Ssdp1 regulates head morphogenesis of mouse embryos by activating the Lim1-Ldb1 complex

Members of the LIM-homeodomain (LIM-HD) family of transcription factors play important roles in various aspects of both vertebrate and invertebrate development (Bach, 2000; Bachy et al., 2002; Hobert and Westphal, 2000). One such factor, Lim1 (Lhx1), is required during early stages of vertebrate development to establish anterior patterning(Shawlot and Behringer, 1995). Lim1-deficient mouse embryos fail to establish the head organizer,the signaling center that initiates and promotes the head differentiation program in naïve ectoderm. In the mouse embryo, head organizer activity resides in two distinct regions, the anterior visceral endoderm (AVE) and the anterior primitive streak. The AVE first confers anterior character to the overlying anterior epiblast at egg cylinder stage(Beddington and Robertson,1999; Tam and Steiner,1999). Following the onset of gastrulation, the anterior definitive endoderm (ADE) and the prechordal plate, which arise from the anterior primitive streak and constitute the anterior region of the axial mesendoderm (AME), intercalate into the outer visceral endoderm layer and displace the AVE, inducing and maintaining anterior neural character in the overlying epiblast. Lim1 mutant embryos fail to specify the AVE and the anterior AME, resulting in the loss of head structures anterior to rhombomere 3 (Shawlot and Behringer,1995; Shawlot et al.,1999). In addition, Lim1 mutants show defects in body axis extension because of impaired cell movements during gastrulation(Hukriede et al., 2003; Tam et al., 2004).

LIM-HD transcription factors contain two LIM domains that regulate their activity. LIM domain-binding proteins, which include Ldb1 (NLI/CLIM2) and Ldb2(CLIM1) in vertebrates and Chip (dLDB) in Drosophila, bind to LIM-HD transcription factors (Agulnick et al.,1996; Jurata et al.,1996) and promote the formation of(LIM-HD)₂-(Ldb1)₂ tetramers through their dimerization domain (Jurata et al., 1998). Formation of this tetrameric complex is required for the transcriptional activator function of LIM-HD proteins in vivo(Hiratani et al., 2003; Milan and Cohen, 1999; Thaler et al., 2002; van Meyel et al., 1999). In addition, it has recently been shown that the single-stranded DNA-binding protein (Ssdp) proteins (Bayarsaihan et al., 1998) bind to Ldb1/Chip(Chen et al., 2002; van Meyel et al., 2003). In Drosophila, a functional complex of Ssdp, Chip and the LIM-HD protein Apterous plays an important role in wing development(Chen et al., 2002; van Meyel et al., 2003). Although a similar mechanism was suggested for vertebrate Ssdp proteins(Chen et al., 2002), their roles in the developmental regulation of LIM-HD proteins in vivo are unknown.

In this study, we generated a novel mouse mutant, headshrinker(hsk), by transgene insertion. At birth, hsk mutants lacked head structures anterior to the ear, a phenotype that is reminiscent of head organizer defects. Mapping of the transgene insertion revealed an intronic disruption of the Ssdp1 locus. Transgenic expression of exogenous Ssdp1 was able to rescue the hsk mutants, indicating that disruption of Ssdp1 is responsible for the hsk phenotype. The head organizer developed normally in Ssdp1^hsk/hskmutants at early stages, but later development was compromised. In addition to exhibiting severe anterior truncations, Ssdp1 mutants also showed reduced cell proliferation, with half of the mutants exhibiting smaller body size than their wild-type littermates. Furthermore, we show that Ssdp1 contains an activation domain and is able to enhance transcriptional activation by the Lim1-Ldb1 complex in a dose-dependent manner in transfected cells. Finally, we show that Ssdp1 interacts genetically with Lim1 and Ldb1 in both head development and body growth. Together, these data indicate that Ssdp1 functions as an activator component of a Lim1-Ldb1-Ssdp1 complex that plays an essential role in head organizer development and body growth in mouse embryos.

Transgenic mice were produced by pronuclear injection of transgene DNA into C57BL/6×C3H/He F2 fertilized eggs(Hogan et al., 1994). The headshrinker mutant line was generated in the process of producing transgenic mouse lines harboring a human copper-zinc superoxide dismutase 1(SOD1) transgene, which consists of an 11 kb genomic DNA fragment with a mutant form of the SOD1 gene that has a 2 bp deletion(Pramatarova et al., 1994). The headshrinker line was maintained by crosses with C57BL/6×C3H/He F1 mice. For the rescue experiments, the Ssdp1transgene was created by cloning the full-length mouse Ssdp1 cDNA(Okazaki et al., 2002) into the pCAGGS vector (Niwa et al.,1991). For genetic interaction studies, Lim1^+/- mutants(Shawlot and Behringer, 1995)and Ldb1^+/- mutants(Mukhopadhyay et al., 2003)were crossed with Ssdp1^+/hsk mutants.

The sequences flanking the transgene insertion site were amplified using the LA-PCR in vitro Cloning Kit (Takara). Specific primers were designed for the 5′ end of the SOD1 transgene: S1(5′-GTCATTAGTTATGACTGAGTTTGGCCACAGCG-3′), S2(5′-TGAGGGTATAGAAAGACGCTACACCTCAATCC-3′).

Two BamHI genomic DNA fragments (1.2 kb and 1.6 kb) were used for sequence analysis.

Embryos were genotyped by PCR using primer P2(5′-GATGAAATGCTGGACTGAGC-3′) and primer P3(5′-TGCTTGGTTACCGTGTTAGC-3′) for the wild-type allele (400 bp),and primer P1 (5′-GTTACTCAGCAATTGGGACGCC-3′) and P2 for the mutant allele (510 bp). The positions of the primers are indicated in Fig. 2E. DNA samples of adults and embryos older than E7.5 were prepared from ear punches and yolk sacs,respectively. PCR products were amplified for 30 cycles of 95°C for 30 seconds, 60°C for 1 minute and 72°C for 1 minute. The genotypes of embryos younger than E7.25 were determined following in situ hybridization analysis. The genotypes of Lim1 and Ldb1 mutants were determined as previously described(Mukhopadhyay et al., 2003; Shawlot and Behringer,1995).

The cartilage and bone of P0 neonates were stained with Alcian Blue and Alizarin Red, respectively, as previously described(Hogan et al., 1994).

Preparation of chromosome spreads and FISH were performed as previously described (Matsuda and Chapman,1995). The 11 kb SOD1 transgene was labeled by nick translation with biotin 16-dUTP (Roche). The hybridized probes were reacted with a goat anti-biotin antibody (Vector Laboratories) and then stained with fluorescein-conjugated donkey anti-goat IgG (Nordic Immunology).

Northern blot analysis of E9.5 total RNA isolated by an RNeasy Mini Kit(Qiagen) was performed using ULTRAhyb™ (Ambion) following the manufacturer's instructions. Radioactive signal was measured with a BAS2500 bio-imaging analyzer (Fuji Film) and normalized using β-actin as a reference.

Total RNA (1 μg) was used for cDNA synthesis with Ready-To-Go You Prime First Strand Beads (Amersham), followed by quantitative PCR using SYBR Premix Ex Tag (Takara) and ABI PRISM 7900HT (Perkin Elmer). The PCR primers used for Ssdp1 were 5′-atggagccccaccacatgaatg-3′ and 5′-ctggaaggagtggaggaagttc-3′; primers for β-actin were 5′-tgtatgcctctggtcgtaccacag-3′ and 5′-gatgtcacgcacgatttccctctc-3′. The signals were normalized usingβ-actin as a reference.

Mouse embryos were staged by morphology(Downs and Davies, 1993). In situ hybridization was performed according to standard procedures(Henrique et al., 1995; Sasaki and Hogan, 1994; Wilkinson, 1992). The following probes were used: Foxg1(Hatini et al., 1994), Cerl (Belo et al.,1997), Dkk1 (Glinka et al., 1998), En2(Joyner and Martin, 1987), Fgf8 (Crossley and Martin,1995), Foxa2 (Sasaki and Hogan, 1994), Foxd4(Kaestner et al., 1995), Gsc (Blum et al.,1992), Hhex (Thomas et al., 1998), Krox20(Nieto et al., 1991), Lefty1 (Meno et al.,1997), Lim1 (Barnes et al., 1994), Otx2 (Ang et al., 1994), Pax6(Stoykova and Gruss, 1994), Six3 (Oliver et al.,1995), Uncx4.1(Mansouri et al., 1997) and Wnt1 (Parr et al.,1993). For Ssdp1, a 343 bp 3′UTR fragment of the Ssdp1 cDNA (nucleotides 1534-1876, Accession Number AK011853) was used.

The transfection of P19 cells with plasmid DNA was performed using the FuGENE6 Transfection Reagent (Roche). Briefly, cells were transfected with 0.4μg reporter plasmid, 0.4 μg effector plasmids and 0.1 μg pCS2-c-β-gal (Turner and Weintraub,1994). pG4-tk-Luc (Sasaki et al., 1999) and –492gsc/Luc(Mochizuki et al., 2000) were used as reporter constructs. pCMV-Gal4-mSsdp1, pCS2-Xlim1, pCS2-Xldb1(Mochizuki et al., 2000) and pCS2-mSsdp1 were used as effectors. mSsdp1 constructs were created by cloning the coding region of mouse Ssdp1 cDNA into the appropriate vectors. The amount of effector plasmid used was adjusted to 0.4 μg by the addition of pCS2.

Detection of apoptotic cells by the TUNEL method and of proliferating cells using anti-mouse PCNA mouse monoclonal antibody (Santa Cruz) was performed as previously described (Kiso et al.,2001).

While constructing a mouse model of human familial amyotrophic lateral sclerosis, we generated several transgenic mouse lines. All the transgenic mice were normal up to 1 year of age, indicating that the expression of the transgene did not disturb embryonic development. However, when we intercrossed these transgenic mice, one line produced neonates that lacked head structures anterior to the ear (Fig. 1B). Skeletal staining of the these mutants revealed that skull derivatives posterior to and including the supraoccipital bone developed normally, but the bone anterior to this point was abnormal and unidentifiable(Fig. 1D,D′). Based on the severe head truncation at birth and the developmental defects that led to it, we named this mutant headshrinker (hsk). In addition to exhibiting truncation of anterior skull bones, some homozygous mutants also showed mild skeletal defects in other body parts, including anteroposterior dislocation of the left and right halves of the sternum(Fig. 1F, asterisks),asymmetrical attachment of ribs to the sternum(Fig. 1F, arrowheads),reduction of sternebra number, bifurcation and/or fusion of ribs(Fig. 1H, arrowheads),absence/reduction or lateral splitting of the vertebral body, and a shortened tail (Fig. 1H; data not shown). The gross morphology of the internal organs of the hsk homozygous neonates was normal (data not shown).

hsk mutant embryos were morphologically indistinguishable from wild-type embryos until E7.75 (Fig. 1M). However, although in wild-type embryos the neural plate folded dorsally to form brain vesicles at E8.5, in mutant embryos the head neural plate remained flat (arrow in Fig. 1N). After E9.5, the absence of anterior head structures in hsk mutants became evident (Fig. 1O,P). While Mendelian ratios were maintained up to E10.5, the survival ratio for homozygotes declined from E11.5 onwards. The recovery of homozygous animals at birth was 3.4% of total progeny. Approximately half of the mutant embryos were significantly smaller at E9.0 than their wild-type littermates (Fig. 1Q). In addition, the neural tube was thin and kinked, and somites were small and irregular or not formed in some cases (data not shown). Because the head defects were completely penetrant, we focused most of our analysis on this aspect of the mutant phenotype, and selected embryos whose defects were restricted to the head region after E8.5.

To identify the gene responsible for the defects observed in hskmutants, we mapped the transgene insertion site. Fluorescence in situ hybridization (FISH) localized the transgene to a single site in the C5-C6 region of chromosome (Chr) 4 (Fig. 2A-D). Cloning the genomic DNA fragments flanking the transgene insertion site localized the affected region to a 105.85 Mb segment of Chr 4,consistent with the chromosomal FISH results. Further analysis showed that four copies of the transgene were inserted into intron 4 of the Ssdp1locus, accompanying a 9 bp deletion (Fig. 2E).

As Ssdp1 appeared to be a good candidate for the gene disrupted in hsk mutants, we analyzed the expression pattern of Ssdp1 in wild-type embryos. Ssdp1 was widely expressed in wild-type embryos from E6.5 to E9.0 (Fig. 2F-M). At E6.5, Ssdp1 mRNA was localized to the basal margin of epiblast cells but was absent from the visceral endoderm, including the early head organizer tissue, the AVE (Fig. 2G,G′). However, Ssdp1 expression was detected in the later head organizer tissues, including the ADE(Fig. 2I,I′), prechordal plate and foregut epithelium (Fig. 2K,K′). By E9.0, Ssdp1 was ubiquitously expressed(Fig. 2M). Whole-mount in situ hybridization of mutant embryos at late head fold (LHF) stage to E8.0 revealed that the expression of Ssdp1 in hsk homozygotes was clearly reduced compared with wild type, although the pattern of expression was unaltered (Fig. 2N). Northern blot analysis of RNA extracted from E9.5 embryos detected two Ssdp1-positive bands at 2.3 kb and 3.8 kb(Fig. 2O), similar to results observed using rat Ssdp1 RNA(Raval-Fernandes et al.,1999). The Ssdp1 RNA species detected in hskhomozygotes were of normal size, but their expression levels were reduced to∼32% of the level seen in wild-type embryos(Fig. 2O,P). No significant differences were detected in the expression levels of the genes surrounding Ssdp1 (Fig. 2E,P). Thus, the transgene insertion results in the specific downregulation of Ssdp1 expression.

To test whether the loss of anterior head structures was caused by reduced expression of Ssdp1, we designed a transgene in which Ssdp1is under the control of a ubiquitous promoter(Fig. 3A) and established two transgenic mouse lines (TG). These transgenic mice were crossed with Ssdp1^+/hsk mutants, and the resultant progeny(TG/Ssdp1^+/hsk) were crossed with Ssdp1^+/hsk mice. In one transgenic line, number 37,TG37/Ssdp1^hsk/hsk embryos did not show any of the head defects or growth retardation observed in Ssdp1^hsk/hskmutants at E9.5 (Fig. 3B,C),and TG37/Ssdp1^hsk/hsk mice developed into healthy and fertile adults (Fig. 3E). In the other line, number 141, which expressed the Ssdp1 at a lower level than number 37 (Fig. 3F),only partial rescue was observed at E9.5(Fig. 3D), and no live-born TG141/Ssdp1^hsk/hsk mice were obtained. Together with the mapping and expression data, these results suggest that the hskmutant phenotype results from reduced expression of Ssdp1.

The severe defects in anterior head development of Ssdp1 mutants suggest that this phenotype is associated with defects in the head induction process. To test this possibility, we examined the development of the head organizer in Ssdp1 mutants. Proper head development first requires the anterior patterning activity of the AVE, which expresses Lefty1, Lim1,Cerl, and Hhex (Fig. 4A,C; data not shown) at early streak (ES, E6.5) and mid-streak(MS, E6.75) stages. Subsequently, the AVE is displaced by the ADE, which is derived from the anterior primitive streak and is marked by the expression of Hhex and Cerl at early bud (EB) stage (E7.25)(Fig. 4E,G). Expression of these genes was unaltered in Ssdp1 mutants(Fig. 4B,D,F,H; data not shown), suggesting that the AVE and ADE developed normally.

Slightly later, at early head fold (EHF) stage (E7.5), another head organizer tissue, prechordal plate, begins to form at the anterior portion of the axial mesoendoderm. Cerl is expressed in the prechordal plate(Fig. 4I, arrow), as well as the anterior mesoderm and anterior definitive endoderm. In Ssdp1mutants, Cerl expression in the prechordal plate was either strongly reduced or absent (Fig. 4J,arrow), while expression in other tissues was not significantly altered. Similarly, Dkk1, which is normally expressed in the anterior-most domain of the prechordal plate and in the anterior margin of the head mesoderm, was specifically absent from the prechordal plate of late head fold(LHF) stage mutants (E7.75, Fig. 4K,L, arrow). Foxa2 and Gsc, which are normally expressed in the prechordal plate in E8.0 (1-2 somites) embryos(Fig. 4M,O, arrow), were also absent from the anterior midline of Ssdp1 mutants, suggesting an abnormal development of the prechordal plate(Fig. 4N,P). The prechordal plate can be morphologically distinguished at E8.0 from its surrounding tissues by two characteristics: (1) it directly contacts the neuroectoderm,and (2) it is thicker than the neighboring notochordal plate(Fig. 4Q,S, arrowheads). In Ssdp1 mutants, the tissue anterior to the notochordal plate either failed to contact the neuroectoderm or had a similar thickness as the notochordal plate, or both (Fig. 4R,T, anterior end of notochordal plate is indicated by an arrowhead; data not shown), indicating the lack of a differentiated prechordal plate. In summary, the development of the AVE and ADE is normal until EB stage in Ssdp1 mutants, but the development of the prechordal plate is compromised from EHF stage onwards.

To understand how this late onset head organizer defect in Ssdp1mutants affects head development, we examined the extent of regionalization of anterior neuroectoderm. From EHF to LHF stage, the anteriormost neuroectoderm of wild-type embryos is marked by Six3 and Foxd4(Fkh2) expression (Fig. 5A; data not shown). Approximately half of the Ssdp1mutants expressed these genes, although expression was generally reduced in strength and in extent (Fig. 5B; data not shown). Otx2 expression, which marks the prospective forebrain and midbrain regions(Fig. 5C), was detected in all Ssdp1 mutants at E8.0, although the size of the expression domain was slightly reduced (Fig. 5D). These results suggest that anterior head development is initiated in Ssdp1 mutants between EHF stage and E8.0, although the most anterior neuroectoderm fates may not be specified in some embryos.

At E8.5 (8-10 somites), Six3 and Otx2 are expressed at the anterior margin of the neural plate, with Six3 marking the forebrain and Otx2 extending more caudally to mark the midbrain(Fig. 5E,G). In Ssdp1mutants, Six3 was not expressed(Fig. 5F), while expression of Otx2 was either absent or confined to the anterior-most tip of the embryo (Fig. 5H; data not shown). Furthermore, Foxg1 (BF1) and Pax6, which normally mark the telencephalon and the diencephalon, respectively, were not expressed in Ssdp1 mutants (Fig. 5I,J; data not shown). The midbrain-hindbrain boundary (MHB)markers En2, Fgf8 and Wnt1 were either absent from the region corresponding to the MHB or expressed at the anterior tip in Ssdp1 mutants (Fig. 5K-P; data not shown). Expression of Krox20 in rhombomeres 3 and 5 of the hindbrain and of Uncx4.1 in the somites was unaltered in Ssdp1 mutants at E8.5-9.5(Fig. 5M,N; data not shown). These results suggest that the head neuroectoderm anterior to the MHB was initially specified, but was lost by E8.5 in some Ssdp1 mutants,while the posteriormost region of the midbrain was maintained in others.

To determine whether the absence of the anterior neuroectoderm of E8.5 Ssdp1 mutants is caused by the loss or reduced growth of previously specified anterior neuroectoderm, we examined apoptosis and cell proliferation using TUNEL and anti-PCNA antibody staining in these embryos. At E8.0 and E8.5, no significant difference in apoptosis was observed between wild-type and Ssdp1 homozygous embryos (Fig. 6A-D). At E9.0 (16-18 somites), however, there was a significant increase in the number of apoptotic cells in the degenerating head neuroectoderm and somites of mutant embryos (data not shown). At E8.0, the level of cell proliferation was similar between wild-type and Ssdp1mutant embryos (Fig. 6G). However, at E8.5 and E9.0, the global increase in cell proliferation observed in wild-type embryos (Fig. 6H;data not shown) was greatly attenuated in all tissues examined in Ssdp1 mutants (Fig. 6E,F,H; data not shown). These results suggest that changes in apoptosis and cell proliferation do not account for the specific loss of anterior neuroectoderm in E8.5 Ssdp1 mutants. The strong, global reduction of cell proliferation and increase in apoptosis in E9.0 somites may,however, contribute to the widespread growth retardation of Ssdp1mutants (Fig. 1Q).

Ssdp1 has been shown to interact biochemically with Ldb1(Chen et al., 2002; van Meyel et al., 2003), a co-factor of LIM domain proteins such as Lim1(Agulnick et al., 1996). As Ssdp1 mutant embryos have defects in head organizer development,which requires Lim1 activity, we hypothesized that Ssdp1 forms a complex with Lim1 through Ldb1 binding and regulates the transcriptional activity of this complex in the late head organizer. Indeed, co-expression of Ssdp1, Lim1 and Ldb1 in the prechordal plate, as revealed by in situ hybridization of Ssdp1 and Ldb1 andβ-galactosidase staining of Lim1-lacZ knock-in embryos, hints at the presence of the Ssdp1-Lim1-Ldb1 ternary complex in this tissue and/or its precursor (Fig. 2K, Fig. 7A-D; data not shown).

To reveal the molecular function of Ssdp1, we first analyzed the transcriptional activity of Ssdp1 using a GAL4-UAS system in P19 embryonic carcinoma cells. The effector, a fusion protein of the DNA-binding domain of GAL4 and full-length Ssdp1, activated reporter gene expression via GAL4-binding sites in a dose-dependent manner in transfected cells(Fig. 7E,F). Thus, Ssdp1 contains a transcriptional activation domain.

To study the role of Ssdp1 in the regulation of Lim1-Ldb1-dependent transactivation, we used the Xenopus Gsc promoter (up to –492 bp) to drive expression of a reporter gene(Fig. 7G). This promoter has previously been shown to be regulated by a Lim1-Ldb1 complex through two elements, UE and DE (Mochizuki et al.,2000). As Gsc expression was lost in Ssdp1mutants (Fig. 4P), it seemed a likely target for a complex including Lim1, Ldb1 and Ssdp1. Neither Ldb1 nor Ssdp1 alone was able to activate the Gsc promoter in transfected cells, while Lim1 was able to activate the promoter by itself, probably through interactions with endogenous Ldb1 and Ssdp1. Co-expression of Ldb1 or Ssdp1 with Lim1 slightly enhanced the activation, while co-expression of all three proteins strongly increased expression of the reporter gene(Fig. 7H). Moreover, Ssdp1 enhanced transactivation in a dose-dependent manner in the presence of Ldb1(Fig. 7I). These results suggest that Ssdp1 enhances the activator function of the Lim1-Ldb1 complex. This supports our hypothesis that Ssdp1 functions by forming a ternary complex with Lim1 and Ldb1, and acting as an activator component of the complex(Fig. 7J). Therefore, it is likely that the reduced expression of Ssdp1 in hskhomozygous embryos decreases the transcriptional activity of the Lim1 complex in the prechordal plate, resulting in its abnormal development.

If Ssdp1 functions by forming a Lim1-Ldb1-Ssdp1 ternary complex during development, the simultaneous reduction of Ssdp1 and either Lim1 or Ldb1 should reduce the amount of this complex and cause similar developmental defects as observed in Ssdp1 mutants. We tested this hypothesis by making compound heterozygous mutant embryos. All of the Ssdp1^+/hsk (n=6) and Lim1^+/-(n=10) embryos obtained by crossing Ssdp1^+/hsk,and Lim1^+/- mice appeared normal at E9.0-9.5 (data not shown), whereas Ssdp1^+/hsk;Lim1^+/- embryos exhibited variable phenotypes that could be classified into three categories. Type I embryos showed no obvious defects(Fig. 8B). Type II embryos exhibited microcephaly with relatively normal body size(Fig. 8C), while type III embryos showed severe growth retardation in addition to microcephaly, and failed to undergo embryonic turning (Fig. 8D).

The majority of Ssdp1^+/hsk (normal/total=6/7) and Ldb1 (n=6/7) mutants obtained from crosses between Ssdp1^+/hsk and Ldb1^+/- mice appeared normal at E9.0-9.5 (Fig. 8E). Other single mutants displayed mild growth retardation and mild microcephaly(data not shown). Phenotypic differences between Ssdp1^+/hsk embryos obtained from different crosses may reflect differences in genetic background between Lim1 (ICR) and Ldb1 (C57BL/6) mutants. Approximately half of the Ssdp1^+/hsk;Ldb1^+/- embryos appeared normal (Fig. 8F), while the remaining mutants showed varying degrees of microcephaly and growth retardation that were difficult to classify(Fig. 8G). The incomplete penetrance of Ssdp1^+/hsk Lim1^+/- and Ssdp1^+/hsk; Ldb1^+/- phenotypes may reflect the fact that hsk is not a null mutation. However, these results indicate that Ssdp1 genetically interacts with Lim1and Ldb1 in both head development and body growth, which is consistent with the model that the Ssdp1-Lim1-Ldb1 complex regulates these developmental processes in vivo.

Head development is initiated by the AVE and maintained by the anterior AME, which is comprised of the ADE and the prechordal plate. The development of all of these head organizer tissues requires the function of Lim1(Shawlot and Behringer, 1995; Shawlot et al., 1999). Both Lim1 and Ldb1 mutant embryos have a constriction at the embryonic/extra-embryonic boundary at E7.5 and lack head structures anterior to rhombomere 3 at E8.5 (Mukhopadhyay et al., 2003; Shawlot and Behringer, 1995), suggesting that Lim1 and Ldb1 make a functional complex in vivo that is crucial for development of gastrulation stage mouse embryos.

Experimental removal or transplantation of anterior AME at EHF stage has demonstrated the crucial role of the prechordal plate in anterior head development (Camus et al.,2000). Our analysis of hsk mutant embryos demonstrates the essential role of Ssdp1 in prechordal plate development after the EHF stage. The head defects observed in Ssdp1 mutants resembled those of chimeric embryos lacking Lim1 in all embryonic tissues, including the prechordal plate, suggesting a strong link between Ssdp1 and Lim1 in prechordal plate development (Shawlot et al., 1999). Co-expression of Ssdp1, Lim1 and Ldb1 in the prechordal plate as well as genetic interactions between Ssdp1 and Lim1 or Ldb1 in head development strongly supports the hypothesis that these three proteins constitute a functional complex in vivo. In addition to the co-activator function of Ssdp1 in transfected cells, downregulation of the Lim1 target genes gsc(Mochizuki et al., 2000) and Foxa2 (Shawlot and Behringer,1995) in Ssdp1 mutants suggests a model in which Ssdp1 is an essential activator component of the Ssdp1-Lim1-Ldb1 complex in the prechordal plate. As each Ssdp molecule has been suggested to bind to a single Ldb1 molecule (van Meyel et al.,2003), the functional complex should consist of(Ssdp1)₂-(Lim1)₂-(Ldb1)₂(Fig. 7J).

The head defects of Ssdp1 mutants also resemble those of chimeric embryos lacking Otx2 function in the embryonic tissues, including the anterior AME (Rhinn et al.,1998), or of mutants lacking Otx2 function specifically in the neuroectoderm (Kurokawa et al.,2004; Suda et al.,1999), raising the possibility that reduced Otx2 activity in the anterior AME and/or neuroectoderm also contributes to the headless phenotype of Ssdp1 mutants. As Lim1 and Otx2 proteins have been shown to interact biochemically (Nakano et al.,2000), it is possible that Ssdp1 also functions by forming a complex that includes Ldb1, Lim1 and Otx2. However, although interaction of Ldb1 with the Otx-type homeodomain protein Pitx1(Bach et al., 1997) suggests that Otx2 function might be regulated by Ssdp1 independently of Lim1, the sequence similarities between Otx2 and Pitx1 are limited to the homeodomain,which is itself insufficient for interaction with Ldb1. To date, there is no direct evidence for an interaction between Otx2 and Ldb1, but this possibility should be addressed in the future.

Studies in lower vertebrates have suggested that the role of the prechordal plate in head development is to protect the overlying anterior neuroectoderm from posteriorizing Wnt signals and ventralizing BMP signals(Glinka et al., 1997). A similar role for the prechordal plate in the mouse was also suggested by the absence of anterior head structures in mice that lack the Wnt antagonist Dickkopf1 and in double mutants lacking the two BMP antagonists Noggin and Chordin (Bachiller et al.,2000; Mukhopadhyay et al.,2001). However, the mechanism causing head defects in these embryos is not known.

In Ssdp1 mutants, specification of anterior neuroectoderm was incomplete, suggesting that the prechordal plate at EHF stage is required for this process. Later loss of head neuroectoderm anterior to the MHB suggests that the prechordal plate is required for maintenance of the anterior neuroectoderm between E8.0 and E8.5. As changes in cell death and/or proliferation cannot account for the loss of anterior head tissues, the most likely reason for the loss of head structures is a change in gene expression patterns leading to the posterior transformation of neuroectoderm normally specified to become fore- or midbrain. This is consistent with the model that the prechordal plate after EHF stage promotes head development by maintaining the expression of anterior neuroectoderm genes by protecting this tissue from posteriorizing signals. Supporting this notion, expression of Dkk1, an antagonist of the posteriorizing signal Wnt, is lost in the prechordal plate of Ssdp1 mutants. In addition, abnormal specification of the anterior neural plate may secondarily cause degeneration of anterior head tissue, as revealed by the increased apoptosis in E9.0 Ssdp1 mutants. A similar increase in pyknotic cells in degenerating head tissue was also reported for two mutants lacking ADE and prechordal plate tissue(Vincent et al., 2003).

Ssdp1 mutants exhibited a global reduction in cell proliferation after E8.5 and an increase in apoptosis in somites at E9.0. These changes may be at the root of the abnormalities such as growth retardation and kinked neural tube that were observed in Ssdp1 mutants. Although the mechanism by which Ssdp1 regulates cell proliferation is unknown at present, growth retardation of Ssdp1^+/hsk;Lim1^+/- and Ssdp1^+/hsk;Ldb1^+/- compound mutants suggests involvement of a Ssdp1-Lim1-Ldb1 complex in this process. A shortened body axis was also observed in embryos lacking either Ldb1 or Lim1, supporting this hypothesis(Mukhopadhyay et al., 2003; Shawlot and Behringer, 1995). However, if the Lim1 complex plays a major role in the regulation of cell proliferation and cell death, it must be through an indirect mechanism, as Lim1 is not expressed in all of the affected cells. It is conceivable that defective gastrulation movements(Hukriede et al., 2003; Tam et al., 2004) or the inability of cells with reduced Lim1 complex activity to induce lateral plate mesoderm genes (Tsang et al.,2000) secondarily affects the proliferation and survival of surrounding cells. Furthermore, it is possible that Ssdp1 may also function independently of Lim1, in which case the Ldb1-Ssdp1 complex may regulate cell proliferation in a cell-autonomous manner by controlling the activities of transcription factors involved in cell cycle regulation and cell survival. Alternatively, Ssdp1 might play a direct role in the DNA replication process as a single stranded DNA-binding protein(Bayarsaihan et al., 1998).

Analysis of hsk mutants showed that disruption of the Ssdp1 gene and the resulting reduction in Ssdp1 expression causes defects in the prechordal plate development and anterior truncations,with some mutants also exhibiting smaller body size. In vitro data demonstrated that Ssdp1 acts as a coactivator that enhances transcriptional activation by the Lim1-Ldb1 complex. Moreover, genetic interactions between Ssdp1 and Lim1 or Ldb1 suggest that the phenotypes observed in Ssdp1 mutants very probably reflect reduced activity of a Lim1 complex. Together, our data demonstrate that Ssdp1 acts as an essential activator component of a Ssdp1-Lim1-Ldb1 complex in the development of the prechordal plate and body growth.

We thank Ms Y. Minami for excellent technical assistance; Dr S. Aizawa and LARGE members for mouse embryo manipulations and the housing of mice; Drs I. Matsuo and C. Kimura-Yoshida for technical advice; Dr P. Tam for help on staging and histology of embryos; Drs A. Sawada and S. Yamamoto for helpful discussions; Drs A. Shimono, E. Lai, E. DeRobertis, C. Niehrs, A. Joyner, G. Martin, B. Hogan, K. Kaestner, P. Thomas, D. Wilkinson, H. Hamada, S.-L. Ang,A. Mansouri, P. Gruss, A. McMahon, I. Matsuo, J.-I. Miyazaki, H. Niwa, M. Taira, K. W. Cho and M. Hibi for reagents. N.N. is a recipient of a fellowship from the Japan Society for the Promotion of Science (JSPS). This work was supported by grants from the Ministry of Education, Science, Culture, Sports,Science, and Technology of Japan (MEXT) (14034230), JSPS (14380345) and RIKEN to H.S., and a grant from MEXT to H.K. (12CE2007).