast to more extensive zic4 expression in other vertebrates. Conclusions The zic1-3 genes were expressed in both overlapping and unique domains within the neural tube, somites and other ectoderm and mesoderm-derived structures in the future head and trunk. zic4 expression, however, was limited to dorso-anterior regions of the future brain. This is the first comprehensive study of zic1-4 gene expression in chick embryos during neurulation and somitogenesis.


Background
The zic genes encode a family of zinc finger transcription factors. Five zic genes are typically found in vertebrates (zic1-5), where they guide a variety of developmental processes [1] and zic genes play significant roles during early neural patterning and neural crest formation [2][3][4][5][6][7][8][9][10][11]. Although zic genes may be able to partially compensate for each other [12,13], mutations in individual Zic genes in mice and humans produce distinct phenotypes [14]. For example, compromised expression of the Zic2 gene results in neural tube defects and mutation of the Zic3 gene causes left-right abnormalities in addition to less severe neural tube defects [15][16][17][18][19][20][21][22][23]. In addition, mutations in the Zic1 or Zic4 genes result predominantly in cerebellar abnormalities [13,24]. Collectively, these phenotypes illustrate the relevance of zic genes to early developmental processes. However, much further study is required to elucidate the mechanisms that underlie the phenotypes associated with mutations of individual zic genes.
Chick embryos represent a major developmental model system that remains largely untapped for the study of zic genes. This is partly due to a lack of characterization of zic gene expression during chick development, which has been mainly restricted to studies of the zic1 gene [25][26][27][28]. Specifically, there are no published micrographs of zic2 gene expression patterns prior to stage 23 [27] and descriptions of zic3 expression are limited to brain and anterior trunk regions at stages 10, 12, and 17 [29], and to a stage 26 embryo [28]. In addition, the expression of zic4 in chick embryos has not been described. In this study, we thoroughly examine zic1-4 gene expression in stage 7-19 chick embryos during neurulation and somitogenesis. Care was taken to use gene-specific probes in order to avoid potential cross reactivity among zic genes. This study differs from previous work by examining zic1-3 expression at earlier stages of chick development (beginning with Hamburger Hamilton HH stage 6+/7), in reporting for the first time on the expression of the zic4 gene, and by providing a comprehensive comparison of zic1-4 expression patterns up to stage 18/19. Thus, this is the most thorough comparative expression study of the zic1-4 genes during early development in chick embryos. As such, it contributes a foundation for future studies of Zic transcription factors during early neural and somite development, where zic genes are known to play important roles.

Genomic location of the zic1-4 genes
The genomes of most vertebrates contain five zic genes (reviewed in Merzdorf, 2007). In the chicken genome, the zic1 and zic4 genes are adjacent and transcribed in opposite directions on chromosome 9, while the zic2 and zic3 genes are located on chromosomes 1 and 4, respectively. In other vertebrates, the zic1 and zic4 genes and the zic2 and zic5 genes are adjacent and transcribed in opposite directions. However, zic5 has not yet been identified in the chicken genome. In mouse, the Zic5 gene is located between the Zic2 and Clybl genes. The equivalent intergenic region in the chicken genome sequence contains several gaps and will thus require additional work to conclusively demonstrate the presence or absence of a zic5 gene.
Members of the zic family show high sequence similarity, particularly throughout their zinc finger domains and in the regions immediately flanking them. Therefore, we used the published cDNA sequence for chicken zic1 [25] and the sequence of the chicken genome to select regions for the zic1-4 genes that were highly specific for each gene. Antisense RNAs for in situ hybridization were synthesized from these regions and used to study zic1-4 gene expression in whole embryos and cryosections.

Comparison of zic1-4 gene expression patterns
We describe zic1-4 gene expression during neurulation and somite formation in chick embryos spanning stage HH 6/7 to stage HH 18/19. Head neural fold formation begins at stage 6/7, somite formation at stage 7, and neural tube closure starts at the level of the mesencephalon at stage 8 (4-6 somites). During stages 6-15, gastrulation, neural plate formation, somite formation, and neural tube closure progress in an anterior to posterior direction. In stage 18/19 embryos, the posterior neuropore has closed, the tailbud has formed and limb buds have begun to develop [30,31]. The expression levels of the zic1-3 genes were extremely low and detectable only slightly above background in early embryos at stages 6/7 and up to stage 12. The zic4 gene was not expressed in these early embryos. In stage 14/15 and 18/19 embryos, the expression of the zic1-3 genes in the head was substantially stronger and the zic4 gene was expressed in head regions.

Brain
In the developing brain, the expression patterns of zic1-3 overlapped extensively, but differed significantly from that of zic4. The expression levels of the zic1-3 genes were very low in embryos at early stages and became more robust in stage 14 and older embryos. At stage 7, zic1 was expressed in the anterior head fold ( Figure 1A). In contrast, zic2 was expressed in the head fold and along the entire anterior-posterior axis of the neural plate, including the midbrain region (Figure 2A, B). At stage 7, zic3 was expressed in discontinuous regions in the neural folds of the future forebrain and hindbrain ( Figure 3A). At stage 8, the zic1 and zic3 genes were expressed in the future forebrain and hindbrain ( Figures 1B; 3B). By stage 9, the forebrain expression domains of zic1-3 were very similar and well defined ( Figures 1C, 2C, 3C). In the midbrain, zic2 expression ceased transiently ( Figure 2C), resulting in a period during which no zic genes were expressed in the midbrain region. In the hindbrain, zic1 expression was limited to rhombomere r4 ( Figure 1C, E; white arrowheads), while zic2 and zic3 were expressed in almost identical patterns in most of the hindbrain at stage 9 ( Figures 2C, 3C) and throughout the brain at stage 11 (Figures 2E, F; 3F). By stage 14/15, the expression patterns of zic1-3 were highly similar and continuous along the forebrain, midbrain, and hindbrain ( Figure 4A-C, E-G). Although expression of zic1-3 in the brain increased in stage 18/19 embryos, the general expression patterns remained largely unchanged from those at stage 14/15 ( Figure 5A-C, I-K). While the expression patterns of zic1-3 overlapped extensively, they differed significantly from that of zic4. Expression of zic4 could only be detected from stage 14, when it was restricted to the anterior telencephalon (arrow, Figure 4D). By stage 18/19, its expression extended to the diencephalon (in a region posterior to that of zic1-3 expression) and to the mesencephalon ( Figure 5D, H, L), but not to the hindbrain.
The observed expression of the zic1-3 genes in the chick forebrain can be related to deficiencies in neural development observed in mice and humans with compromised Zic gene expression. Among the zic genes in the chick, zic2 showed the earliest and most extensive expression in the anterior neural plate/head fold ( Figure 2A) and, accordingly, compromised Zic2 expression in mice or humans causes holoprosencephaly, in which part of the forebrain does not develop properly [16,20,23,32]. Further, mutations in the Zic1 or Zic3 genes alone do not cause obvious forebrain abnormalities but Zic1/Zic3 compound mutant mice show forebrain defects that are very similar, although not as severe, as those observed in Zic2 hypomorphs [11]. This indicates that Zic1, 2, and 3 are involved in forebrain development of mouse embryos. The expression patterns of the zic1-3 genes in chick embryos suggest that these genes are involved in forebrain development as well.
The importance of Zic transcription factors in brain development is also suggested by experiments in zebrafish and Xenopus. In zebrafish, the zic2 gene is required during formation of the anterior diencephalon [33] and the zic2 and zic5 genes are important during dorsal midbrain development [10]. Further, the zic1 and zic4 genes are necessary for hindbrain ventricle morphogenesis from the level of r2 towards the posterior [3]. These results corroborate the importance of zic1 and zic2 during brain development. Xenopus laevis embryos express zic1-5 during early development [34]. Several genes that are regulated by zic genes in Xenopus embryos are expressed in domains throughout the brain. For example, Zic1 induces expression of the wnt1 and en-2 genes and, therefore, is likely to play a role in midbrain/ hindbrain boundary (MHB) development [4,5]. Since chicken zic1 does not appear to be expressed in the region of the early MHB ( Figure 1B), zic2 or zic3 ( Figures  2B, 3B) may potentially play roles in MHB formation in birds. Other genes that are induced by zic genes in Xenopus include the forebrain/midbrain boundary gene wnt8b [5], the dorsal neural tube gene pax3, the hindbrain genes krox20 [4] and Xfeb, which regulates hoxB1 expression [35]. hoxB1 is expressed specifically in r4 [36], where Figure 1 Expression of the zic1 gene in whole mount chick embryos at HH stages 7-11 determined by in situ hybridization. A: In stage 7 embryos, zic1 is expressed at very low levels in the most anterior portion of the head fold (white arrow). B: zic1 is expressed in the neural folds of the future forebrain (white arrow) and hindbrain (white arrowheads) at stage 8. It is not expressed in the future spinal cord or in somites (black arrows). C: By stage 9, zic1 is expressed in the anterior region of the future forebrain and in a domain of the future hindbrain (white arrowheads). D: Very weak zic1 expression in anterior somites is first observed at stage 10 (black arrowheads). E (dorsal view), F (side view): zic1 expression in anterior somites becomes more robust by stage 11 (black arrowheads), as does zic1 expression in the anterior portion of the trunk neural tube (E). In the brain, zic1 continues to be expressed in the anterior prosencephalon (best seen in F, white arrow) and in the limited hindbrain domain (best seen in E, white arrowheads). zic1 is not expressed in the optic vesicles. G: At stage 11, anterior somites express zic1, while it is not expressed in posterior somites. zic1 expression in the trunk neural tube diminishes towards posterior regions. H: zic1 is not expressed in the neural tube, neural folds, or somites of the posterior trunk in stage 11 (and older) embryos. Overall, expression levels of zic1 are very low in these early stage embryos. FB: future forebrain; MB: future midbrain; HB: future hindbrain; HN: Hensen's Node; PS: primitive streak; NF: neural folds; Pro: prosencephalon; Mes: mesencephalon; Rhom: rhombencephalon; white arrowheads: hindbrain expression domain of zic1; white arrow: forebrain expression domain of zic1; black arrows: somites not expressing zic1; black arrowheads flank somites expressing zic1; the anterior of the embryos in all figures is to the left. early chick zic1 was expressed ( Figure 1C, E). Further, direct targets of Zic1 in Xenopus include genes that modulate retinoic acid signaling, suggesting further roles for zic genes in hindbrain development [37]. Thus, the expression patterns in the future brain of early chick embryos supports numerous roles for zic genes in neural development.

Eyes
The developing eyes in stage 9 to stage 18/19 chick embryos did not express zic1 or zic4 (Figures 1, 4, 5). zic2 expression began in the optic vesicles at stage 11 ( Figure   2E, F) and zic3 expression at stage 13 ( Figure 3H). All four zic genes were expressed in the optic stalk ( Figure 5E-H). The significance of zic2 and zic3 expression for optic vesicle development remains to be determined.

Neural tube of the trunk
In chick, the zic1-3 genes were expressed in the neural tube of the trunk. zic3 expression preceded zic1 expression in the anterior trunk neural tube (stage 7/8 for zic3; stage 9 for zic1) and, while expression levels were low, the expression of both genes tapered towards the caudal region ( Figures 1C; 3B, C). zic2 expression in the trunk is not yet expressed in somites of stage 10 embryos. E, F: dorsal view (E) and side view (F) of stage 11 embryo; although zic2 expression appears weaker in the mesencephalon region (E), the side view shows that zic2 expression is present along the entire dorsal brain and trunk neural tube (F). The optic vesicles express zic2 at stage 11. G: Weak zic2 expression begins in the anterior somites of stage 11 embryos. H: zic2 is expressed in the most posterior neural tube and neural folds. I: in stage 12+ embryos, zic2 expression in anterior somites is more robust and zic2 is expressed in the periotic mesoderm (bracket). Overall, expression levels of zic2 are very low in early stage embryos. Abbreviations as in legend for Figure  1; NP: neural plate; black arrows: somites not expressing zic2; black arrowheads flank somites expressing zic2; bracket: periotic mesoderm.
neural tube was unique, since it was expressed at low levels very early during formation of the neural plate (stage 6+; Figure 2A) and continued to be expressed along the entire trunk neural tube, trunk neural folds and neural plate throughout further early development ( Figure 2B  It is expressed in the neural folds of the hindbrain and more weakly in the anterior trunk. C, D, D': In stage 9 embryos (7-9 somites), the first epithelial somites (somites I-III and sometimes somite IV) express zic3, while more anterior somites do not express zic3 (black arrows, C). zic3 is expressed at low levels in the presomitic mesoderm and at a slightly increased level in the area that will form the next somite (red arrowheads). D and D': posterior region of a stage 9 embryo. In D', the area that represents the neural plate and expresses zic2 (see Figure 2B) is outlined with red dots. The outer limit of the adjacent presomitic mesoderm, which expresses zic3, is marked by black dots. E: At stage 10, low levels of zic3 expression continue in presomitic mesoderm and in epithelial somites I-III. The neural folds do not express zic3. F: zic3 is first expressed along the entire brain in stage 11 embryos. Anterior somites do not express zic3 (black arrows). G, H: By stage 13, anterior somites (H, black arrowheads) and the optic vesicles first express low levels of zic3. The presomitic mesoderm expresses zic3, while there is very little or no expression of zic3 in epithelial somites (G). The brain and anterior neural tube express zic3 (H), while the posterior neural tube does not (G). Abbreviations as in legend for Figure 1; NP: neural plate; black arrows: somites not expressing zic3; black arrowheads flank somite showing weak zic3 expression; red arrowheads: increased zic3 expression in presomitic mesoderm.
hindbrain was discontinuous ( Figure 1E Cross-sections of stage 14/15 embryos allowed more precise location of zic gene expression following in situ hybridization. The zic1-3 genes were expressed in the dorsal neural tubes of the mid-trunk and anterior trunk ( Figure 6D, E, J, K, P, Q) and in the dorsal hindbrain (Figure 6F, L, R), whereas zic4 was neither expressed in the trunk nor hindbrain ( Figure 6S-X). In the tail tip region, zic2 was expressed not only in the dorsal portion, but also throughout the remainder of the neural tube ( Figure 6G).
In mice, decreased zic2 expression causes both holoprosencephaly and spina bifida, a posterior neural tube closure defect [16,20]. Thus, expression of chicken zic2 in the neural plate and neural folds ( Figure 2B, H) may be required for neural tube closure. This appears to be true for Zic proteins in Xenopus and zebrafish, since zebrafish zic2a and zic5 are necessary for hinge point formation during neurulation [38]. Further, we found that Zic factors directly induce an aquaporin gene (aqp-3b) in Xeno-pus that is specifically expressed in the neural folds and may contribute to neural tube closure [37].

Somites
During chick somite development, the presomitic mesoderm segments into epithelial somites (somites I-V). As new somites form, older somites compartmentalize into an epithelial dermomyotome and mesenchymal sclerotome. zic gene expression in the somites proceeded in two phases. During the first phase only zic3 expression occurred in newly formed somites, while in the second phase zic1-3 were expressed in more mature somites. Phase 1: In stage 7-10 embryos, zic3 was expressed at low levels in presomitic mesoderm, which included a stripe of slightly increased zic3 expression, where the next somite pair would form (red arrowheads, Figure 3C, D, D', E). In addition, zic3 was expressed in two or three of the most recently formed somites ( Figure 3C-E). By stage 12, zic3 expression was largely limited to somite I and was uniformly expressed in presomitic mesoderm (not shown). The expression of zic3 in the presomitic mesoderm (Figure 3C, D) surrounded the expression domain of zic2 in the neural plate ( Figure 2B). The relationship between these expression patterns is shown in Figure 3D', where red dots outline the zic2 expression domain (neural plate; see Figure 2B) and black dots indicate the outer limit of zic3 expression (mesoderm). Phase 2: The zic1-3 genes were expressed in the most anterior, mature somites. The onset of very low, but detectable expression in anterior somites was stage 10 for zic1 ( Figure 1D), stage 11 for zic2 ( Figure 2G), and stage 12/13 for zic3 ( Figure 3H). In stage 14/15 embryos, zic1 expression extended to the most posterior somites ( Figure 4I, M), including the dorsomedial region of recently formed somites ( Figure 6C) [26]. zic2 and zic3 expression, however, was absent from more posterior somites ( Figures 4J, K, N, O; 6I, J, O). In stage 14/15 embryos, zic3 continued to be expressed in pre-somitic mesoderm (Figures 4O; 6M, N). Cryosections of mature somites in the anterior trunk showed that zic1 and zic2 were expressed in both the dermomyotome and in the sclerotome ( Figure 6D, E, K) [26], while zic3 was expressed in the sclerotome with some expression in the dermomyotome of the anterior trunk ( Figure 6P, Q). In stage 18/19 embryos, zic1-3 were expressed in somites along most of the anterior/posterior axis ( Figure 5M-O,  Q-S). Since zic4 was not expressed in the trunk, no expression was detected in somites (Figures 4L, P; 5D, P, T; 6U-W). Migrating neural crest cells do not express zic1 [26] and serial sections suggest that zic2 and most likely zic3 are not expressed in migrating neural crest cells (not shown).
In mice, skeletal abnormalities occur when Zic1, Zic2, or Zic3 expression is compromised. This is presumably due to loss of Zic expression in the sclerotome [20,39,40], a region where expression of the zic1-3 genes was found in chick embryos ( Figure 6D, E, K, P, Q). The ability of Zic proteins to stimulate cell proliferation [10,11,25] may be important for the high rate of proliferation in the dorsomedial lip (DML) of the dermomyotome, where zic1 and zic2 were expressed in chick embryos ( Figure 6D, E, K). In mice, Zic2 and Zic3 are required for proper somite formation, particularly with respect to early establishment of somitic integrity [11]. It is likely that zic3 in chick plays a comparable role due to its similar expression in presomitic mesoderm and epithelial somites. However, the significance of the stripes of zic gene expression in the presomitic mesoderm is not known and further analysis will be needed to understand their role in somite formation.

Limb buds
zic2 was expressed in stage 18/19 wing and hindlimb buds ( Figure 5B, N, R), while zic1, 3, and 4 transcripts were not detected in limb buds. In a previous study [26], we detected zic1 expression in the limb buds of stage 21 embryos. Further studies suggested that zic1 expression in limb buds was transient during stages 20 /21-24 (unpublished). This is consistent with a lack of zic1 expression in the limb buds of stage 26 embryos [28].

Expression levels of zic genes
The zic1-3 genes were expressed at very low levels in chick embryos at early developmental stages (stages 7-11 in particular). Thus, minimal expression levels may be sufficient to mediate Zic function during early neural tube and somite formation. zic1-3 expression levels increased significantly with progressing maturity of the cranial region during stages 13-15. Increased zic1-3 expression is also seen in the trunk region, although this is not apparent until stages 18-19. This delay may be due to the anterior to posterior gradient of maturity in chick embryos, where cranial regions develop ahead of trunk regions. Higher expression levels of zic1-3 during stages 13-15 anteriorly and stage 18-19 in the trunk may be required to help direct cellular decisions relating to proliferation versus differentiation in the maturing cranial and trunk neural tube, and in the maturing somites [25,26,41]. Increased Zic protein levels may impact binding to regulatory sites in Zic target genes and may also affect the stoichiometry with interacting proteins [42][43][44]. This, in turn, may modulate Zic protein function as a transcriptional activator or repressor and influence the affinities of Zic proteins for specific target sites, leading to context-dependent regulation of target gene sets. That transcript levels are dynamic and can vary at different developmental stages is also illustrated by expression of the tbx1 transcription factor gene that is transcribed at very low levels in the cranial region in stage 10 embryos and shows greatly increased expression at stage 12 [45]. Before the significance of the differing zic gene expression levels can be truly understood, additional work is needed to identify the relevant transcriptional targets, to correlate zic mRNA levels with Zic protein levels, to identify Zic functional partners and potential post-translational modifications, to determine the effect of these factors on Zic protein activity, and to determine their effects on the expression of known target genes.

Comparison of zic gene expression in different species
The current study in chick embryos indicates that the zic1-3 genes are expressed during dorsal axial development, both in the neural tube and in somites, while expression of zic4 appears limited to the dorsal brain. In mouse and Xenopus embryos, zic4 is expressed in the brain and in the trunk and its expression pattern is generally thought to resemble that of zic1 at a weaker expression level [34,46,47]. Thus, it was suggested that the adjacent zic1 and zic4 genes are subject to a certain degree of coordinate regulation. Indeed, in zebrafish, the adjacent zic2a and zic5 genes possess common regulatory elements [10]. In chick embryos, the expression patterns of the zic1 and zic4 genes appeared to be quite different, suggesting that coordinate regulation is unlikely for the zic1 and zic4 genes in birds.
In general, the expression patterns of zic genes in the neural tube and somites are relatively similar across species. The similarities of zic1, zic2, and zic3 gene expression in chick to that in other organisms suggests conserved functions for these genes. However, there are also variations in zic gene expression across species. These are particularly evident in the presomitic mesoderm and eyes. Zic2 and Zic3 are expressed in mouse presomitic mesoderm [11], while only zic3 was expressed in chick presomitic mesoderm (this study). Further, Zic1-3 are expressed in mouse eyes [47], zic1 and zic2 in Xenopus eyes [34], and zic2 and zic3 in chick eyes (this study). While these differences may be species-specific, other differences in zic gene expression are less consistent. For example, Warner et al. (2003) report zic1-3 expression in chick periotic mesoderm, while our findings and a study in mouse showed only zic2 expression in periotic mesoderm [46]. Besides species-specific variations, additional explanations for observed differences in zic gene expression between species and within the same species may be based on more technical considerations. These may include the possibility of different degrees of probe specificity for individual zic genes that were used in previous studies (the probes used in this study were specifically designed to preclude cross-hybridization). Another possible explanation for differences in zic gene expression patterns includes the possibility of inaccuracies in the precise staging of embryos within the same species and inaccuracies inherent in comparing equivalent stages across species, which could be very important for transient expression features of particular zic genes. Further, it is possible that zic genes are alternately spliced, which might cause probes to reveal different expression domains. However, to date, alternate splice forms have not been reported for zic genes. Finally, since zic genes may be able to compensate for each other, it is possible that the roles of zic genes may be distributed slightly differently among zic genes in different species, resulting in different expression patterns.

Conclusions
We have followed the expression of the zic1-4 genes during early chick development, resulting in a comprehensive side-by-side study of zic1-4 gene expression throughout neurulation and somitogenesis. We find that the zic1-3 genes are expressed in partially overlapping domains in the dorsal neural tube and in dorsal portions of somites. In addition, the zic2 gene is uniquely expressed along the entire early neural plate and zic3 is uniquely expressed in the surrounding presomitic mesoderm, suggesting that Zic2 and Zic3 specifically regulate developmental genes during initial formation of the neural tube and somites, respectively. Further, zic2 is expressed in the periotic mesoderm and in limb buds and both zic2 and zic3 are expressed in developing eyes, suggesting involvement of these genes in regulating the formation of these tissues. We also show that the zic4 gene is expressed in dorsal regions of the future head, but does not appear to be expressed in the chick hindbrain or trunk. Overall, zic gene expression in chick and other organisms shows significant similarities, indicating that the particular strengths of the chick developmental system will complement current studies of zic genes in other organisms. At the same time, the species-specific differences in zic gene expression that we observe may point to important evolutionary differences, which are of interest in their own right.

Generation of Gene-Specific Antisense RNA in situ Hybridization Probes
Design of a gene-specific antisense RNA probe for the chicken zic1 gene was based on its published sequence [25]. For the chicken zic2, zic3, and zic4 genes, homology to these genes in mouse, human, and Xenopus was used to identify exon regions for zic2, zic3, and zic4 in the chicken genome. The identification of homologous regions for chicken zic2 and zic3 was additionally aided by unpublished partial sequence for these genes, kindly provided by Dr. Sara Ahlgren (Children's Memorial Research Center, Chicago). As expected, we found high homology among the four chicken zic genes in the zinc finger region and in the regions flanking the zinc fingers. Comparing the identified sequences, we carefully selected regions that were sufficiently divergent to result in antisense RNA probes that would not cross-react. At the same time, the chosen regions had to be long enough to produce useful probes. Not all regions chosen gave rise to working probes and multiple regions were tested to obtain functional gene-specific probes. For zic1 and zic2, PCR products were synthesized from stage 18 chick cDNA, which was obtained by reverse transcription of isolated total RNA. These PCR products were TA cloned into pGEM-T (Promega) and transcribed from the resulting plasmids to generate antisense RNA probes. For zic3 and zic4, T7 promoter-containing PCR products were synthesized from stage 18 chick cDNA. The gel-purified PCR products were used as templates for synthesis of antisense RNA probes using T7 polymerase. The primers used to clone pieces of zic1 and zic2 and for synthesis of PCR products for zic3 and zic4 were: zic1 forward: 5'-GCGCTAAAACAAAACAGCGA-3'; zic1 reverse: 5'-CTGTATTTACAAGAGGGAGTGGG-3' (497 bp in 3'UTR) zic2 forward: 5'-CCCTCCTCTCCCTCCTCCT-3'; zic2 reverse: 5'-ACGCTGATTTCCTCACAACC-3' (441 bp in 3'UTR) zic3 forward: 5'-CAGCAAGGACTCCACGAAAAC-3'; zic3 reverse: 5'-CTAATACGACTCACTATAGGCGAC-CCCATCAGATGAGAAT-3' (ca. 730 bp; little 3' coding region and mostly 3'UTR).
The bold bases represent T7 RNA polymerase promoter sequence.
Extensive comparison of the probe sequences that were amplified by these primer sets among each other, comparison of each of these sequences to the regions of all zic genes in the chicken genome and to the complete sequences of the zic2-4 genes in mouse, human and Xenopus (since we do not have complete sequence information for these three genes in chicken) allowed us to conclude that any cross-reactivity of these probes with other zic genes was extremely unlikely. A lack of crossreactivity of our probes was further suggested by the gene expression patterns generated by each probe. The expression patterns produced by each of the four probes showed unique features such that none of the staining patterns could be a subset of the staining pattern generated by another one of the probes.
Finally, our exhaustive sequence comparisons of each of the four chicken zic genes with all zic genes of mouse, human and Xenopus allowed us a very high degree of certainty that we had correctly identified each chicken zic gene. Further, our extensive independent comparisons were in agreement with the annotations provided in the Ensembl chicken genome database.

In situ Hybridization
White Leghorn chicken embryos were staged according to [30]. Whole mount in situ hybridization was performed as in [48] with the modifications described in [26]. NBT/BCIP substrate (Sigma) was used for color detection. The embryos were not post-treated with alcohol, since such treatment proved to remove too much of the fainter stain, which was critical for assessing zic gene expression domains. Hence the red-brown color of the NBT/BCIP reaction product in our images. A pink shadow was digitally softened in Figure 2F and the orange color in Figure 3H was diminished slightly. Both manipulations did not alter the data content of either panel.

Cryosections
Embryos were sectioned following in situ hybridization. Stained embryos were cryoprotected in graded sucrose solutions and embedded in OCT compound. Cryosections of 14 μm thickness were collected on Superfrost Plus (VWR) slides. The sections were rehydrated in 1× PBS and mounted in 1:1 glycerol/PBS.