The zinc-finger protein CNBP is required for forebrain formation in the mouse

Craniofacial abnormalities affect hundreds of thousands of children each year and result in physical, emotional and economic hardships for affected individuals and their families (Cohen,1993). The isolation of genes underlying mouse mutants that resemble the human syndromes promises to identify many important players in normal and abnormal craniofacial development(Jabs et al., 1993;Dattani et al., 1998). Similarly, gene knockout techniques have proved to be powerful tools for identifying the molecular regulation of many developmental processes. For example, genes such as Lim1 (Lhx1 — Mouse Genome Informatics) (Shawlot and Behringer,1995), Otx2 (Matsuo et al., 1995; Acampora et al.,1995), Smad2 (Waldrip et al., 1998; Nomura and Li,1998), Nodal (Brennan et al., 2001), Foxh1(Hoodless et al., 2001;Yamamoto et al., 2001),Arkadia (Episkopou et al.,2001), Hex (Martinez Barbera et al., 2000), Oto(Zoltewicz et al., 1999),Hesx1 (Dattani et al.,1998) and Dkk1(Mukhopadhyay et al., 2001;Shawlot et al., 1999;Yamamoto et al., 2001) have been demonstrated to be essential for normal head development by targeted gene disruption in mice. However, studies using such knockout techniques are limited to known genes. Through retroviral insertional mutagenesis and genetic screening approaches, we have identified a null-mutant mouse strain with complete forebrain truncation. Heterozygous mutants survive to birth with various craniofacial abnormalities, including the absence of the lower jaw and eyes. The mutated gene has been determined to be that encoding the cellular nucleic acid binding protein (CNBP); this was confirmed by transgenic rescue.

The Cnbp gene encodes a 19 kDa protein containing seven tandem zinc-finger repeats of 14 amino acids(Covey, 1986). The amino acid sequence of CNBP is highly conserved; the sequence of human CNBP is 94.1%identical to that of Xenopus laevis(Flink et al., 1998), 99%identical to that of the chick (van Heumen et al., 1997) and 100% identical to the mouse protein. Despite its discovery over a decade ago, little is known about CNBP function. CNBP was initially postulated to function as a negative-transcription regulator in the coordinate control of cholesterol metabolism(Rajavashisth et al., 1989)but this has not been confirmed(Ayala-Torres et al., 1994;Warden et al., 1994). CNBP was subsequently shown to be a single strand-specific DNA-binding protein that interacts with the sequence CCCTCCCCA (termed the CT element), a segment of DNA that enhances Myc promoter activity(Michelotti et al., 1995). Recently, Konicek et al. reported that CNBP upregulates CSF1 promoter activity in a tissue-specific manner through specific DNA-binding protein interactions(Konicek et al., 1998). Expression studies during embryogenesis, determined that Xenopus CNBP(XCNBP) was located in the ectoderm, endoderm and mesoderm during early development, and in a wide variety of cell types during lateXenopus embryogenesis (Flink et al., 1998). De Dominicis et al. further reported that, inXenopus embryos, CNBP mRNA accumulation during development decreases before the mid-blastula stage and increases again thereafter(De Dominicis et al., 2000). Although the in vivo role and expression pattern of CNBP in mammalian development remains unclear, the extraordinary level of conservation and the expression pattern in Xenopus embryos suggest a potentially important role for CNBP during early embryonic development across different species.

The biological events that control anterior and posterior patterning in vertebrate embryos is one of the most intriguing questions to challenge biologists. Recent evidence from studies in the mouse suggests that anterior patterning precedes gastrulation(Beddington and Robertson,1999). In mouse embryos, an increasing number of genes have been identified that are expressed in the anterior visceral endoderm (AVE) before,or coincident with, the start of gastrulation(Lu et al., 2001). Mutations in a number of transcription factor genes, such as Otx2, Lim1, Hexand Hesx1, that are first expressed in the AVE and, subsequently, in the node and node derivatives, affect anterior development and exhibit anterior truncation. The AVE region is located at the distal tip of the conceptus prior to primitive streak formation, and, subsequently, undergoes a morphogenetic movement toward the proximal/anterior region. These movements have been proposed to be extremely important for the anteroposterior patterning of the embryo (Beddington and Robertson, 1999). For example, in Otx2^-/-embryos at egg cylinder stages, the posterior rotation of epiblast seems to occur normally but the AVE remains distal(Acampora et al., 1998), and the resulting embryos lack midbrain and forebrain. The AVE cells have been suggested to detach from the epithelial sheet and move toward the anterior region (Kimura et al., 2000). It is currently not understood what mechanisms drive either of these processes.

The mouse node structure is homologous to the Spemann's organizer inXenopus. It gives rise to a similar repertoire of embryonic tissues:prechordal mesoderm, notochord and gut endoderm(Beddington, 1981;Beddington, 1994;Lawson et al., 1991). However,it is unable to induce secondary anterior structures even when node precursor cells are transplanted from an early gastrula stage(Tam and Steiner, 1999). AVE appears to repress posterior signals in the epiblast. However, it is unable to pattern the neuroectoderm or cause formation of anterior embryonic structures(Lu et al., 2001;Moon and Kimelman, 1998;Piotrowska and Zernicka-Goetz,2001). The anterior definitive endoderm (ADE) arises from the anterior streak region before node formation and notochord extension, and moves anteriorly to displace the AVE and underlie the prospective neuroectoderm during gastrulation (Lawson and Pedersen, 1987; Tam and Beddington, 1992; Lu et al.,2001). The ADE expresses many of the same genes as the AVE, such as Hex and Cer1, making it an attractive candidate tissue from the anterior streak for patterning the anterior epiblast(Martinez Barbera et al.,2000; Lu et al.,2001). Although, the AVE, ADE and node tissues are essential for head development, the precise function and interaction of these three tissues remain unresolved. We report a new mouse mutant, generated by retroviral insertion into the locus of the Cnbp gene, that displays impaired anterior movement of AVE, lack of both ADE and anterior neuroectoderm (ANE)tissues, and forebrain truncation.

The Cnbp mutant mouse strain, also termed A8, was generated as described (Harbers et al.,1996). Briefly, J1 embryonic stem (ES) cells(Li et al., 1992) were grown on a monolayer of mytomycin C-treated Psi-2 cells producing mp 10 virus(Barklis et al., 1986) in ES cell culture medium containing 8 ug/ml polybrene. ES cell lines with a single-copy proviral genome were injected into BALB/c or C57BL/6J blastocysts,and injected embryos were then transferred into the uteri of pseudopregnant F1(C57BL/6J×CBA) foster mothers as described(Li et al., 1992) to obtain transgenic lines. The Cnbp transgenic line was obtained by breeding a male chimera with a female C57BL/6J mouse. The Cnbp mutation was repeatedly backcrossed (>12 generations) onto the C57BL/6J inbred genetic background to improve phenotypic consistency. Cnbp^+/-inbred mice were intercrossed to produce Cnbp^-/- mice.

Initially, a genomic DNA fragment flanking the 5′ end of the mp 10 provirus was cloned by inverse PCR. This fragment, designated 5′ fA8(see Fig. 1), was subcloned into the Bluescript vector (Stratagene). To obtain λ-phage clones representing the A8 locus from wild-type mice, a 129/Sv mouse genomic library in lambda FIXII (Stratagene) was screened with a radiolabeled probe derived from 5′ fA8. Positive plaques were purified by three rounds of screening and sub-fragments from the λ clones were subcloned according to standard procedures. The physical map of the A8 locus shown inFig. 1 was obtained by both sequence analyses of the plasmid clones and Southern blot analyses of restriction enzyme digested wild-type and mutant genomic DNA.

For genotyping, DNA was isolated from the yolk sac of dissected embryos or from the terminal tail region of adult animals and analyzed by PCR or by Southern blot using 5′ fA8 as a probe(Fig. 1). Embryos were obtained from timed matings; the day of plug detection was counted as day 0.5 of gestation. The presence of a single 8 kb fragment indicates a homozygote genotype (Fig. 1). Oligonucleotide primers P1 (ATAGGACCCGTAGGTTGTCA), P2 (CTCTGAGTGATT-GACTACCC)and P3 (AGTCTCTCCAGAATTGGGTC) were used to give diagnostic amplification products of 500 bp for the wild-type Cnbp allele and 300 bp for the disrupted Cnbp allele (Fig. 1). Data from this study were from C57/B6J inbred mice.

Total cellular RNA was isolated from adult tissues and mouse embryos by the guanidinium isothiocyanate procedure. Extracted RNA was fractionated (15 μg per lane) by electrophoresis in 1% agarose gels containing formaldehyde and then transferred onto nylon membranes (Li et al., 1999). Blots were hybridized for 18-20 hours at 65°C in a standard hybridization solution without formamide(Li et al., 1999).

Embryos and tissues were fixed in 4% paraformaldehyde. Tissues were embedded in paraffin wax and sectioned at 7 μm. Sections were stained with Hematoxylin and Eosin according to standard procedures.

Tissue section immunostaining was performed as described(Li et al., 1999) using anti-CNBP polyclonal anti-peptide antibodies raised against a 20 amino acid peptide from the C terminus of mouse CNBP (CYRCGESGHLARECTIEATA).

Whole-mount in situ hybridization was performed as described(Deng et al., 2001). The full-length mouse Cnbp cDNA was subcloned and linearized withNotI and transcribed with T3-RNA polymerase. The Krox20 cDNA was linearized with BamHI and transcribed with T3-RNA polymerase.En1 and Hnf3b cDNA were linearized and transcribed with T7-RNA polymerase. Other antisense probes used were for: Myc, Mox1(Meox1 — Mouse Genome Informatics), Otx2, Brachyury (T),Hex, Lim1, Six3, Dkk1, Gsc, Hesx1 and Cer1. At least five embryos with the same genetic background were analyzed for each probe.

Cnbp transgenic mice were used to rescue the forebrain truncation in Cnbp mutants. The transgenic vector construct that was used contained 10 kb of the CNBP promoter and 11 kb of the entire Cnbpgene. The vector DNA was linearized and used for pronuclear microinjection to obtain Cnbp transgenic mice. Transgenic (TG) mice were crossed toCnbp^+/- mutants and the resultant progeny(TG/Cnbp^+/-) were backcrossed toCnbp^+/- mice. Litters were examined at E9.5.

BrdU incorporation and TUNEL assays were performed as described(Shen-Li et al., 2000). At least five embryos with the same genetic background were analyzed for each stage.

Transfection study was performed as described previously(He et al., 1998). Wild-type and Cnbp^-/- mutant embryonic fibroblasts (MEF) were isolated from E13.5 embryos in C57B1/6J and 129Sv hybrid background using 0.05% trypsin/EDTA digestion and then maintained in MEM/10% FBS. Wild-type andCnbp^-/- mutant embryonic fibroblasts (MEF) were transfected with a Myc promoter-luciferase reporter plasmid or co-transfected with the luciferase reporter DNA and a mouse Cnbpexpression plasmid (pCMV-CNBP) as described(He et al., 1998). The pCMV-CNBP was constructed by inserting the Cnbp cDNA under the transcriptional control of a CMV promoter in the pCDNA3.1 vector (Invitrogen).Cnbp cDNA was obtained by screening a day 17.5 mouse embryo cDNA library (Clontech). DNA co-transfections were performed in duplicate and repeated at least four times.

We have generated a mutant mouse strain (A8) that exhibits severe forebrain truncation and facial anomalies (Fig. 1A-E). The external morphological deficiencies were variable but limited to the forebrain, eyes and lower jaw. The A8 mutant mice were generated using retroviral insertional mutagenesis(Harbers et al., 1996). The 500 bp DNA fragment flanking the 5′ LTR of the provirus was cloned by inverse PCR and designated 5′ fA8(Fig. 1L). When used as a probe on Southern blots, 5′ fA8 hybridizes to a single 4 kb band in genomic DNA from wild-type mice (Fig. 1M), and an additional 8 kb band, resulting from the insertion of the provirus, in DNA from heterozygous animals, which was used for genotyping mice and embryos (Fig. 1M). Mutant embryos were also genotyped by PCR analysis(Fig. 1N). To ensure genetic uniformity for closely linked loci, heterozygous mice were backcrossed to C57BL/6J inbred strain 12 times. To define the role of Cnbp in mouse development, the offspring from intercrossed heterozygous(Cnbp^+/-) C57BL/6J inbred mice (n=12) were examined at various developmental stages. We found that A8 homozygous embryos had a severe forebrain truncation and died around E10.5(Table 1). No A8 homozygous newborns were ever found. About 40% of heterozygous newborn mutants exhibited multiple defects, including growth retardation and craniofacial defects (e.g. a smaller mandible and complete lack of eyes), and died shortly after birth. The haploinsufficiency suggests that the Cnbp gene must be expressed above a threshold level to ensure normal development. The remaining heterozygous mice either grew normally to adulthood or had mild eye and skeleton defects (data not shown).

In order to identify the gene that is responsible for the mutation,additional genomic sequences flanking 5′ fA8 were isolated by screening a λ-phage library of mouse genomic DNA with the 5′ fA8 probe. Theλ clones were dissected into subfragments and used as probes to hybridize to northern blots containing poly(A)⁺ RNA that was extracted from newborn mice. A 3.0 kb subfragment, designated PA832 (as shown in Fig. 1), hybridized to a 1.65 kb RNA transcript. Sequencing the PA832 fragment indicated that the proviral insertion created a mutation in the previously describedCnbp gene (Rajavashisth et al.,1989). In order to map the proviral integration site relative to the transcriptional unit of the Cnbp gene, the exon-intron junctions of the gene were identified by comparing Cnbp gene sequences withCnbp cDNA sequences. The results summarized schematically inFig. 1 indicate that the provirus was inserted into the first intron. To test whether the proviral insertion affected levels of Cnbp transcription, total RNA was isolated from E9.5 embryos (derived from Cnbp heterozygous mutant parents) and analyzed on northern blots. Compared with their wild-type littermates, the 1.65 kb Cnbp transcripts were significantly reduced in heterozygous embryos and could not be detected in homozygous embryos(Fig. 1O). In order to examine CNBP protein levels, immunostaining of tissue sections was performed using an anti-CNBP polyclonal anti-peptide. We found that, Cnbp was normally expressed in the ANE and ADE of E7.25 embryos(Fig. 1P) but was absent in E7.25 Cnbp^-/- mutant embryos(Fig. 1Q). These results indicate that the Cnbp mutation was a null mutation.

Morphological analysis showed that Cnbp^-/- mutant embryos were distinguishable from normal embryos. At E7.5, the abnormal-looking embryos were smaller than their normal littermates(Fig. 1F,G). A constriction was seen between the embryonic and extra-embryonic regions(Fig. 1G). A similar extra-embryonic/embryonic constriction was also observed in Hnf3bmutants (Ang and Rossant, 1994)and Otx2 mutants (Ang et al.,1996) and to a lesser extent in Lim1 mutants(Shawlot and Behringer, 1995). Truncations were also seen in the anterior neural folds at early somite stages(Fig. 1H,I) and in the anterior regions of E9.5 Cnbp^-/- embryos(Fig. 1J,K). However, the trunk and tail of the mutant embryos were relatively well formed.

To clarify the role of CNBP in mouse head development, we analyzed the expression of Cnbp at pre-gastrulation and gastrulation stages using whole-mount RNA in situ hybridization and tissue section immunostaining. We found that the expression of Cnbp during pregastrulation and gastrulation stages was very dynamic. Cnbp was expressed in visceral endoderm located at the distal tip of the E6.0 embryo in pre-primitive streak stage (Fig. 2A). At E7.0, the early-primitive streak stage, Cnbp was expressed in the AVE(Fig. 2B). At E7.25, the late-primitive streak stage, CNBP protein was localized to the ADE, underlying the future forebrain, and in the overlying ANE, where the forebrain will form(Fig. 1P). At early neural plate stages (E7.5), Cnbp was expressed in the anterior axial mesendoderm, ADE and ANE (Fig. 2C,D). At 8-10 somites (E8.25-8.5), Cnbp expression became progressively restricted to the headfold region(Fig. 2E-G). By E9.25,Cnbp was predominately expressed in the forebrain(Fig. 2H). The expression ofCnbp was also detected in midbrain by E9.5(Fig. 2L). In addition to the head region, Cnbp expression was also detected in limb bud and tail at low level when organogenesis occurs(Fig. 2L). The expression pattern of Cnbp during early mouse development suggests that CNBP plays a role in patterning the anterior central nervous system (CNS), which is consistent with its role in forebrain formation.

To confirm that the forebrain truncation was indeed caused by a disruption of the Cnbp gene instead of by some unknown genetic or epigentic mutations, we generated Cnbp transgenic (TG) mice to test whether theCnbp transgene could rescue the forebrain defect inCnbp^-/- mutants. The transgene contained a 10 kbCnbp promoter and the entire 11 kb Cnbp gene(Fig. 2I). The TG mice were crossed with Cnbp^+/- mutants and the resultant progeny(TG/Cnbp^+/-) were then crossed withCnbp^+/- mice. Litters were examined at E9.5. As previously described, Cnbp^-/- embryos showed forebrain truncations;however, transgene-positive Cnbp^-/-(TG/Cnbp^-/-) embryos were normal(Fig. 2M,N). In situ hybridization revealed an almost identical Cnbp-expression pattern between wild-type and TG/Cnbp^-/- embryos(Fig. 2L,N). The forebrain truncation was rescued in TG/Cnbp^-/- embryos, which confirms that knockout of the Cnbp gene was responsible for the forebrain truncation phenotype.

We examined neuroectoderm formation and anteroposterior patterning inCnbp^-/- embryos between 10-25 somite stages by the expression analysis of a number of CNS and mesoderm marker genes. Bf1mRNA, a marker for telencephalon forebrain, was entirely absent in E8.5 and E9.5 Cnbp^-/- embryos when compared with wild-type littermates (Fig. 3A,B,K,L). Loss of Bf1 expression in mutant embryos at E8.5 suggests loss of the telencephalon. We examined the expression of other forebrain markers, such asHesx1 and Six3, which mark the diencephelon, to determine whether this tissue is also missing in the mutants. Hesx1 andSix3 were not detected in the mutants, indicating that diencephelon is also missing in the mutant embryos (Fig. 3C-F). To determine the anterior truncation level, engrailed1(En1), a marker for posterior midbrain and anterior hindbrain was employed. En1 was expressed in the anterior region of both E9.0 normal and Cnbp mutant embryos(Fig. 3G,H), indicating that anterior hindbrain was not affected by the mutation. However, we could not determine whether midbrain is affected in the mutants from the analysis ofEn1 expression. Another hindbrain marker, Krox20(Egr2 — Mouse Genome Informatics), was detected in rhombomeres 3 and 5 of Cnbp^-/- embryos at E9.0(Fig. 3I,J). Thus, the anterior hindbrain regions are present in homozygous mutants. We then used the mesoderm specific markers, Mox1 and Brachyury (T) to determine whether trunk and tail development was affected. Both genes were expressed normally in homozygous mutants. Mox1 expression was detected in paraxial mesoderm cells of E9.5 mutant embryos. Mox1expression was similar in the trunk regions of homozygous mutants as in their wild-type littermates (Fig. 3M,N). T expression was detected in the notochord and posterior (tail) mesoderm cells of both mutant E9.5 embryos, and in their wild-type littermates (Fig. 3O,P), indicating that notochord development is not affected inCnbp^-/- homozygous embryos(Fig. 3P). Collectively, our expression analysis indicates that the Cnbp mutation results in forebrain truncation but does not affect posterior patterning beyond the midbrain, as development of hindbrain, trunk and tail ofCnbp^-/- embryos was essentially normal.

In order to investigate the onset of the forebrain phenotypes, we analyzed the expression of a number of markers at early developmental stages when morphological abnormalities are not yet visible. We analyzed the expression of AVE markers Hex and Lim1 at pre- and early-streak stages. At E6.0, AVE formation was initiated normally at the distal end ofCnbp^-/- embryos (Fig. 4A,B). The defects were first detected at mid-primitive streak stages (E6.5), when Hex expression did not complete a morphogenetic movement toward the proximal anterior region in Cnbp^-/-mutants when compared with wild-type littermates(Fig. 4C,D). The expression ofLim1 in the anterior of mutants was also detected more distally when compared with that in wild-type embryos(Fig. 4E,F). Interestingly, the posterior expression of Lim1 appeared to be more proximal, and closer to the extra-embryonic/embryonic junction in the mutant embryo compared with its expression pattern in wild-type embryos(Fig. 4E,F). The ectopic expression may be caused by the Cnbp mutation. In a similar case,ectopic expression of Hesx1 was found throughout the ectodermal layer of the distal region of the egg cylinder at E6.75 Cripto mutants(Ding et al., 1998). Others have reported that Otx2-null mutant embryos also failed to execute movement of the AVE from the distal end to proximal region of the embryo and that they lack anterior structures(Perea-Gomez et al., 2001). We conclude from these data, that CNBP is important for the correct localization of the AVE.

To understand further the developmental origins of theCnbp^-/- phenotype, we investigated ADE induction(Lu et al., 2001). The ADE expresses many of the same genes as the AVE, such as Hex andCer1 (Martinez Barbera et al.,2000). The expression of Hex inCnbp^-/- mutant embryos failed to occur at the late streak stage E7.5 (n=6) (Fig. 4K,L), indicating the ADE must be absent inCnbp^-/- mutant embryos. Cer1 is also normally expressed in the ADE at the late streak stage but was also absent inCnbp^-/- mutant embryos (n=5)(Fig. 4M,N), which further confirms the lack of the ADE in Cnbp^-/- mutant embryos. Currently, we do not know why Hex and Cer1 were not expressed in the mutants. The difference in expression of Hex andCer1 in mutants compared with wild-type embryos could be caused by a delay of development. In order to take into account the problem of developmental delay in the mutants, we then analyzed expression of these genes at E7.25 to determine whether the AVE is correctly positioned in the mutant embryos at this stage. Our results showed that at E7.25 stage, Hexand Cer1 were expressed in the AVE and ADE of E7.25 wild-type embryos(Fig. 4G,I). By contrast,Hex mRNA was expressed at the distal tip in the E7.25 mutant embryos and this leads, in Cnbp^-/- E7.25 embryos, to the ectopic confinement of Hex-expressing cells to the region where the node is normally located (Fig. 4H). The expression of Cer1 was also detected more distally when compared with that in wild-type embryos (Fig. 4J). Interestingly, the expression of Hex andCer1 in the ADE is not detected in the mutants at this stage(Fig. 4H,J). Mislocalization of the AVE and absence of the ADE indicate a defect in anterior displacement of the AVE instead of a developmental delay. The critical AVE movement could perhaps be a prerequisite for ADE formation. Its absence inCnbp^-/- embryos supports this hypothesis.

To determine whether the induction of the ANE was affected inCnbp^-/- embryos, we examined the expression of an ANE marker, Otx2, at E7.5. In all Cnbp^-/- mutants examined Otx2 expression was undetectable (n=8)(Fig. 4O,P), which indicates that the cells destined for an anterior neural fate failed to form in the mutant embryos. Our data indicate that CNBP is required for ADE formation and anterior neural fate induction.

Recent transplantation experiments have demonstrated that a mixed graft of cells from the AVE, the anterior epiblast and the anterior streak can induce anterior neural genes (Tam and Steiner,1999). In addition, removal of the ADE, together with prechordal plate and axial node derivatives, at the late gastrula stage results in truncation of the anterior neuroectoderm(Camus et al., 2000),indicating that a reciprocal interaction between these tissues is required for anterior patterning. To examine whether the Cnbp mutation affects the formation of anterior mesendoderm (AME), prechordal mesoderm, node and axial node derivatives, we analyzed the expression of a number of anterior mesendoderm markers, including Lim1, T, Hnf3b, Gsc and Dkk1,at primitive streak and early somite stages. Lim1, T andHnf3b were all expressed in the node of wild-type embryos at E7.5(Fig. 5A,C,E)(Ang et al., 1993;Monaghan et al., 1993).Hnf3b and Lim1 were also expressed in midline cells anterior to the notochord, known as anterior mesendoderm or prechordal mesoderm cells(Fig. 5A,C). In homozygous mutants, all three genes were expressed in the node and in the anterior region, but only a short distance from the node(Fig. 5B,D,F). This is in sharp contrast to wild-type embryos in which labeled head-process cells had migrated much farther anteriorly (Fig. 5B,D,F). In particular, the anterior-most midline expression ofLim1 and Hnf3b in AME cells is missing in the mutants(Fig. 5B,D), indicating that the AME fails to develop. Later, during early somite stages, the absence ofHnf3b signal indicates defects in anterior axial mesoderm cells and the rostral portion of the neural tube(Fig. 5G,H). The rostral expression of Hnf3b in the mutant embryo appears to be limited to the prospective hindbrain (Fig. 5H). This suggests that the midbrain development may also be affected in the mutant embryos. However, the potential defect in midbrain should be only a partial truncation based on the above morphological analysis(Fig. 1). The reducedLim1 and Hnf3b expression in the anterior embryo suggests a defect in the AME. To analyze this structure further, we used prechordal plate markers Gsc and Dkk1 to assess prechordal plate development.Gsc and Dkk1 were not expressed in E7.75 and E8.0 mutant embryos, indicating a defect in prechordal plate development(Fig. 5I-L). Although loss ofCnbp expression leads to defects in forebrain and midbrain development, the more posterior CNS is normal.

We next investigated the cellular and molecular basis of the forebrain truncation defect in Cnbp^-/- embryos. The forebrain truncation may potentially result from defects in anterior neural cell differentiation, excess cell death, decreased cell proliferation or a combination of these processes in the developing forebrain region. Morphological and histological analysis indicated that the AME and ANE tissues of E7.5 Cnbp^-/- embryos were missing(Fig. 6A,B). Sagittal sections of E8.5 Cnbp^-/- embryos revealed defects in headfold formation and prechordal mesoderm formation(Fig. 6C,D). The rest of the body axis appeared normal. To compare the proliferative and apoptotic profiles in Cnbp^-/- and wild-type littermates, BrdU incorporation and TUNEL assays were performed on sections of E7.5 and E8.5 embryos. Wild-type E7.5 and E8.5 embryos exhibit many BrdU-positive nuclei throughout the embryonic structures (Fig. 6E,G). By contrast, the mutants have fewer BrdU-positive nuclei in the ANE region (Fig. 6F,H). However, there is no significant difference in the number of BrdU-positive nuclei between the trunk region of wild-type and mutant embryos. The ratio of proliferating cells (BrdU-positive nuclei) to total cell number in the anterior of E7.5 embryos was calculated to be 84% for three wild-type embryos compared with 28% for three Cnbp^-/- mutant embryos(Fig. 6E,F,Q). As cells have been estimated to have a 10-12 hour division cycle during this period, a 10-20% decline in the proportion of S-phase cells during early post-implantation could result in a 25% decline in embryo size over a period of 1 day. The lack of cell proliferation may result in the observed reduction in size of the headfolds at E8.5 (Fig. 6H). TUNEL assays showed minimal apoptosis in normal and mutant E7.5 and 8.5 embryos (Fig. 6I-L), which suggests that programmed cell death does not contribute significantly, if at all, to the null phenotype. These findings indicated that the Cnbp mutation leads to a dramatic reduction in cell proliferation in the AME and ANE tissues, and headfold. To address further whether the impaired anterior movement of the AVE observed inCnbp mutant embryos is related to defects in cell proliferation in AVE, we performed BrdU incorporation assays on E6.0 wild-type and mutant littermates. BrdU-positive nuclei were rarely seen in the prospective anterior region of the AVE in E6.0 mutant embryos(Fig. 6P). By contrast, the greatest density of BrdU-positive nuclei was observed in the anterior region of the AVE in normal E6.0 embryos (Fig. 6O). Our results suggest that reduced cell proliferation in anterior regions of Cnbp mutant embryos might account for defects in formation of the AVE, ADE, AME and ANE.

CNBP was shown to regulate the CT element of the human MYCprotooncogene through its binding to the element found in the MYCpromoter (Michelotti et al.,1995). In addition to regulating cell proliferation and apoptosis,Myc can also promote differentiation of stem cells into transit-amplifying cells specific for the sebaceous and interfollicular epidermal lineages(Arnold and Watt, 2001;Gandarillas and Watt, 1997),and the Myc^-/- mutant has defects in development of anterior structures (Davis et al.,1993; Gandarillas and Watt,1997). These reports lead us to hypothesize that Myc is a downstream target gene of Cnbp during forebrain development, which may promote cell proliferation and differentiation in forebrain induction. We therefore examined the possible involvement of Myc in forebrain neuroectoderm induction and specification. We observed that Myc is expressed in anterior neuroectoderm at E7.25, and that the expression pattern of Myc in E8.5 and E9.5 mouse embryos was similar to that ofCnbp during forebrain development(Fig. 2C,E,H;Fig. 7A,C,E). Notably,expression of Myc in the anterior neuroectoderm and the headfold region of E7.25 and E8.5 Cnbp^-/- mutant embryos was absent, whereas the expression of Myc in the allantois was normal(Fig. 7B,D). However, the loss of the anterior tissues by E8.5 and E9.5 could equally be the mechanism that results in reduced Myc expression. The regions where Mycexpression was downregulated also showed reduced BrdU labeling, indicating that CNBP might regulate anterior cell proliferation through Myc.

To test whether CNBP regulates Myc expression at the transcription level, we transfected wild-type and Cnbp^-/- mutant embryonic fibroblasts (MEF) with a Myc promoter-luciferase reporter plasmid (He et al., 1998). A lower level of luciferase activity was observed in Cnbp^-/-MEF than in wild-type cells (Fig. 7G). Co-transfection of Cnbp^-/- MEF with the luciferase reporter DNA and a mouse Cnbp-expression plasmid(CMV-CNBP) elevated Myc expression to a level higher than that seen in Cnbp^+/+ cells (Fig. 7G). Therefore, we conclude that Cnbp expression enhancesMyc-promoter activity. Although the mechanism by which Cnbpexpression enhances Myc promoter activity during anterior patterning remains to be elucidated, it is plausible that CNBP is one of the necessary transcription factors that bind to the Myc promoter to regulate its transcription.

Our work is the first to demonstrate the role of CNBP in rostral head formation during mouse embryonic development by generating Cnbpmutant mice, performing in vivo functional studies and transgenic mouse rescue, and characterizing the potential mechanism by which CNBP induces and specifies the forebrain through Myc expression and regulation of cell proliferation. Our results demonstrate that the Cnbp mutant mouse provides a valuable model for insight into anterior patterning related to AVE localization, ADE formation, neuralization of anterior ectoderm and forebrain induction.

Our study has shown that ablation of Cnbp function in the mouse results in severe truncation of the forebrain. This finding provides direct genetic evidence that Cnbp, a zinc-finger protein, plays an essential and novel role in mouse forebrain development. De Robertis and colleagues have recently shown that mouse embryos carrying null mutations in the genes encoding BMP antagonists Noggin and Chordin fail to maintain a functional AVE and display forebrain defects (Bachiller et al., 2000). In addition, Mukhopadhyay and colleagues have recently shown that mouse embryos carrying null mutations in the genes encoding Dkk1 display forebrain defects (Mukhopadhyay et al., 2001). Molecular marker analysis showed that expression ofBf1, Hesx1 and Six3 is completely absent; however,expression of En1 is detected in Dkk1-mutant embryos. The identical expression pattern of the marker genes in both Dkk1- andCnbp-mutant embryos indicates that both Dkk1- andCnbp-mutant embryos show a similar forebrain phenotype. AlthoughCnbp is predominately expressed in the forebrain, Cnbpexpression is also detected in midbrain region of E9.5 embryos. Moreover, E7.5Cnbp-mutant embryos show a complete lack of Otx2 expression,and the rostral level of Hnf3b expression in the mutant embryos appears to be limited to the hindbrain region, indicating a defect in midbrain tissues in the Cnbp mutants.

Cnbp expression in the early embryo is first noted in cells corresponding to a region of the early gastrulating embryo (at E6.0) where the AVE abuts the epiblast. However, no morphological defect can be detected in the Cnbp^-/- embryos prior to the early-streak stages. The defects were first detected at mid-primitive-streak stages (E6.5), whenHex expression did not complete a morphogenetic movement toward the proximal anterior region in Cnbp^-/- mutants when compared with wild-type littermates. The more distal Hex expression could be caused by a delay in development of the mutants. However, our results could not rule out the possibility of a delay in the development of mutants, based on the fact that: (1) Hex expression in E6.0 mutants is normal, which indicates the delay did not happen at this stage of development; (2) at E7.25,expression of Hex and Cer1 was incorrectly positioned at the distal end in mutant embryos, which indicates defects in corresponding tissues, whereas we would expect that the AVE would persist and fully elongate, and that the ADE would be induced if there was a delay in development; and (3) forebrain truncation in the E9.5 and newborn mutants is consistent with defects in the anterior tissue, whereas the trunk develops normally.

It is notable that Otx2 expression is absent in Cnbpmutants at E7.5. As Otx2-null mutant embryos both failed to execute the movement of the AVE from the distal end to proximal region of the embryo(Perea-Gomez et al., 2001) and lack anterior structures (Ang et al.,1996), we suspect Otx2 may act downstream of CNBP. However, inOtx2 mutants the brain truncation was extended to anterior hindbrain as the expression of En1 marker gene was not detected inOtx2-mutant embryos (Ang et al.,1996). Thus, the head defect phenotype in Otx2-mutant embryos is more severe than that in Cnbp-mutant embryos. The difference between the two mutations might be explained by residual Otx2 protein or reduced Otx2 expression in Cnbp mutants that was not detected by our in situ methods. An alternative possibility is that CNBP might only regulate Otx2 expression in certain tissues and at specific stages. To address this question, mutant embryos at early stages will be analyzed in further studies. Nevertheless, the absence of Otx2expression in the prospective ANE cells of late-streak mutant embryos at E7.5 suggests that CNBP function is required for specification of the ANE during forebrain development. Forebrain patterning in the mouse is initiated by the inductive activity of the AVE and, subsequently, requires the function of the node-derived ADE (Ang et al.,1994; Shawlot et al.,1999; Tam and Steiner,1999; Thomas and Beddington,1996). The severe anterior phenotype ofCnbp^-/- embryos suggests that CNBP is a key factor in the head developmental process. However, it is not clear from this analysis whether CNBP is required in the AVE and/or the ADE for forebrain development. The generation of chimeric embryos composed of extra-embryonic and embryonic tissues of different genotypes would resolve this issue in future studies.

An abnormal constriction at the extra-embryonic/embryonic boundary is observed in Cnbp^-/- mutants. The constriction was also reported in Otx2, Hnf3b and Lim1 mutants. However, the cause of the constriction remains unknown. Our cell proliferation data identify a substantial reduction in the cell proliferation of the AME and ANE, which is also associated with the loss of Myc expression in a tissue-specific manner where the constriction is observed. As no difference in apoptosis was evident between Cnbp^-/- and wild-type embryos, we conclude that the constriction arises as a result of reduced proliferation of the AVE and ANE during expansion of the ANE. The fact that CNBP upregulates CT elements in the Myc promoter and regulates cell proliferation highlights potential links between CNBP and Myc. InCnbp^-/- embryos, CNBP appears to regulate proliferation through Myc. Myc is an important regulator of cell proliferation; however,others have recently shown that Myc is also involved in differentiation(Arnold and Watt, 2001;Gandarillas and Watt, 1997). Myc may be involved in ANE tissue specification. In homozygous Cnbpmutants, the lack of Myc may hinder neuralization in the anterior epiblast and, thus, further exacerbate the forebrain defect. Our data suggest a forebrain induction mechanism by which CNBP induces the expression ofMyc, which in turn stimulates cell proliferation and differentiation of the anterior epiblast and neuroectoderm cells during forebrain induction and specification. Although we propose that CNBP regulates forebrain formation through the Myc pathway, we could not rule out involvement of other CNBP target genes that have not yet been characterized. Interestingly, someMyc-null mutant embryos die at E10.5 with anterior neural fold truncation (Davis et al.,1993) whereas other Myc mutant embryos do not show obvious forebrain defects. One possible explanation is that CNBP targets a group of genes, including Myc, to regulate forebrain development. Another explanation is that an unknown factor may compensate for Mycloss in C57B1/6J and 129Sv hybrid or inbred 129Sv background(Davis et al., 1993). The role of Myc in forebrain formation remains to be investigated further.

We find that AVE, ADE and ANE defects in Cnbp^-/- mice result in forebrain truncation initiated from early gastrulation stages. Other genes, such as Lim1, Otx2, Nodal, Smad2, Foxh1, Arkadia, Hex, Oto, Dkk1,Hesx1, Nog and Chrd, are also essential for murine head development (Episkopou et al.,2001; Hoodless et al.,2001; Shawlot et al.,1999; Yamamoto et al.,2001). However, the brain defects are considerably different among these mutants. Embryos homozygous for mutations in Lim1, Otx2, Foxh1or Arkadia exhibit truncations of the forebrain, midbrain and rostral hindbrain. By comparison, forebrain truncation in Hex^-/-,Oto^-/- and Hesx1^-/- embryos is relatively mild (Zoltewicz et al.,1999; Martinez Barbera et al.,2000). The defects observed in Cnbp^-/- embryos are clearly different from other mutants as Cnbp^-/-mutants showed complete forebrain truncation. The developmental origins of the defects are also considerably different among these mutants. The developmental defects in Otx2 mutants originate from an inability of the AVE to complete its anteriorward movement and a failure to form the node, prechordal mesoderm, notochord and ADE. Foxh1 and Arkadia mutants have normal AVE but impaired ADE, node and notochord. Hex^-/-,Oto^-/- and Hesx1^-/- mutants display absence or early regression of the ADE and normal AVE, node and notochord. Compared with other mutants that have brain defects, the developmental origin of the forebrain defects in Cnbp^-/- embryos is clearly unique. Cnbp mutants exhibit impaired AVE and ADE, with normal development of the node and notochord. The unique forebrain phenotype and developmental origin of the defects in Cnbp mutants indicate that theCnbp mutation may affect a different genetic pathway when compared with any known mutation resulting in forebrain defects. Therefore,Cnbp^-/- embryos provide a unique and valuable mouse model for studying forebrain formation.

We are grateful to Ms Justine Dobeck for her excellent histology assistance and Dr S. P. Oh for his assistance with genomic gene cloning. We thank Dr Margaret A. Thompson for her assistance with transgenic mouse rescue work. We thank Dr Ryoji Moroi for his assistance with the manuscript. We thank Dr Kenneth W. Kinzler for the Myc promoter-luciferase plasmid, Dr Sarah Millar for the Dkk1 probe and Dr Guillermo Oliver for theSix3 probe. This work was supported by NIH grant AR44741 (Y.-P. L.)and AR-48133-01 (Y.-P. L.).