Zic4, a zinc-finger transcription factor, is expressed in the developing mouse nervous system

Abstract

Zic genes comprise a family of transcription factors, characterized by the presence of a zinc-finger domain containing two cysteines and two histidines (C2-H2). Whereas the embryonic expression patterns of Zic1, 2, 3, and 5 have been described in detail, Zic4 has not yet received close attention. We studied the expression of Zic4 by in situ hybridization during mouse embryogenesis. Zic4 mRNA was first detected at low intensity at embryonic day (E) 9 and, by E10.5, expression was up-regulated in the dorsal midline of the forebrain with a strong, expanded expression domain at the boundary between the diencephalon and telencephalon, the septum, and the lamina terminalis. The choroid plexus of the third ventricle expresses Zic4, as does the dorsal part of the spinal neural tube, excluding the roof plate. The dorsal sclerotome and the dorsomedial lip of the dermomyotome also express Zic4 whereas dorsal root ganglia are negative. At E12.5, Zic4 continues to be expressed in the midline of the forebrain and in the dorsal spinal neural tube. Postnatally, Zic4 is expressed in the granule cells of the postnatal day 2 cerebellum, and in the periventricular thalamus and anterior end of the superior colliculus. We conclude that Zic4 has an expression pattern distinct from, but partly overlapping with, other members of the Zic gene family. Developmental Dynamics 233:1110–1115, 2005. © 2005 Wiley-Liss, Inc.

INTRODUCTION

Zic (zinc-finger protein of the cerebellum) genes were first identified in the mouse, owing to their restricted expression in the granule cell neurons of the cerebellum (Aruga et al., 1994). To date, five members of the Zic gene family have been identified in mouse: Zic1, 2, 3, 4, and 5 (Aruga et al., 1994, 1996a, b). Zic genes are characterized by the presence of a zinc-finger domain (ZFD) that typically contains two cysteines and two histidines (C2-H2) that fold in a tetrahedral configuration around a central zinc ion (Pavletich and Pabo, 1993). Each Zic transcription factor contains five tandem repeats of the C2-H2 ZFD, which is highly homologous to the ZFD sequences of the Gli and Glis transcription factor families (Aruga et al., 1994, 1996a, b; Kim et al., 2002; Zhang et al., 2002).

In the mouse, a role for Zic2, 3, and 5 in neurulation, and for Zic2 in early forebrain development, has been demonstrated through gene targeting and positional cloning studies. Neurulation is the embryonic process responsible for neural tube formation, with impairment of this process, leading to neural tube closure defects (Copp and Bernfield, 1994; Copp et al., 2003). Zic2 expression is down-regulated in the gene-targeted mouse Zic2^Kd (Nagai et al., 2000), whereas a second ENU-induced allele Zic2^Ku carries a point mutation in the ZFD, producing loss of Zic2 function (Elms et al., 2003). Homozygous embryos for either allele exhibit lumbosacral spina bifida and an open cranial neural tube (exencephaly). In humans, heterozygous mutations in ZIC2 cause holoprosencephaly, a congenital malformation that arises from the disturbance of early forebrain midline induction, that precludes normal separation of the telencephalic vesicles (Brown et al., 1998, 2001; Nanni et al., 2000).

Zic3 is also essential for neural tube closure. Positional cloning of the Bent tail mutant, a well-known model of neural tube defects (Copp et al., 2003), identified a deletion that comprises the entire Zic3 locus as well as neighboring genes (Carrel et al., 2000; Klootwijk et al., 2000). Direct targeting of Zic3 produced a null mutant with a phenotype closely similar to the Bent tail phenotype, confirming the critical role of Zic3 (Purandare et al., 2002). Both Bent tail and Zic3 null embryos exhibit exencephaly, together with a low frequency of lumbosacral spina bifida. In addition, homozygous embryos display left–right asymmetry defects, including aberrant positioning of the organs with respect to each other (situs ambiguus) or mirror-image reversal of the structures (situs inversus; Carrel et al., 2000; Klootwijk et al., 2000). Heterozygous mutations of ZIC3 have also been identified in humans, causing X-linked heterotaxy (HTX1; Gebbia et al., 1997), a phenotype reminiscent of homozygous Bent tail and Zic3 knockout embryos.

Zic5 has been disrupted recently by gene targeting in mice, with approximately 15% of the homozygous null embryos exhibiting exencephaly (Inoue et al., 2004). Zic5 null embryos also exhibit a reduction in size of neural crest-derived facial bones such as the mandible, which correlates with a reduction in the number of Sox10- and Cadherin6-positive cells migrating to the first branchial arch (Inoue et al., 2004). Neural crest defects were also detected in Zic2^Ku/Ku embryos, which show delayed emigration and reduced numbers of Foxd3- and Sox10-positive neural crest cells compared with wild-type embryos (Elms et al., 2003).

At postnatal stages, Zic genes have been shown to play a key role in patterning the cerebellum. Thus, Zic1^−/− homozygotes and Zic1^+/−;Zic2^+/Kd compound heterozygotes both exhibit an abnormal foliation pattern with cerebellar hypoplasia and a reduced number of granule cell neurons (Aruga et al., 1998, 2002).

Zic4 is less well-understood than the other mouse Zic genes, as its expression pattern has not been described in detail. Aruga et al. (1996b) detected Zic4 mRNA expression in the adult mouse cerebellum by RNase protection assay. Moreover, recent work has identified a contiguous deletion of the neighboring genes ZIC1 and ZIC4 in some patients with the developmental hindbrain disorder Dandy Walker malformation. Mice doubly heterozygous for targeted mutations of Zic1 and Zic4 exhibited cerebellar defects consistent with these findings in humans (Grinberg et al., 2004). Hence, Zic4 like the other Zic genes seems likely to play an important role in nervous system development. To further investigate this role, we examined the expression profile of Zic4 mRNA during mouse embryonic/fetal development by in situ hybridization on whole embryos and on histological sections.

RESULTS AND DISCUSSION

Expression of Zic4 in the Developing Mouse Embryo

Zic4 expression can first be detected by whole-mount in situ hybridization at embryonic day (E) 9 (Fig. 1A). This onset of expression is later than for other members of the Zic gene family, which are expressed from gastrulation (Nagai et al., 1997; Elms et al., 2003, 2004). At E9 (12-somite stage), Zic4 mRNA transcripts are expressed at low intensity in the dorsal midline of the forebrain neural tube (Fig. 1C) and in the dorsal part of the spinal neural tube (Fig. 1D). By E9.5 (25 somites), Zic4 expression already is noticeably more intense, with continued localization to the dorsal midline of the cranial neural tube and dorsal spinal neural tube (Fig. 1B,E), although the caudal-most region of neural tube is entirely negative for Zic4, as is the posterior neuropore (Fig. 1B,F). E9.5 is the first stage that Zic4 expression can be detected in the dorsomedial lip of the dermomyotome and an adjacent area of the dorsal sclerotome (Fig. 1B,F).

Figure 1

A–F: Localization of Zic4 mRNA at E9 (A,C,D) and E9.5 (B,E,F). A: Zic4 transcripts are first detected at embryonic day (E) 9. A,C,D: Sections at the level of the dotted lines in A show expression confined to the dorsal midline of the forebrain neural tube (arrow in C; anterior to bottom of figure) and the dorsal region of the low hindbrain/upper spinal neural tube (arrow in D). D: The roof plate is negative for Zic4 at this level (arrowhead). B: By E9.5, expression of Zic4 is more intense, with transcripts detected in the telencephalon (arrow) and midbrain neural tube (arrowheads). Zic4 mRNA is first detected in the somites at this stage. E: Sections at the level of dotted lines in B show expression of Zic4 dorsally in the neural tube at the level of the heart (between arrows in E), with diminished expression in the roof plate. F: In the caudal spinal neural tube, posterior to the forelimb bud, Zic4 expression is absent from the entire neural tube, although the dorsomedial lip of the dermomyotome and the dorsal part of the sclerotome are positive at this level (arrowhead; inset is higher magnification of somite expression). dm, dermomyotome; fb, forebrain; mb, midbrain; nt, neural tube. Scale bars = 250 μm in B (applies to A,B), 100 μm in C–F.

By E10.5, Zic4 expression has become further up-regulated (Fig. 2A). Transcripts are restricted to the dorsal midline of the cranial neural tube (Fig. 2A,G), with expanded expression domains at the boundary between diencephalon and telencephalon (including the dorsal region of prosomere 3; Fig. 2B,D) and at the anterior forebrain extremity (Fig. 2B,F). In sections, the choroid plexus of the third ventricle and the septum, including the lamina terminalis, are strongly positive for Zic4 (Fig. 2E,F). The spinal neural tube expresses Zic4 strongly, although transcripts are absent from the roof plate (Fig. 2H,I) and from the entire caudal portion of the neural tube, including the closing posterior neuropore (Fig. 2A and data not shown). The dorsal root ganglia are also negative for Zic4, whereas the somites are positive (Fig. 2H–J).

Figure 2

Localization of Zic4 mRNA at embryonic day (E) 10.5 (30-somite stage). A–C: Views from the side (A), top of the head (B), and dorsal aspect of the trunk (C) show that Zic4 expression has become up-regulated compared with E9 or E9.5. A–C: Particularly intense expression is present dorsally in the cranial and spinal neural tube (red and white arrowhead in B and C, respectively), although the extreme caudal region does not express Zic4 (between arrowheads in A). C: The dorsal sclerotome and dorsomedial dermomyotome are positive for Zic4 (arrow) along most of the body axis. B: A laterally expanded expression domain is seen at the anterior extremity of the brain (i.e., the lamina terminalis and septum, black arrowhead) and the dorsal part of the telencephalon/diencephalon boundary (prethalamus; arrow). D–J: Sections at the level of the dotted lines in (A) show Zic4 transcripts localized at the diencephalon/telencephalon boundary (arrow in D), in the choroid plexus of the third ventricle (arrow in E), and at the septum including the lamina terminalis (between arrows in F). G: At the level of the midbrain/hindbrain boundary (i.e., the isthmus), Zic4 expression is localized to the dorsal neural tube; note that the roof plate is Zic4-positive at this level (arrow; dorsal to bottom). H,I: A transition to a negative roof plate occurs around the level of the hindbrain/spinal cord boundary, so that the neural tube at the level of the heart (H) and, at the mid-trunk level (I), shows Zic4 transcripts restricted to the dorsal part of the neural tube (between arrows in H,I) but with a negative roof plate. A,I,J: The somites are strongly positive, whereas the dorsal root ganglia are negative for Zic4 (A,I), confirmed in a tangential section (J). cp, choroid plexus; d, diencephalon; dm, dermomyotome; drg, dorsal root ganglia; hb, hindbrain; nt, spinal neural tube; rp, roof plate; sc, sclerotome; tv, telencephalic vesicle. Scale bars = 300 μm in A–C, 100 μm in D–J.

At E12.5, the expression of Zic4 mRNA has become more localized, with transcripts restricted to midline regions of the forebrain, in particular the septum, and the medial walls of the telencephalic vesicles (Fig. 3A–D). Zic4 positive regions include the lamina terminalis, choroid plexus, and the medial edge of the hippocampal primordium, the latter corresponding to the dorsal part of the expression domain (arrowheads in Fig. 3D). In the spinal region, Zic4 continues to be expressed in the dorsal neural tube, dorsomedial dermomyotome, and dorsal sclerotome, with exclusion of Zic4 transcripts from the caudal-most neural tube (data not shown).

Figure 3

A–G: Localization of Zic4 mRNA in embryonic day (E) 12.5 embryos (A–D) and in the early postnatal (day 2) brain (E–H). A–D: Sections at the level of the dotted lines in A. B–D: Zic4 mRNA transcripts are largely restricted to midline regions of the E12.5 brain, with expression in the septum (arrows), including the choroid plexus (arrowheads in B). C,D: The Zic4-positive region of the septum includes the medial edge of the hippocampal primordium (arrowheads). E–G: Sagittal sections through the early postnatal brain show Zic4 expression in the cerebellum (arrows in E), anterior end of the superior colliculus, subcommissural organ (black arrowheads in E,G), and periventricular thalamus (red arrowheads in E,G). F: Higher magnification view of the cerebellum shows Zic4 expression in the external granular cell layer (arrowheads). H: A Zic4 sense control probe gave no specific hybridization signal on cerebellum. ba, branchial arch; cb, cerebellum; cc, cerebral cortex; egl, external granule cell layer; igl, internal granule cell layer; sp, anterior end of superior colliculus; pt, periventricular thalamus; tv, telencephalic vesicle. Scale bars = 200 μm in A–D; 1.5 mm in E; 100 μm in G,H.

Expression of Zic1, 2, and 3 has been reported in the postnatal cerebellum (Aruga et al., 1998; Dahmane and Ruiz-i-Altaba, 1999), and at postnatal day 2, we find Zic4 mRNA transcripts specifically in the external granular layer of the cerebellum, which is populated by granule cell neurons (Fig. 3E,F). In contrast, expression is not detected in the cerebellar anlage at E10.5, this region of the hindbrain appearing specifically negative for Zic4 (Figs. 2A,B, 4D). Zic4 is strongly positive in the anterior end of the superior colliculus and in the periventricular thalamus (Fig. 3E,G). A Zic4 sense probe did not give specific hybridization signal (Fig. 3H).

Figure 4

A–P: Comparison of mRNA expression patterns of Zic1 (A,E,I,M), Zic2 (B,F,J,N), Zic3 (C,G,K,O), and Zic4 (D,H,L,P) at E9.5. A–D: Zic1, 2, and 4 transcripts are present along the dorsal midline of the cranial and spinal neural tube. A,B,D: Zic1 (A) and Zic4 (D) are absent from the caudal spinal region, whereas Zic2 is expressed in the neural folds of the posterior neuropore (between red arrowheads in B). C: Zic3 is expressed in the rostral telencephalon, at the telencephalon/diencephalon boundary and in the presomitic mesoderm and notochord of the caudal-most region (between red arrowheads in C). A–D: All four Zic genes are expressed in the developing somites (black arrowheads). A,B: Zic1 and Zic2 are localized to the hindbrain (arrows). E–P: Transverse sections at the level of the heart (E–H) and mid-trunk (I–L, and enlarged in M–P) show overlapping but distinct expression patterns for Zic genes. E–H: Zic1, 2, and 3 are all expressed dorsally in the neural tube, but at the level of the heart, the Zic1 expression domain occupies the dorsal third of the neural tube (between arrows in E), whereas Zic2 and Zic4 are expressed in the dorsal quarter of neural tube (between arrows in F and H, respectively). I,J: At the mid-trunk level, the expression domains of Zic1 and Zic2 are comparable (between arrows). Moreover, Zic4 expression is excluded from the roof plate, whereas this structure expresses Zic1 and Zic2. All Zic genes are expressed in the somite, but the domains of expression differ. E,F,H,I,L,M,N,P: Zic1, 2, and 4 are expressed in the sclerotome at the level of the heart (asterisks in E,F,H) and mid-trunk (asterisks in I,M,N), Zic4 being expressed in only the dorsal most part of the sclerotome at this level (L,P). E,G,I,K,L,M,O,P: Zic1, 3, and 4 are expressed in the dorsomedial portion of the dermomyotome (arrowheads in E,G,I,K,L, arrows in M,O,P). dm, dermomyotome; nt, neural tube; rp, roof plate. Scale bars = 1 mm in A (applies to A–D), 200 μm in E (applies to E–L and represents 350 μm in M–P).

Comparison of Zic4 Expression With Other Members of the Zic Gene Family

We compared the expression pattern of Zic4 with that of Zic1, 2, and 3. E9.5 embryos were processed for whole-mount in situ hybridization using probes for each Zic gene in parallel, under identical conditions.

Nervous system.

Zic1, 2, and 4 transcripts are detected along the dorsal midline of the entire brain and spinal cord. The only exception is the caudal-most neural tube and the unclosed neural folds at the posterior neuropore, where Zic2 and Zic3 are present (Fig. 4B,C), whereas Zic1 and Zic4 are absent (Fig. 4A,D). Zic3 is not expressed in the spinal neural tube, although in the cranial neural tube, expression is restricted to the extreme rostral end of the forebrain and at the diencephalon–telencephalon boundary (Fig. 4C).

Transverse sections through the upper spinal neural tube, at the level of the heart (Fig. 4E–H), reveal a different pattern of mRNA expression for each Zic gene. Although Zic1, 2, and 4 are all expressed in the dorsal neural tube, the expression domain of Zic1 extends approximately one third of the way down the neural tube (Fig. 4E), whereas Zic2 and Zic4 are expressed in only the dorsal quarter of the neural tube (Fig. 4F,H). Both Zic1 and Zic2 are expressed in the roof plate (Fig. 4E,F), but Zic4 is absent from the roof plate, with the Zic4 expression domain being restricted to a dorsal horizontal stripe (Fig. 4H). Different domains of expression of the Zic genes in the neural tube are also observed in the spinal neural tube at the mid-trunk level (Fig. 4I–L). While the Zic1 and Zic2 expression domains persist (Fig. 4I,J), expression of Zic3 and Zic4 is not detected in the neural tube at this axial level (Fig. 4K,L).

Somites.

At E9.5, the developing somites can be subdivided into the ventromedial sclerotome, which forms the vertebrae, and the dorsolateral dermomyotome, which gives rise to muscle and dermal derivatives (Christ et al., 1998). Zic1, 2, and 4 transcripts are all detected in a population of sclerotomal cells adjacent to the neural tube (Fig. 4E,F,H–J,M,N), although the precise extent of sclerotomal expression differs subtly between the three genes. For example, at the level of the heart, Zic4 is expressed more ventrally in the sclerotome than either Zic1 or Zic2 (Fig. 4E,F,H), whereas at the mid-trunk level, Zic1 appears most ventrally expressed and Zic4 is restricted to the most dorsal part of the sclerotome and the dorsomedial dermomyotome (Fig. 4I,L,M,P). The perinotochordal region appears negative for all Zic genes at all axial levels. Zic3 is not expressed in the sclerotome but is strongly expressed in the medial dermomyotome (Fig. 4G,K,O), a cell population that also appears to express Zic1 and Zic4 but not Zic2 (Fig. 4E,F,I,J,L–N,P).

Place of Zic4 in the Zic Gene Subfamily of Transcription Factors

In the present study, we found that Zic4 has a pattern of expression that is distinct from, but partially overlapping with, that of Zic1, 2, and 3. Hence, although Zic genes belong to a single subfamily of transcription factors, with closely similar ZFD regions, their distinct domains of expression could underlie specific developmental roles. This idea is strongly supported by the different phenotypic abnormalities observed in mutants for the various Zic genes. Hence, Zic1^−/− mice display a hypoplastic cerebellum but survive to birth (Aruga et al., 1998), a defect that appears similar to the recently reported Zic4 mutant, although this finding was described only in heterozygous form (Grinberg et al., 2004). In contrast, embryos homozygous for knockdown and ENU-induced mutant alleles of Zic2 exhibit holoprosencephaly, together with cranial and spinal neural tube closure defects and neural crest anomalies (Nagai et al., 2000; Elms et al., 2003). The recently described Zic5 knockout, with its exencephaly and neural crest defects (Inoue et al., 2004), seems reminiscent of Zic2 defects. Hence, of the known Zic genes, Zic1 and Zic4 seem to resemble each other functionally, while Zic2 and Zic5 form a second functionally related group. Zic3 appears distinct from the other genes in exhibiting right–left asymmetry defects in homozygous/hemizygous mutants, although the presence of exencephaly is reminiscent of Zic2 and Zic5 mutants.

The homozygous loss of function phenotype of Zic4 has yet to be described (Grinberg et al., 2004). We predict that it is unlikely to include open neural tube defects, because Zic4 has not begun to be expressed by the stage of cranial neural tube closure and is absent from the posterior neuropore region, where the spinal neural tube closes. This finding is in contrast to Zic2, which is expressed at the sites of cranial and spinal neurulation, and which exhibits closure defects at both ends of the body axis in null mutants (Nagai et al., 1997; Elms et al., 2003). On the other hand, the strong expression of Zic4 in the sclerotome suggests a possible involvement of Zic4 in vertebral development. It will be interesting to determine whether vertebral anomalies are among the features of the Zic4 homozygous knockout mouse.

EXPERIMENTAL PROCEDURES

Mouse Strains and Embryos

CD1 mouse embryos from timed pregnancies were dissected in Dulbecco's modified Eagles Medium (DMEM) containing 10% fetal calf serum, washed with phosphate buffered saline (PBS) and fixed overnight (E8–E10.5) or for 48 hr (E12.5 and postnatal brain) in 4% paraformaldehyde in PBS. Embryos were dehydrated to 100% methanol and stored at −20°C for whole-mount in situ hybridization or dehydrated to 100% ethanol for slide in situ hybridization.

Zic1–4 mRNA Riboprobes and In Situ Hydridization

After isolation of CD1 E9.5 RNA with TRIzol reagent (Gibco BRL), reverse transcriptase was used to generate cDNA. Riboprobes were designed as follows: (1) primers 5′-TCTCTGGGGCTTCAGCTTTT-3′and 5′-GGAAACTAAAGTGTACATACG-3′ amplified a 642-base pair fragment between nucleotides 2244-2886 of Zic1 (GenBank accession no. NM_009573); (2) primers F5′-GGCCAGGCCTTTCTCCCATT-3′ and R 5′-TGTGAAAAGGAAGGCGTCCG-3′ amplified a 354-base pair fragment between nucleotides 1838-2192 of Zic2 (GenBank accession no. D70848); (3) primers 5′-TCTAGATTCCTTACAATGTCAG-3′and 5′-AAGAAGCACTTTAACCATGAG-3′ amplified a 471-base pair fragment between nucleotides 2825-3296 of Zic3 (GenBank accession no. D70849); (4) primers 5′-CACCCCTTGGTGTTGGTGGA-3′ and 5′-GTCATCCCCTAGCCACTTGCA-3′ amplified a 706-base pair fragment between nucleotides 225 and 931 of Zic4 (GenBank accession no. D78174). All polymerase chain reaction products were cloned into the p-GEMt vector (Promega, UK), and identification of the cloned fragments was verified by sequencing.

Zic4 expression pattern in E9 to E10.5 embryos was performed by whole-mount in situ hybridization. Embryos were photographed, embedded, and Vibratome sectioned at 50 μm as described (Henderson et al., 1999). Expression of Zic4 in E12.5 and postnatal brain was performed on 10-μm microtome sections by slide in situ hybridization as described previously (Breitschopf et al., 1992). A control sense probe gave no specific hybridization signal.