Mutations in ZFHX1B cause Mowat–Wilson syndrome (MWS) but the precise mechanisms underlying the aberrant functions of mutant ZFHX1B proteins (also named Smad-interacting protein-1, SIP1) in patients are unknown. Using mass spectrometry analysis, we identified subunits of the NuRD corepressor complex in affinity-purified Zfhx1b complexes. We find that Zfhx1b associates with NuRD through its N-terminal domain, which contains a previously postulated NuRD interacting motif. Interestingly, this motif is substituted by an unrelated sequence in a recently described MWS patient. We show here that such aberrant ZFHX1B protein is unable to recruit NuRD subunits and displays reduced transcriptional repression activity on the XBMP4 gene promoter, a target of Zfhx1b. We further demonstrate that the NuRD component Mi-2β is involved in repression of the Zfhx1b target gene E-cadherin as well as in Zfhx1b-induced neural induction in animal caps from Xenopus embryos. Thus, NuRD and Zfhx1b functionally interact, and defective NuRD recruitment by mutant human ZFHX1B can be a MWS-causing mechanism. This is the first study providing mechanistic insight into the aberrant function of a single domain of the multi-domain protein ZFHX1B/SIP1 in human disease.
Loss-of-heterozygosity of ZFHX1B (encoding ZFHX1B, Zinc finger homeobox 1B, also named SIP1 and ZEB2) is implicated in the etiology of Mowat–Wilson syndrome (MWS). All patients with typical MWS present severe mental retardation and display distinctive facial dysmorphism in addition to multiple variable features, including heart defects, agenesis of the corpus callosum, seizures, urogenital anomalies and Hirschsprung disease (1). Analysis of the phenotype of Zfhx1b conventional knockout mouse embryos revealed defects in the neural crest cell lineage (2,3). In addition, the reduced expression of the neuroectodermal gene Sox2 and the sustained expression of E-cadherin in the Zfhx1b defective neural plate indicated defects in early neurogenesis in these embryos (2). Both classes of defect are relevant to the neurological problems seen in MWS patients. Zfhx1b was also found to be a crucial regulator of early neural tissue formation in Xenopus (4).
We have identified Zfhx1b as a protein that binds to TGFβ family receptor activated Smad proteins and acts as a DNA-binding transcriptional repressor (5,6). After δEF1 (Zfhx1a), Zfhx1b became the second and only other member of the Zfhx1 family in vertebrates. These multi-domain repressors are characterized by the presence of a homeodomain-like domain and two flanking C2H2-type zinc finger clusters. Both clusters are necessary for Zfhx1b to repress target genes by binding to spaced CACCT(G) sequences in their regulatory region (6). Though studies have focused mainly on Zfhx1b repressive function, transcriptional activation activity has been ascribed to Zfhx1b as well (7,8). Indeed, Zfhx1b interacts not only with the corepressor CtBP (C-terminal binding protein) (9,10), but full-length Zfhx1b also binds to the coactivators p300 and pCAF (p300/CBP associated factor) (11). Taken together, Zfhx1-type proteins are complex multi-domain proteins: not all of their functions may depend on interaction with Smad proteins, and association with different partners via different domains may provide selectivity for target gene repression or activation.
The vast majority of the clinically characterized human mutations are all gene deletions, nonsense or frame shift mutations affecting a single allele. In nearly all cases, this leads to drastic C-terminal truncation of ZFHX1B in humans, and possibly instability of the protein (12). Recently, also non-truncating mutations in ZFHX1B in patients presenting with a mild form of MWS have been identified, leading to the assumption of a possible correlation between more subtle mutations and a milder or atypical phenotype (13–16). Despite the emerging list of mapped truncating and non-truncating mutations in ZFHX1B, the precise molecular mechanism(s) underlying the clinical features of MWS patients carrying these mutations are still unknown. In a global approach to address Zfhx1b’s mechanism(s) of action, we decided to purify Zfhx1b protein complexes and identify candidate Zfx1b direct binding partners or proteins in its complex by mass spectrometry (MS). Doing so, we found a novel association of Zfhx1b with multiple subunits of the nucleosome remodeling and histone deacetylation complex NuRD. NuRD is composed of at least eight subunits, including the ATPase Mi-2β, which confers the chromatin remodeling activity (17). NuRD has been purified from both mammalian and amphibian cells and is considered to play a key role in transcriptional repression (18–21). Results from our study allow us to propose defective NuRD recruitment by Zfhx1b as part of Zfhx1b’s mechanism of action and—here—as an underlying cause of a form of MWS.
Affinity purified Zfhx1b complexes contain multiple NuRD subunits
We generated a tagged Zfhx1b protein in which a strep tag (S) is preceded by a 3XFlag epitope (F) at the N-terminus (Fig. 1A). Protein complexes containing this 3XFS-Zfhx1b were isolated by means of a variant tandem affinity purification approach (Fig. 1B). Briefly, total extracts of human embryonic kidney (HEK293T) cells transiently producing 3XFS-Zfhx1b proteins were subjected to anti-Flag agarose affinity gel-based purification. Following elution with excess 3XFlag peptide, the eluted proteins were applied to a streptactin agarose column as a second purification step. After gel electrophoresis and staining of the proteins, the bands were excised and the polypeptides were identified by MS (Fig. 1C). The major polypeptide, migrating as a protein of 170 kDa present in the purified 3XFS-Zfhx1b sample, was shown to correspond with 3XFS-Zfhx1b. Amongst the proteins that copurified with Zfhx1b were the known partners CtBP1 and -2 (9,10). Additionally, we identified novel Zfhx1b partners that are subunits of the NuRD complex, i.e. Mi-2β, p66β, the metastasis associated proteins MTA1, -2 and -3, the histone deacetylases HDAC1 and -2, and the retinoblastoma associated protein RbAp48.
Zfhx1b associates with NuRD
The identification of many subunits of NuRD in Zfhx1b complexes made us focus on a role of NuRD as a novel cofactor for Zfhx1b that is possibly involved in Zfhx1b-dependent repression. We therefore probed for Zfhx1b association with endogenous NuRD subunits. For this, we performed a Flag-based purification of extracts of HEK293T cells containing transiently produced 3XFS-Zfhx1b proteins (or only 3XFS tag as control). Western-blot analysis using NuRD subunit-specific antibodies confirmed the association of Zfhx1b with all the subunits identified by MS, with the exception of p66β (Fig. 2A). Also RbAp46 of the NuRD complex co-purified with 3XFS-Zfhx1b. Subsequently, the interaction of endogenous Zfhx1b with endogenous NuRD subunits was analyzed by subjecting extracts of HEK293T cells to immunoprecipitation using anti-Zfhx1b antiserum. WB analysis showed that endogenous Zfhx1b associated with Mi-2β, MTA1, MTA2, HDAC1 and RbAp48 (Fig. 2B). None of the NuRD subunits were detected when extracts were incubated with control antibody. Hence, our results show that Zfhx1b associates with the NuRD complex at normal physiological levels. Similarly, we have also observed interaction between δEF1 and several NuRD subunits (data not shown).
Mi-2β is a specific cofactor for Zfhx1b-mediated transcriptional repression of E-cadherin promoter
Zfhx1b represses E-cadherin transcription in epithelial cells in culture, and an inverse correlation is seen between Zfhx1b and E-cadherin mRNA (as well as their protein levels) in cells of epithelium derived tumors (22). We previously observed a persistence of E-cadherin mRNA and protein in the neuroepithelium of Zfhx1b knockout mouse embryos, which has been linked with the neurological problems seen in MWS patients (2). These observations, together with DNA-binding data in vitro, make the E-cadherin gene a direct target for Zfhx1 proteins (6,10). To address a potential role for NuRD in Zfhx1b-regulated E-cadherin transcription, we included a mutant NuRD component, Mi-2βK750C, in an E-cadherin promoter-based reporter assay. Mi-2βK750C contains a mutation within its ATPase domain and was shown recently to act in a dominant-negative manner with respect to Mi-2β’s involvement in NAB2-mediated repression (29). We found that the level of repression by Zfhx1b of E-cadherin promoter-based reporter constructs was reduced in the presence of Mi-2βK750C, whereas Mi-2βK750C alone did not have a substantial effect on this reporter (Fig. 3A). Overexpression of wild-type Mi-2β restored Zfhx1b-mediated transcriptional repression of the reporter construct (Fig. 3A). 3TPlux, which is another Zfhx1b-sensitive reporter (6), was not affected in the presence of Mi-2βK750C (Fig. 3B). These results demonstrate that Mi-2β can function as a target gene-specific corepressor of Zfhx1b.
A mutant Zfhx1b protein causing atypical MWS is unable to interact with NuRD
To identify the domain of Zfhx1b necessary for association with NuRD, a series of Zfhx1b deletion mutants (from mouse, Fig. 4A) were transiently produced and tested for interaction with MTA2, HDAC2 and RbAp48. As shown in Figure 4B, all peptides containing at least the Zfhx1b N-terminal domain co-precipitated with these subunits, whereas a long Zfhx1b polypeptide (from amino acid 715 till 1214) did not. Hence, the first 289 amino acids of Zfhx1b were found to be sufficient for interaction with the NuRD subunits tested here. An N-terminally 12 amino acid-long motif was recently found to be necessary and sufficient for NuRD recruitment by the transcriptional repressor Sall1 (30). Most notably, the critical residues within this motif that, upon substitution, abolished NuRD binding and decreased transcriptional repression by Sall1, are fully conserved within the Zfhx1b N-terminal domain. This putative NuRD-binding motif is also strongly conserved in Zfhx1b orthologues, including in H. sapiens and X. laevis (Fig. 5A; bold type sequences), as well as in δEF1.
A ZFHX1B mutation disrupting a splice site was recently identified in a patient presenting a mild form of MWS. The resulting protein was found to be identical to ZFHX1B except for the first 24 amino acids, which in this patient were replaced by a heterologous sequence (16). Alignment of wild-type and mutant sequences shows that this atypical ZFHX1B protein (referred to as AZmut) lacks the NuRD interaction motif (Fig. 5A). To test this experimentally, we generated a construct encoding a similar ZFHX1B mutant protein (see Fig. 5B), and probed for interaction with endogenous NuRD subunits. To this end, HEK293T cells extracts containing either transiently produced 3XFS-ZFHX1BWT or 3XFS-ZFHX1BAZmut proteins were subjected to immunoprecipitation with anti-Flag antibodies. Western-blot analysis using NuRD subunit-specific antibodies revealed defective NuRD association by ZFHX1BAZmut proteins, whereas CtBP binding was unaffected (Fig. 5C). Immuofluorescence analysis showed that wild-type as well as mutant ZFHX1B proteins both localize to the nucleus of HeLa cells (data not shown). Together, these findings support the importance of the conserved motif for NuRD recruitment by Zfhx1b protein, and further suggest defective NuRD recruitment by mutant ZFHX1B human protein as the underlying cause of this atypical MWS in this patient (16).
XMi-2β is involved in XZfhx1b-induced neural tissue formation
Zfhx1b plays an essential role in early neural development in the amphibian embryo (4). We therefore also used Xenopus as a model to investigate a potential role for NuRD in XZfhx1b’s neural inducing activity. First, we addressed whether the expression domain of endogenous XMi-2β overlaps with that of XZfhx1b during Xenopus embryogenesis. In situ hybridization analysis showed that XMi-2β mRNA, like XZfhx1b, is initially broadly expressed at gastrula stage, whereas at neurula stage it becomes more strongly expressed in neural tissue and neural crest cells. At tadpole stage, the strong expression of XMi-2β and XZfhx1b mRNA persists in the neural tube and also in neural crest cells and, also similar to mouse Zfhx1b mRNA, in the somites, in line with its documented role in somitogenesis (Fig. 6A) (31). This specific expression pattern thus supports a role for XMi-2β in early neurogenesis, while the strong overlap between XZfhx1b and XMi-2β RNA expression suggests that the encoded proteins may indeed act together in neural development and possibly somitogenesis.
Part of XZfhx1b’s neuralizing activity can be ascribed to the transcriptional repression of BMP4 and BMP-dependent genes (4,25,32). We therefore addressed also the ability of ZFHX1BAZmut proteins to repress expression driven from XBMP4 promoter-based reporter constructs in HEK293T cells. While wild-type ZFHX1B proteins repressed XBMP4 reporter constructs in a concentration-dependent manner, the repression activity of ZFHX1BAZmut proteins on this reporter construct was significantly reduced (Fig. 6B).To further investigate a role for NuRD in XZfhx1b-neuralizing activity, we used a loss-of-function approach involving previously used XMi-2β anti-sense MOs (28). Both the XZfhx1b-dependent repression of BMP4 and the BMP-dependent gene XVent-1, and the Zfhx1b-mediated induction of the neural marker N-CAM were reduced in animal cap explants derived from XMi-2β MOs-injected Xenopus embryos (Fig. 6C). Co-injection of standard control MOs did not reduce XZfhx1b activities to a similar extent. We therefore conclude that endogenous XMi-2β is required for XZfhx1b neuralizing activity. Co-injection of hMi-2β RNA failed to rescue XZfhx1b activity in the animal caps derived from XMi-2β MOs injected embryos (Fig. 6B). Most likely, this is due to the mesoderm inducing activity seen upon Mi-2β overexpression, as described for XMi-2 β recently (28).
Although the defects in neural crest and neurogenesis in conventional Zfhx1b knockout mice helped to explain the developmental defects that relate to certain clinical features of MWS patients, the precise molecular mechanism(s) underlying this congenital disorder is still unknown. Here, we followed a proteomics approach and identified multiple subunits of the NuRD complex as novel nuclear partners for Zfhx1b. We confirmed association of endogenous Zfhx1b with endogenous NuRD, and demonstrated that this interaction is mediated via the N-terminal domain of Zfhx1b, which harbors a recently defined NuRD-interacting motif. NuRD association has been reported for multiple transcriptional repressors, including Ikaros (33), KAP-1 (34), p53 (35), Bcl-6 (36), FOG-1 (37) and BCL11B (38). Previous purifications of the NuRD complex did not identify Zfhx1b proteins (18–21). This suggests that Zfhx1b is not a bona fide component of this complex but rather transiently associates, possibly depending on the cell type or its activation status, with NuRD to regulate transcription of a selective set of Zfhx1b target genes. In addition, previous searches for proteins containing the proposed NuRD-interaction motif yielded neither Zfhx1b nor δEF1 (Zfhx1a) (30,37). The reason may be 2-fold. First, all C2H2 type zinc finger transcriptional repressors that also interact with NuRD, such as Friend of GATA proteins (FOG-1 and FOG-2), Chicken Ovalbumin Upstream Promoter Transcription Factor Interacting Proteins (CTIP1 and CTIP2) and zinc finger protein 423 (Ebfaz), contain the motif at their extreme N-terminal end, whereas in Zfhx1b 11 amino acids unrelated to the proposed motif precede it. Second, the motif is in fact not strictly conserved within Zfhx1b but still contains the critical residues found within Sall1 (30).
Patients presenting with typical MWS all carry either whole gene deletions or a translocation, nonsense mutation or a frame shift mutation in one allele (12). The clear lack of a correlation between the genotype and the clinical features of the patients previously suggested ZFHX1B haplo-insufficiency as the major mechanism causing MWS (39). Here, we present evidence suggesting that defective NuRD recruitment by ZFHX1B is implicated in the etiology of at least a mild form of MWS. First, a ZFHX1B protein, similar to the aberrant ZFHX1B protein in a patient presenting with atypical MWS (16), is unable to interact with NuRD. This aberrant protein in this patient is identical to the wild-type protein from amino acid 25 onwards, and therefore retains both zinc finger clusters required for DNA-binding and all previously known interaction domains for non-NuRD proteins in Zfhx1b (16). Second, we found that the repression of E-cadherin promoter constructs by Zfhx1b is less efficient in the presence of a dominant-negative form of Mi-2β. Third, XZfhx1b fails to efficiently induce a neural program in the absence of endogenous XMi-2β in frog embryos, which might be relevant regarding the neurological problems seen in MWS patients. Fourth, we found that the ability of ZFHX1BAZmut proteins to repress expression driven from a XBMP4 promoter reporter construct is reduced compared to ZFHX1BWT proteins. Given the apparent specificity with which Zfhx1b and NuRD act together to selectively regulate promoters, this may very well explain the milder phenotype. We have shown previously that CtBP-mediated repression is only one of the mechanisms underlying Zfhx1b action, as observed in both Zfhx1b-induced neurogenesis in Xenopus (25) and E-cadherin repression (10). In line with this, multiple repression modes have also been reported for other transcriptional zinc finger repressors that recruit CtBP, including Knirps and FOG (40–42). The AZ patient still contains one functional ZFHX1B allele and the ZFHX1BAZmut protein misses only the first 24 amino acids. Hence, the resulting protein still contains all previously characterized domains, including a functional CtBP interacting domain (also shown in Fig. 5C). It is interesting to note that this patient has characteristic facial appearance but speaks full sentences. This indicates that the patient has also neural crest cell deficiencies, but the condition of mental retardation is quite different from that of typical MWS patients, who have severe mental retardation and limitation of language. Based on the presence of a potential N-myristoylation site within the highly evolutionary conserved N-terminal region affected in the AZmut patient, a critical role of ZFHX1B N-myristoylation in neural crest cell differentiation and migration was previously discussed (16). Our work, however, strongly supports a role for NuRD ZFHX1B association in neural crest cell development. Hence, the production of an aberrant ZFHX1B protein lacking the putative NuRD interacting motif within its N-terminal region but containing all previously characterized functional domains may well explain the clinical condition of the AZmut patient.
Given the evidence of NuRD-Zfhx1b association at physiological levels presented here, and the effect of XMi-2β knock-down on XZfhx1b activity in a process previously used to document Zfhx1b activity, our work strongly supports a role for the NuRD complex as a corepressor for XZfhx1b activity, at least during neural tissue formation. Consistent with this, we previously found that this XZfhx1b activity is sensitive to the HDAC inhibitor Trichostatin A (25). In addition to the very recently discovered role of Mi-2β in the positioning of the mesodermal/neurectodermal boundary by controlling on its turn XZfhx1b transcripts (28), we show here a new aspect of Mi-2β as being a partner of Zfhx1b during early neural amphibian development. Taken together, our study provides new molecular insight into XZfhx1b neural inducing activity, which depends on its N-terminal domain and on interaction with NuRD. Finally, this study, which links a biochemical approach with mutation analysis in MWS patients, suggests that subtle mutations leading to milder or atypical MWS can be informative for the function of individual domains of the multi-domain Zfhx1-type proteins. Such more subtle mutations can in principle be introduced in mouse embryonic stem cells within the Zfhx1b gene itself for testing their effect on embryogenesis or postnatal development.
MATERIALS AND METHODS
The mouse 3XFS-Zfhx1b expression construct was generated by insertion of a strep tag encoding double-stranded oligonucleotide (AGCTTATGGCTAGCTGGAGCCACCCGCAGTTCGAAAAAGGCG) between the HindIII and EcoRI site of p3XFLAG-CMV-7.1 (Sigma), and subcloning of the full open reading frame of mouse Zfhx1b in frame with the N-terminally 3XFS tag (using BglII and EcoRV restriction sites). Constructs encoding Zfhx1b mutants were generated by subcloning the Zfhx1b cDNA fragments (eventually corresponding with amino acids 1–289, 1–317, 1–436, 1–493, 1–714 and 715–1214) in-frame with six Myc tags and the nuclear localization signal of SV40 T-antigen in pCS2-NLS-Myc. The cDNA encoding human ZFHX1B was first subcloned into pPCR-Script and via EcoRV and SmaI restriction digestion inserted into p3XFS (cut with EcoRV). The ZFHX1BAZmut was generated by insertion of selected double-stranded oligonucleotides containing the sequences as described (16). pCDNA3.1-FLAG-MTA2 and -RbAp48 were generated by insertion of the MTA2 and RbAp48 coding regions (pCMV Sport6 MTA2/RbAp48; Open Biosystems) into pCDNA3.1-FLAG. pMT2-HA-HDAC2 was received from S. Schreiber. Constructs used in luciferase assays are the pGL3 wild-type E-cadherin promoter(−178+44)-luciferase reporter (22) and the p3TPLux reporter (23). From the 2 Kb-long XBMP4 (−2000 to +54) promoter-based luciferase reporter (received from W. Knochel), a 488 bp-long promoter fragment was amplified and cloned into pGL4.16. A cytomegalovirus promoter-driven lacZ reporter was used for normalization. pCMVSport6 with inserts of wild-type mouse Mi-2β and mutant mouse Mi-2βK750C cDNA were a kind gift from J. Svaren (Wisconsin-Madison; USA). The pCMV Sport6 hMi-2β was obtained from Open Biosystems. Constructs used in this study that have been described elsewhere are pCS2 with inserts of 6Myc-tagged wild-type mouse (5) and Xenopus Zfhx1b cDNA (24).
Cell culture and transient transfections
HEK293T cells were cultured in DMEM (Invitrogen) with 10% fetal bovine serum. Transient transfections for association studies were done with 25 kDa branched polyethyleneimine (PEI; Aldrich), using 4 µl of 1 mg of PEI/ml of stock solution per µg of DNA, unless stated elsewhere. Reporter assays were done according to van Grunsven et al. (25) using 7.5 × 104 cells per well of a 96-well plate.
For double affinity purification, HEK293T cells were seeded in 9 cm cell culture dishes (2 × 106 cells/dish) and transiently transfected after 24 h with 3XFS-tag or 3XFS-Zfhx1b-encoding plasmids. Cells were harvested 1 day later, washed twice with ice-cold phosphate-buffered saline and incubated for 20 min on ice with lysis buffer (170 mm NaCl, 10 mm EDTA, 50 mm Tris pH 7.4, 50 mm NaF, 0.2 mm dithiothreitol and 0.5% NP-40) supplemented with protease and phosphatase inhibitors. Lysates were collected by centrifugation (21.000 g for 20 min at 4°C). Supernatant was incubated with avidin (Biochemika) (4°C, 1 h) and avidin complexes were collected by centrifugation (20 min, 4°C). Extracts were subjected to affinity purification by overnight incubation with M2 anti-FLAG agarose beads (Sigma) on a rotating wheel at 4°C. Beads were washed and complexes were eluted by competition with 3XFLAG peptides (Sigma) (3 h, 4°C). Extracts were applied to a streptactin-agarose column for a second purification (carried out according to the manufacturer’s protocol; IBA), and precipitated by incubation with two volumes of ice-cold acetone at −20°C. Semi-endogenous immunoprecipitation (IP) was done according to the conditions described for the tandem affinity purification, except for the avidin incubation, which was omitted here. Flag-affinity purified complexes were precipitated using acetone.
Co-IP. Whole cell extracts were precleared for 30 min using Protein G-Sepharose beads (Amersham Biosciences) and precipitated overnight at 4°C on a rotating wheel using antibodies against Myc (Santa Cruz) or Flag (Santa Cruz). Complexes were precipitated with blocked protein G Sepharose beads. For endogenous IP, HEK293T cells were plated in T-175 flasks and precipitation was carried out as described for co-IP. Precleared lysates were precipitated with anti-Zfhx1b rabbit polyclonal antiserum (10) and anti-Flag (Santa Cruz) antibodies. Antibodies used in western blotting detected CtBP (Sigma), Flag (Sigma), HDAC1 (Santa Cruz), HDAC2 (Santa Cruz), Mi2 (Santa Cruz), MTA-1 (Santa Cruz), MTA-2 (Santa Cruz), p66 (Upstate), RbAp46 (Santa Cruz), RbAp48 (Abcam) and Zfhx1b (10), respectively.
Mass spectrometry analysis
Purified Zfhx1b complexes were denatured for 5 min and subjected to gel electrophoresis. Staining of the gel was done with Brilliant Blue G-colloidal concentrate (Sigma), and bands were excised and subjected to MS (26). A nano-scale C18 reversed phase peptide separation system, on-line connected with the ion source of a Q-TOF1 mass spectrometer, was used (Micromass UK Cheshire, UK). Peptides were identified from their fragmentation spectra using the MASCOT search engine and the Swiss-Prot database. Peptides identified by a MASCOT score that exceeded the identity threshold score of MASCOT at the 95% confidence level were considered as positively identified.
Xenopus manipulations and injections
Embryos were obtained from adult frogs by hormone-induced egg-laying and in vitro fertilization using standard methods (27). Capped transcripts were synthesized in vitro using the mMessage mMachine kit (Ambion, Austin, TX, USA) and injected in the animal hemisphere of all blastomeres of four cell stage embryos. Animal cap explants were prepared at late blastula stage (stage 9.5) and cultured in 1× Steinberg medium supplemented with 0.1% bovine serum albumin (Steinberg) until early neurula stage. The XMi-2β and standard control morpholino oligonucleotides (MOs) were obtained from Gene Tools LLC and are described elsewhere (28).
In situ hybridization
Embryos were fixed in MEMFA, stained for β-galactosidase activity with 5-bromo 4-chloro-3-indolyl-β-galactopyranoside (X-Gal) and processed for in situ hybridization using digoxygenin-labeled antisense RNA probes (27). Production of the probe for detection of XZfhx1b transcripts has been described elsewhere (24).
The Qiagen RNeasy mini kit was used for RNA isolation from animal caps at stage 10. All RNA preparations were treated with DNAse I (Qiagen). Real-time RT–PCR was performed on an ABI PRISM 7700 (Applied Biosystems) using Platinum SYBR Green qPCR SuperMix-UDG (Invitrogen). The following primers were used: BMP4 (F) 5′-GAGATTGTCCATTTCCCTTGGC-3′, (R) 5′-TCAGTGGAAAGAAGTCCAGCCG-3′; Histone H4, (F) 5′-CAAGCGCATCTCAGGTCTCAT-3′, (R) 5′-ACCTTGAGAACGCCACGAGT-3′; N-CAM (43); XVent1, (F) 5′-CGATCCAACTCTACAGCCCC-3′, (R) 5′-GGGACTCAGAGATACGGAGGC-3′.
Fund of Scientific Research-Flanders (G.0279.04, G.0288.07, to D.H.); the interuniversity networks of the Federal Prime Minister (IUAP-P5/35 to D.H. and E.B., IUAP-P6/20 to D.H.).
We thank Magda Puype for expert technical assistance and Luc Nelles for valuable discussions. We also thank J. Svaren and W. Knochel for plasmids pCMVSport6 with inserts of wild type mouse Mi-2β and mutant mouse Mi-2βK750C cDNA, and the BMP4 promoter, respectively. G.V. was holder of a IWT predoctoral fellowship and was supported by a grant from the E. Vanderschueren Foundation, awarded by the Flanders League against Cancer. L.A.v.G. was holder of a post-doc position from the Fund of Scientific Research-Flanders and was also supported by a Krediet aan Navorsers project from the same agency. J.S. was holder of a post-doc position awarded by the Fonds de la Recherche Scientifique.
Conflict of Interest statement. The authors declare that they have no financial conflict of interests.