The murine homolog of SALL4, a causative gene in Okihiro syndrome, is essential for embryonic stem cell proliferation, and cooperates with Sall1 in anorectal, heart, brain and kidney development

Early differentiation of the mammalian embryo leads to the development of two distinct lineages - the inner cell mass (ICM) and the trophectoderm. Cells of the ICM are pluripotent and give rise to the epiblast and eventually to all the fetal tissues, while trophectoderm cells have restricted potential and give rise to the trophoblast cell layers of the placenta. Embryonic stem (ES)cells are established from the ICM and have the potential to generate chimeric mice when introduced into a blastocyst. When cultured in vitro, ES cells differentiate into a variety of cell lineages; thus, human ES cells are possible candidates for cell therapies, although there are some ethical issues to overcome. Maintenance of mouse ES cell pluripotency requires constant suppression of their differentiation by both extrinsic and intrinsic factors. Leukemia inhibitory factor (LIF) and bone morphogenetic protein 4 (BMP4) are the two major extrinsic signals (Niwa et al., 1998; Matsuda et al.,1999; Ying et al.,2003). LIF induces phosphorylation and nuclear localization of STAT3, a transcription factor essential for the LIF-dependent pathway for self-renewal. BMP4 induces phosphorylation and nuclear localization of Smad1,and the subsequent upregulation of Id helix-loop-helix proteins that block neural differentiation and maintain the pluripotency of ES cells in the presence of the LIF signal. Intrinsic factors include Oct3/4, Nanogand Eras. Oct3/4 and Nanog are expressed at high levels in ES cells and have been shown to be essential for maintaining pluripotency in ES cells (Chambers et al.,2003; Mitsui et al.,2003; Nichols et al.,1998). Eras and Utf1 are also expressed abundantly in ES cells, and they play crucial roles in the proliferation of stem cells (Takahashi et al.,2003; Nishimoto et al.,2005).

The spalt (sal) gene was first isolated from Drosophila and it encodes a protein characterized by multiple double zinc-finger motifs of the C2H2 type. sal acts as a region-specific homeotic gene, and is required for the specification of the head and tail regions during early development (Jurgens,1988; Kuhnlein et al.,1994). During the later stages of development, salregulates pattern formation and cell fate decisions in the wing disc(de Celis et al., 1996; Nellen et al., 1996), trachea(Kuhnlein and Schuh, 1996) and sensory organ development (de Celis et al.,1999). sal is expressed at the anteroposterior boundary of the wing imaginal discs, and its expression is controlled by the dpp (decapentaplegic) gene (de Celis et al., 1996; Nellen et al., 1996).

Humans and mice each have four known Sal-related genes (known as SALL1-SALL4 in humans and Sall1-Sall4 in mice). Mutations in SALL1 on chromosome 16q12.1 have been associated with Townes-Brocks syndrome, an autosomal dominant disease characterized by dysplastic ears, a preaxial polydactyly, imperforate anus and, less commonly, kidney and heart anomalies (Kohlhase et al.,1998). Mice deficient in Sall1 show kidney agenesis or severe dysgenesis, but other phenotypes observed in the human disease are not apparent (Nishinakamura et al.,2001). This discrepancy could be explained by the formation of truncated SALL1 proteins as a result of mutations in SALL1, as comparison with Sall1-null mice showed that mutant mice producing a truncated Sall1 protein exhibited more severe defects, including renal agenesis, exencephaly, as well as limb and anal deformities(Kiefer et al., 2003). It has been reported that Sall1 also functions as a transcriptional repressor by localizing in the heterochromatin and interacting with components of chromatin remodeling complexes such as histone deacetylase (HDAC)1, HDAC2,retinoblastoma-associated protein 46/48 (RbAp46/48), metastasis-associated protein (MTA)1 and MTA2 (Kiefer et al.,2002).

It is still unclear if SALL2 is associated with human disease;however, it has been reported that Sall2 functions as a tumor suppressor(Li et al., 2001; Li et al., 2004). Sall2-deficient mice show no apparent phenotype, and mice lacking both Sall1 and Sall2 show kidney phenotypes comparable with those of Sall1 knockout mice(Sato et al., 2003). Although no diseases are thus far directly linked to SALL3, this gene is located in a region that is commonly deleted in cases of 18q deletion syndrome(Kohlhase et al., 1999). Individuals with this deletion exhibit hearing loss, cardiac problems, mental retardation, midfacial hypoplasia, delayed growth and limb abnormalities(Strathdee et al., 1997). Sall3-null mice die on the first postnatal day and deficiencies in the cranial nerves and abnormalities in the oral structures are present(Parrish et al., 2004). Mutations in SALL4 cause an autosomal dominant disorder known as Okihiro syndrome, which is characterized by limb deformity, eye movement deficits and, less commonly, anorectal, ear, heart and kidney anomalies(Al-Baradie et al., 2002; Kohlhase et al., 2002).

To investigate the roles of Sall family genes and their functional redundancy in organogenesis, we generated Sall4-deficient mice. In this study, we report that the mouse Sall4 gene was found to be unexpectedly essential for early embryogenesis, and for the proliferation of ES cells. We further reveal the importance of the heterodimerization of Sall4 and Sall1 in vivo, which could explain the underlying genetic mechanisms involved in Townes-Brocks syndrome that is caused by truncations of SALL1.

A Sall4-targeting vector was constructed by incorporating the 5′ Aor51HI-EcoRI 8.2-kb Sall4 fragment and the 3′ BamHI-EcoRV 2.3 kb fragment into a vector that contained the β-galactosidase gene (lacZ), the neomycin resistance (Neo^r) gene (pGK-Neo) and the diphtheria toxin A subunit (pMC1DTA) in tandem. This construct deletes all the eight zinc-finger domains of Sall4 and results in the fusion of 39 amino acids at the N terminal of Sall4 and β-galactosidase(Fig. 1A). This strategy is identical to the one we used to delete Sall1(Nishinakamura et al., 2001). Five out of 114 G418-resistant E14.1 ES clones were correctly targeted and two were used to generate germline chimeras that were bred with C57BL/6J females. E14.1 ES cells were maintained on mitomycin C-treated primary embryonic fibroblasts in the presence of LIF (10³ U/ml) and serum in all the procedures and experiments described in this paper. The primer sequences used for genotyping were as follows: 5′-GAGGACTCCATACCGGTGAA-3′,5′-GTGCCCAGCTTCTTCAAGTC-3′ and 5′-CCTCTTCGCTATTACGCCAG-3′ (the length of the amplified segment was 272 bp in the wild-type allele and 379 bp in the mutated allele). As the Sall4-deficient mice (Sall4-del) lacked proper lacZexpression, we also generated mice using the Sall4-IRES-β-geo vector described below.

The Sall4-IRES β-geo vector was constructed by incorporating the 5′ PacI-EcoRI 4.2 kb fragment,the EcoRI-SalI 6.0 kb IRES-β-geofragment and the 3′ ApaLI-EcoRV 6.0 kb fragment into a vector that contained pMC1DTA (see Fig. S1 in the supplementary material). The Sall4-IRES-Hyg vector was generated by a similar method. These vectors confer drug resistance only when they are incorporated into the promoter regions of genes that are expressed in ES cells, including Sall4 (thus facilitating the isolation of homologous recombinants). Indeed, when tested in wild-type ES cells, homologous recombinants were obtained at a frequency of 68.8% (11/16) by using the Sall4-IRES-β-geo vector and at 66.7% (8/12)with the Sall4-IRES-Hyg vector. The Sall4-IRES-β-geo clones were used to generate another strain of mice (Sall4-βgeo) for monitoring Sall4 expression, as shown in Fig. 4D,E. This strain showed phenotypes identical to the original Sall4-deficient mice (Sall4-del).

The Sall4-flox vector was constructed by incorporating the 5′ HindIII-HindIII 3.5 kb fragment, the HindIII-ApaLI 4.7 kb fragment and the 3′ ApaLI-EcoRV 6.0 kb fragment into a vector that contained pGK-Neo flanked by Frt and loxP sequences and pMC1DTA (see Fig. S1 in the supplementary material). LoxP sequences were placed so that exon 2 and 3 were excised upon Cre treatment, resulting in disruption of all zinc-finger motifs. Five out of 170 clones were correctly targeted (flox/+) and one of them was further transfected with the Sall4-IRES-Hyg vector. Both resulting clones, flox/- and +/-, proliferated normally. Neo^rplaced in intron 2 did not affect Sall4 expression as shown in Fig. 3C. These ES cells were infected with adenovirus expressing Cre under the CAG promoter (AxCANCre provided by RIKEN Bioresource Center)(Niwa et al., 1991; Kim et al., 2002) at a m.o.i(multiplicity of infection) of 50. After incubation for 1 hour, cells were diluted and plated onto 6-well plates coated with mitomycin C-treated embryonic fibroblasts.

Histological examination was performed as described earlier(Nishinakamura et al., 2001). In situ hybridization was performed using the AmpMap Kit and an automated Discovery System (Ventana) according to the manufacturer's protocols. Blastocyst culture and immunosurgery were also performed as described earlier(Nichols et al., 1998). cDNA was synthesized using SuperScript III CellsDirect cDNA Synthesis System(Invitrogen). Primer sequences used for RT-PCR are available upon request. 5-Bromo-2′-deoxyuridine (BrdU) labeling and detection kit I (Roche) was used to examine proliferation of the blastocysts upon 30 minutes of BrdU incorporation.

Two independent Sall4-deficient clones were transfected with pCAG-GFP-IRES-puro and selected on puromycin-resistant embryonic fibroblasts(Tucker et al., 1997). Cells retaining ubiquitous green fluorescent protein (GFP) expression from each Sall4-deficient clone were injected into blastocysts, and consistent results were obtained. Frozen sections of the chimeras were stained by an anti-GFP antibody (Molecular Probes) and detected using ImmunoPure metal enhanced DAB substrate kit (Pierce).

For the proliferation assay, 1×10⁴ cells were plated per well in 24-well plates in triplicate on mitomycin C-treated primary embryonic fibroblasts in media containing LIF and serum, and passaged at the same density every 4 days for 16 days to determine the cumulative cell number. A BrdU Flow Kit (BD Biosciences) was used for cell cycle analysis of ES cells after 2 hour incorporation of BrdU. For rescue analysis, the Sall4expression vector (Sall4 in pCAG-IRES-puro) or GFP expression vector(negative control) was introduced into Sall4-deficient cells by electroporation. Multiple clones were selected on puromycin-resistant embryonic fibroblasts and expanded.

The siRNA duplexes were designed to target the coding region of mouse Sall4 cDNA at nucleotide 2761-2785 and synthesized by Invitrogen. Sall4-siRNA or control siRNA containing the same GC content was transfected into D3 embryonic stem cells using Lipofectamine2000 (Invitrogen) according to the manufacturer's instruction, except for the maintenance of D3 cells, which were cultured on gelatin-coated plates in the presence of LIF. D3 cells at a density of 1×10⁴ cells per well in 24-well plates were transfected in triplicate with 25 pmol siRNA, and counted every day. The negative control siRNA showed no growth or morphological impairment compared with the mock-transfected or the non-transfected cells.

The anti-Sall1 monoclonal antibody(Sato et al., 2004) and an anti-Sall4 polyclonal antibody raised against polypeptide (MAKHQFPHFLEENKI)corresponding to amino acids 1050-1064 were used. A monoclonal anti-Sall4 antibody was generated using Sall4 cDNA encoding amino acids 95-216. The following additional antibodies were used for staining blastocysts and ES cells: monoclonal anti-Oct3/4 (Santa Cruz), rabbit antisera against Oct3/4(Niwa et al., 2005) and monoclonal anti-Cdx2 (BioGenex). GFP-fused Sall4 and Sall1zinc-finger mutants in which the cysteine residues were replaced by glycine were produced using PCR. NIH 3T3 cells were plated onto six-well plates at a density of 1×10⁵ cells per well 1 day prior to transfection. The cells were transfected with 3 μg of the plasmids using FuGENE 6 (Roche)and cultured for 48 hours prior to analysis by confocal microscopy. Construction of Sall1-GFP and Sall1^1-435-DsRed, immunoprecipitation,immunocytochemistry and confocal microscopy were performed as described previously (Sato et al.,2004).

Fig. 1A,B depict the strategy used to delete all the eight zinc-finger domains of Sall4. As Sall4-null mice did not survive beyond embryonic day (E) 6.5(Table 1), the embryos were examined at E5.5-6.5 (Fig. 1C,D). Although wild-type embryos developed an egg cylinder with a central proaminiotic cavity, seven out of the 26 embryos (26.9%) derived from the heterozygous intercross barely retained a structure that resembled the epiblast. These embryos were identified as Sall4-null by in situ hybridization, while Sall4 expression was observed in the epiblast,and less abundantly in visceral endoderm and extra-embryonic ectoderm in the wild-type embryos (Fig. 1D). Though the areas occupied by the epiblast markers (Oct3/4, Fgf4 and Nodal), the trophectoderm and extra-embryonic ectoderm marker(H19), and the extra-embryonic ectoderm marker (Bmp4) were significantly reduced, they were still present in Sall4-null embryos(Fig. 1D and data not shown). These data demonstrate that Sall4 is essential for embryonic development during the peri-implantation period, but not for commitment to the epiblast or extra-embryonic lineages.

As the drastic impairment of Sall4^-/- embryos hindered further examination, blastocysts from Sall4^+/-intercrosses (E3.5) were examined. At this stage, the Sall4 protein was expressed both in the inner cell mass (ICM) and trophectoderm in the wild-type(Fig. 2A). Homozygotes were,however, indistinguishable from wild-type or heterozygotes, and expression of Oct3/4 (ICM marker) and Cdx2 (trophectoderm marker) was not altered (Fig. 2A), indicating that lineage commitment between these two lineages occurs normally. Next,cultured blastocysts were investigated for phenotypic changes. The embryos hatched from their zonae and attached to the plates, and the trophectoderm grew in an identical manner. By day 5 in culture, all homozygotes(n=14) showed significantly reduced outgrowth of ICM compared with wild-type and heterozygote (n=43)(Fig. 2B). At day 3 of culture,when the phenotype was becoming apparent, Sall4-deficinet ICM was positive for Oct3/4, as determined by immunostaining(Fig. 2B), and expression of the lineage markers examined by RT-PCR was not impaired: Oct3/4(ICM), Gata6 and Gata4 (primitive endoderm), eomesodermin and H19 (trophectoderm) (Fig. 2C). These results suggest that lineage commitment does occur in the absence of Sall4. By contrast, at day 3 of culture, BrdU incorporation was significantly reduced in Sall4-deficient ICM when compared with wild type or heterozygote, while an apparent increase in apoptosis, as determined by TUNEL assay, was not detectable (Fig. 2D; data not shown), indicating that Sall4 is required for ICM proliferation. Next, the trophectoderm was removed by immunosurgery to rule out secondary effects of trophectoderm abnormality. Of 54 ICMs, 41 (76%)grew into a spheroid structure with a primitive endoderm surrounding it, and all the ICMs were genotyped as wild-type or heterozygote. By contrast, 13 out of 54 (24%) ICMs showed very few signs of growth(Fig. 2C rightmost column),suggesting that, independent of trophectoderm, Sall4 is essential for ICM outgrowth.

The above data prompted us to investigate Sall4 functions in ES cells, as these cells are derived from the ICM. To obtain Sall4-null ES cells,a vector containing the hygromycin resistance gene (Hyg) was introduced into heterozygous cells containing the Neo^rallele (see Fig. S1 in the supplementary material). We also attempted the reverse order, and both experiments resulted in a relatively frequent isolation of heterozygous cells in which the initial targeted allele was retargeted by the second vector. However, Sall4-null ES cells were isolated at a very low frequency (Table 2). To confirm these results, we constructed other versions of the vectors that contained promoterless β-geo or Hyg, thus facilitating the isolation of homologous recombinants (see Fig. S1 in the supplementary material). However, when introduced into heterozygous cells,most homologous recombinants showed retargeting of the mutated allele, and Sall4-null cells were obtained at a low frequency(Table 2). Therefore, the absence of Sall4 is disadvantageous to ES cells.

We next tried inducible knockdown of Sall4 by introducing siRNA oligonucleotides into ES cells. siRNA against Sall4 efficiently reduced Sall4 expression by day 2 after transfection, but Sall4 expression recovered by day 4(Fig. 3A), which was also confirmed by Sall4 immunostaining on ES colonies (data not shown). Sall4- siRNA-treated cells showed transient reduced growth with similar kinetics to the Sall4 expression, while negative control siRNA had no effect (Fig. 3B). By contrast, colony morphology and Oct3/4 staining were not altered in the absence of Sall4 (data not shown). These data suggest that Sall4 may be required for proliferation of ES cells.

Next, heterozygous cells were generated containing a floxed allele of Sall4 by homologous recombination (see Fig. S1 in the supplementary material). When the Sall4-IRES-Hyg vector was introduced into this clone, both alleles were targeted with a similar frequency (wild-type allele,10/23; floxed allele, 11/23), resulting in two types of cells: flox/- and +/-(Fig. 3C). As shown by western blot, Neo^r in intron 2 of the floxed allele did not affect Sall4 expression, and there was no difference in proliferation between the two types of cells. Upon infection with adenovirus expressing Cre, flox/- cells became almost Sall4-null by day 3, while +/- cells served as a negative control, as determined by western blot(Fig. 3C). Indeed, when cells were subsequently replated and single clones were picked, 44 out of 47 clones were Sall4 null, and all these Sall4-null clones grew significantly slower than the clones retaining Sall4 despite being treated identically (Fig. 3D),but could be cultured for a prolonged period of time (more than 1 month). None of the clones from the other groups (flox/- without Cre, +/- with or without Cre) showed impaired proliferation, and re-expression of Sall4 cDNA under a ubiquitous promoter in Sall4-null cells restored proliferation (data not shown), confirming that this phenotype was Sall4 dependent. Cell cycle analysis revealed that Sall4-null cells showed a decreased S-phase and increased G1-phase compared with heterozygous cells (Fig. 3E). These data suggest that Sall4 absence in ES cells leads to inefficient G1/S transition, which may possibly explain the phenotypes observed in blastocyst culture and embryos in vivo.

These Sall4-null ES cells formed compact colonies and were morphologically indistinguishable from heterozygous or wild-type cells(Fig. 3F). No significant differences were observed in LIF-induced STAT3 phosphorylation or BMP4-induced Id family upregulation between wild-type and Sall4-null ES cells,suggesting that the absence of Sall4 does not affect the response to the two major extrinsic signals involved in maintaining the pluripotency of ES cells (data not shown). In the Sall4-null ES cells, Oct3/4and Nanog, which are essential for pluripotency, and Erasand Utf1, which are important for the growth properties of the ES cells, continued to be expressed, which was confirmed by northern blot and immunostaining (Fig. 3F,G). Although expression of Oct3/4 and Utf1 was unaltered in the absence of Sall4, the expression of Nanog and Eraswas slightly decreased, and this was detailed in the discussion. Expression of markers for primitive endoderm (Gata6), mesoderm (T), and trophectoderm (Cdx2) was not detected (data not shown), which suggests that the proliferation defect observed in the absence of Sall4 is not secondary to aberrant differentiation. Embryoid bodies were formed, though were smaller, from Sall4-null ES cells, and Sall4-deficient cells produced markedly smaller tumors than did heterozygous cells when transplanted into nude mice (data not shown), again suggesting that Sall4 absence does not affect pluripotency of ES cells but proliferation. To further confirm that Sall4-null cells are undifferentiated, we injected these cells into blastocysts to generate chimeras. At E7.5, Sall4-null cells tagged with GFP contributed highly to the embryos (Fig. 3H). Taken together, our data indicate that Sall4 is essential for proliferation but not pluripotency of ES cells.

Genotyping from heterozygous crosses showed that nearly half the Sall4 heterozygous mice died in utero(Table 1). Out of 86 born heterozygous mice, 13 were runt and died or were eaten by their mothers within 3 weeks (Table 3). Eight more died within the next 3 weeks and six of these mice had significantly dilated bowels and apparent anal stenosis (Fig. 4A, open arrowhead). The number of enteric ganglia in these heterozygotes was not affected (data not shown); hence, gastrointestinal dilation is likely to be a secondary effect of anal dysplasia. When examined at E17.5-18.5, ventricular septum defects were also observed in some of the heterozygotes (Fig. 4B and Table 4). These phenotypes partially mimic the Okihiro syndrome caused by SALL4 mutations in humans, and, thus, anorectal and heart anomalies could be caused by SALL4 haploinsufficiency. Other phenotypes observed in Okihiro syndrome were not detected in Sall4-heterozygous mice. All the heterozygotes that survived beyond 3 weeks (n=73) had normal appearing extremities and closer examination of newborns did not show any anomalies in digit, metacarpus or metatarsus formation (n=6)(Fig. 4C). Though abducens nerves and nuclei are reported to be responsible for the abnormal eye movements in Okihiro syndrome (Al-Baradie et al., 2002; Kohlhase et al.,2002), the abducens nuclei in adult heterozygous mice(n=6) were properly formed, and oculomotor and trochlear nuclei, eye and ocular muscles were also unaffected(Fig. 4D and data not shown). Inner ear structures at E17.5 (n=9) and in adults (n=6) were not impaired (Fig. 4E; data not shown). Thus, in mice, not all Okihiro phenotypes were caused by Sall4 haploinsufficiency.

We also found other phenotypes not reported in humans. Some heterozygotes(three out of 26 examined at E11.5-15.5) exhibited exencephaly(Fig. 4F), and four out of 73 Sall4 heterozygous mice that survived up to 3 weeks exhibited tail flexion anomalies (data not shown). As exencephaly and a kinked tail are caused by failure of the neural tube to close, Sall4 may also play an important role in this event.

Compound heterozygotes were generated to investigate functional redundancy among members of the Sall family. Surprisingly, no Sall1/4compound heterozygotes survived after birth, whereas mice having other genetic combinations (Sall1/2, Sall1/3, Sall2/3, Sall2/4 and Sall3/4) survived. The Sall1/4 heterozygotes exhibited uni-or bilateral renal agenesis (Fig. 5A), exencephaly (data not shown), anorectal malformations(Fig. 5B) and ventricular septum defects (data not shown); the incidence of these phenotypes was significantly increased in comparison with Sall4 heterozygotes(Table 4), suggesting a genetic interaction of Sall4 and Sall1 in vivo. Next, we determined if the expression of Sall4 and Sall1 overlapped in the affected organs. At E8.5, a stage at which the neural tubes close, both Sall4 and Sall1 were expressed in the mesenchyme of the anterior portion (Fig. 5C,filled arrowhead) and in all tissues of the tail region(Fig. 5C, open arrowhead). At E11.5, Sall4 and Sall1 were expressed in the anorectal region (Fig. 5D, arrowhead). In the heart at E11.5, Sall4 was detected in the myocardium, including the developing interventricular septum, while Sall1 was expressed not only in the myocardium but also in the endocardium, thus overlapping with Sall4 in the myocardium (Fig. 5E). We further checked to determine if Sall1 and Sall4 were colocalized in ES cells. Endogenous Sall4 was localized in the punctate nuclear foci that colocalized with 4,6-diamidino-2-phenylindole (DAPI)(Fig. 5F), indicating that Sall4 is localized in the constitutive heterochromatin. Endogenous Sall1 was also localized in the heterochromatin and overlapped with Sall4. Immunoprecipitation using lysates from ES cells showed that endogenous Sall4 binds to Sall1 (Fig. 5G). Therefore, these two genes probably form heterodimers in the developing brain,heart and anorectal regions. However, in the developing kidney, the two genes did not overlap (data not shown). Thus, heterodimer-independent mechanisms may exist in kidney development.

When various Sall4 and Sall1 constructs with mutations in the zinc-finger domains were generated, we observed that the most C-terminal double zinc-finger domain (Zn4) of Sall4 was essential and sufficient for localization to the heterochromatin (Fig. 6A). In the case of Sall1, two double zinc fingers (Zn 4 and 5)were required, and were sufficient for heterochromatin localization(Fig. 6B). Thus, these C-terminal double zinc fingers constitute heterochromatin localization domains. SALL1 mutations in Townes-Brocks syndrome are likely to produce C-terminally truncated proteins. Truncated Sall1 fused to DsRed (Sall1^1-435-DsRed) was ubiquitously located throughout the cytoplasm and euchromatin(Fig. 6C), as this mutant lacked the C-terminal heterochromatin localization domain. Co-transfection of Sall1^1-435-DsRed and Sall4-GFP showed disturbance in Sall4 localization in the heterochromatin. Thus, the C-terminally truncated SALL1 proteins in Townes-Brocks syndrome probably dimerize with SALL4 in a dominant-negative manner, resulting in mislocalization of SALL4 in the heterochromatin. Considering the anal and heart anomalies and exencephaly in Sall1/4 double heterozygotes, the phenotypes observed in Sall1 truncations could be explained by the functional reduction of Sall4.

In this study, we showed that mouse Sall4 is essential for early embryogenesis and for proliferation of ES cells. We also provided both in vivo and in vitro evidence for the dimerization of Sall4 and Sall1 and revealed that the C-terminally truncated Sall1 in Townes-Brocks syndrome caused mislocalization of Sall4 in a dominant-negative manner. As human SALL4 is a causative gene for the Okihiro syndrome, characterized by limb deformity and eye movement abnormalities, the indispensable role of mouse Sall4 in early embryogenesis and proliferation of ES cells was entirely unexpected. Blastocyst culture analysis combined with the trophectoderm removal technique identified the importance of Sall4 in the inner cell mass that is the origin of the embryo proper. As ES cells are derived from the inner cell mass, it is understandable that Sall4-null ES cells showed significantly reduced proliferation, which may possibly explain the phenotypes observed in blastocyst culture and embryos in vivo.

How can Sall4 regulate the proliferation of ES cells? As Sall1 is localized in the heterochromatin and binds to the components of chromatin remodeling complexes [namely HDAC1, HDAC2, RbAp46/48, MTA1 and MTA2(Kiefer et al., 2002)], Sall4 could also bind HDAC complexes and deacetylate the histone lysine residues,followed by recruitment of histone-methylating and DNA-methylating complexes. Centromeric and pericentromeric regions, which comprise constitutive heterochromatin, are required for spindle formation in mitosis, and it is possible that the absence of Sall4 affects these structures. Alternatively,Sall4 could suppress downstream genes, including cell cycle inhibitors, by inducing heterochromatin formation in the euchromatic promoters (known as facultative heterochromatin formation). More detailed examination in the heterochromatin structure in Sall4 mutant ES cells, as well as identification of downstream target genes of Sall4, is required for further elucidation of Sall4 functions. It is interesting that Nanog and Eras were slightly decreased in Sall4-deficient cells. Nanog heterozygous cells show no reduction in proliferation, and Nanog-deficient cells differentiate into primitive endoderm(Mitsui et al., 2003). Though Eras could be partly responsible for impaired proliferation of Sall4-null ES cells, Eras-null mice had no abnormalities in vivo (Takahashi et al., 2003). Thus, it is unlikely that a reduction of these genes could explain the phenotypes of Sall4 deficiency. As ES cells are suitable for quantitative biochemical approaches, we expect to apply the detailed analysis of Sall4 function in ES cells to the study of organ formation during later stages of development.

We demonstrated that Sall4 and Sall1 form heterodimers and that truncated Sall1 proteins altered Sall4 localization in a dominant-negative manner. Townes-Brocks syndrome caused by SALL1 mutations exhibits limb, anal,ear, kidney and heart anomalies, and mice retaining truncated Sall1 proteins show similar phenotypes as well as exencephaly. As human families with severe phenotypes are unlikely to survive, exencephaly could be one of the phenotypes of SALL1 truncations. Considering the anal and heart anomalies and exencephaly in Sall1/4 double heterozygotes, at least these three phenotypes observed in Sall1 truncations can be explained by the functional reduction of Sall4. Thus, we propose that some symptoms of Townes-Brocks syndrome caused by SALL1 truncations result from the inhibition of SALL4 functions that is due to heterodimer formation.

SALL4 mutations in humans cause the autosomal dominant disorder Okihiro syndrome, which is characterized by limb deformity, eye movement(abducens nerve) abnormalities, and anorectal, ear, heart, and kidney anomalies. Some Sall4 heterozygous mice had anal and heart anomalies,suggesting that these phenotypes are caused by Sall4haploinsufficiency. This is important because most phenotypes in Townes-Brocks syndrome could be caused by the dominant-negative effect of the truncated Sall1, but not by haploinsufficiency of Sall1. However, other anomalies to limbs, abducens nuclei, inner ears and kidneys were not detected in Sall4 heterozygous mice; thus, these phenotypes are not explained by Sall4 haploinsufficiency in mice. In humans, C-terminally truncated SALL4 proteins could function in a dominant-negative manner, as is the case for SALL1 truncations. Recently, a gene trap allele retaining truncated Sall4 was reported and this strain exhibited digit and heart anomalies (Koshiba-Takeuchi et al.,2006). If this allele serves as a dominant-negative form, it is possible that this trap allele has more severe phenotypes than ours, although other phenotypes are not described. Direct comparison of the two mouse strains on the same genetic background would test this possibility.

We thank H. Niwa for providing the plasmids(IRES-βgeo, IRES-Hyg, Oct3/4and H19) and antisera against Oct3/4; H. Suemori, for the anti-mouse antibody for immunosurgery; A. P. Monaghan for providing Sall3-deficient mice; H. Masai, M. Nakao, T. Aoto and K. Okita for technical advice; N. Takeda, Y. Kataoka, H. Meguro, S. Yamada, A. Nakane and R. Sakamoto for technical assistance; and T. Hara for critical reading of the manuscript. This work was supported in part by the Ministry of Education,Culture, Sports, Science, and Technology; and by the Ministry of Health,Labor, and Welfare of Japan.