306 (2007) 34–49www.elsevier.com/locate/ydbio

Developmental Biology

XSip1 neuralizing activity involves the co-repressor CtBP and occursthrough BMP dependent and independent mechanisms

Leo A. van Grunsven a,b,1, Vincent Taelman c, Christine Michiels a,b, Griet Verstappen a,b,Jacob Souopgui c, Massimo Nichane c, Emmanuelle Moens c, Karin Opdecamp c,

Jessica Vanhomwegen c, Sadia Kricha c, Danny Huylebroeck a,b,⁎, Eric J. Bellefroid c,⁎

a Department of Developmental Biology (VIB7), Flanders Interuniversity Institute for Biotechnology (VIB) and Laboratory of Molecular Biology (Celgen),Division of Molecular and Developmental Genetics, K.U. Leuven VIB, Campus Gasthuisberg (Bldg. Ond and Nav2, box 812),

Herestraat 49, B-3000 Leuven, Belgiumb Department of Human Genetics, K.U. Leuven VIB, Campus Gasthuisberg (Bldg. Ond and Nav2, box 812), Herestraat 49, B-3000 Leuven, Belgiumc Laboratoire d'Embryologie Moléculaire, Université Libre de Bruxelles, Institut de Biologie et de Médecine Moléculaires (IBMM), rue des Profs,

Jeener et Brachet 12, B-6041 Gosselies, Belgium

Received for publication 1 July 2006; revised 13 February 2007; accepted 16 February 2007Available online 20 March 2007

Abstract

The DNA-binding transcription factor Smad-interacting protein-1 (Sip1) (also named Zfhx1b/ZEB2) plays essential roles in vertebrateembryogenesis. In Xenopus, XSip1 is essential at the gastrula stage for neural tissue formation, but the precise molecular mechanisms that underliethis process have not been fully identified yet. Here we show that XSip1 functions as a transcriptional repressor during neural induction. Weobserved that constitutive activation of BMP signaling prevents neural induction by XSip1 but not the inhibition of several epidermal genes. Weprovide evidence that XSip1 binds directly to the BMP4 proximal promoter and modulates its activity. Finally, by deletion and mutationalanalysis, we show that XSip1 possesses multiple repression domains and that CtBPs contribute to its repression activity. Consistent with this,interference with XCtBP function reduced XSip1 neuralizing activity. These results suggest that Sip1 acts in neural tissue formation through directrepression of BMP4 but that BMP-independent mechanisms are involved as well. Our data also provide the first demonstration of the importanceof CtBP binding in Sip1 transcriptional activity in vivo.© 2007 Elsevier Inc. All rights reserved.

Keywords: BMP4; CtBP; Neural induction; Sip1; Sox2; Zfhx1b; ZEB2

Introduction

The vertebrate neural plate arises from the embryonic ecto-derm during gastrulation. Molecular studies have revealed that

⁎ Corresponding authors. D. Huylebroeck is to be contacted at Department ofDevelopmental Biology (VIB7), Flanders Interuniversity Institute for Biotech-nology (VIB) and Laboratory of Molecular Biology (Celgen), Division ofMolecular and Developmental Genetics, K.U. Leuven VIB, Campus Gasthuisberg(Bldg. Ond andNav2, box 812), Herestraat 49, B-3000 Leuven, Belgium. Fax: +3216 345933. E. J. Bellefroid, fax: +32 2 650 9733.

E-mail addresses: danny.huylebroeck@med.kuleuven.be (D. Huylebroeck),ebellefr@ulb.ac.be (E.J. Bellefroid).1 Present address: Cell Biology and Histology (CYTO), Faculty of Medicine

and Pharmacy, Vrije Universiteit Brussel, Laarbeeklaan 103, B-1090 Jette,Belgium.

0012-1606/$ - see front matter © 2007 Elsevier Inc. All rights reserved.doi:10.1016/j.ydbio.2007.02.045

inhibition of BMP signaling in the ectoderm is required forneural fate acquisition. This BMP inhibition occurs throughdistinct mechanisms, including inhibition of BMP ligands byvarious secreted ligand-binding proteins, inhibition of transcrip-tion of the BMP gene itself, and negative modulation ofintracellular signaling by Smad proteins (Munoz-Sanjuan et al.,2002; De Robertis and Kuroda, 2004; Linker and Stern, 2004;Stern, 2005). Downstream of neural induction, the intracellularcomponents that establish neural cell fate are not wellcharacterized. Several neural effectors have been identifiedsuch as Geminin (Kroll et al., 1998), Zic1 to 3 (Nakata et al.,1997; Kuo et al., 1998), Sox1 to 3 (Penzel et al., 1997; Kishi etal., 2000; Bylund et al., 2003; Graham et al., 2003), SoxD(Mizuseki et al., 1998a,b), and the Smad-binding protein Sip1

35L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

(Eisaki et al., 2000; van Grunsven et al., 2000; Sheng et al.,2003).

Sip1 (Zfhx1b/Zeb2) belongs to the Zfhx1 family of multi-domain transcriptional repressors characterized by a homeo-domain-like domain and by two zinc finger clusters each ofwhich binds with high affinity to CACCTG and CACANNTGbinding sites and can form complexes with Smads (Remacleet al., 1999; Verschueren et al., 1999), the co-repressor CtBP(C-terminal binding protein) (Postigo and Dean, 2000; vanGrunsven et al., 2003), and the co-activators p300 andpCAF (p300/CBP associated factor) (van Grunsven et al.,2006).

In embryos of Xenopus, chick and mouse, Sip1 mRNA isdetected at gastrula in the prospective neurectoderm (Eisaki etal., 2000; van Grunsven et al., 2000; Sheng et al., 2003; Van dePutte et al., 2003). Homozygotic deletion of Sip1 in the mouseis embryonic lethal and the embryos show severe neural crestcell defects and fail to generate or maintain intact neural

Fig. 1. XSip1 acts as a repressor during neural differentiation of the ectoderm. Whole-14). Embryos were injected at the four-cell stage with the indicated RNA. (A–D) Injdoes not induce Sox2 in animal caps. (E, F) Embryos injected with 250 pg of XSip1performed to reveal distribution of the injected RNA. Note the expansion of Sox2 exreduction in the XSip1-VP16 injected embryo. (G–I) Animal caps derived from em250 pg of XSip1-VP16 RNA. XSip1-VP16 prevents induction by XSip1 or tBR of So0%, n=44; (E) 80% embryos with expanded Sox2, n=18; (F) 100% embryos with d

ectoderm (Van de Putte et al., 2003). In human, mutations inZFHX1B exon coding sequences, most of which cause C-terminal truncation of the protein, lead to Mowat–WilsonSyndrome (Mowat et al., 2003).

Although Sip1 has been documented primarily as atranscriptional repressor (Verschueren et al., 1999; Comijn etal., 2001; van Grunsven et al., 2003; Vandewalle et al., 2005), itcan also act as a transcriptional activator in vivo (Long et al.,2005; Yoshimoto et al., 2005). Overexpression of Sip1 incertain epithelial cells induces epithelial to mesenchymaltransition by directly repressing E-cadherin (Comijn et al.,2001) and other genes coding for crucial proteins of epithelialcell–cell junctions (Vandewalle et al., 2005). In animal capexplants of Xenopus early embryos, ectopic synthesis of XSip1induces neural specific gene expression and represses in theembryo the expression of the panmesodermal marker genebrachyury (Xbra), of BMP4 and of other genes in thepresumptive epidermis (Eisaki et al., 2000; Lerchner et al.,

mount in situ analysis of Sox2 expression in embryos (St. 11) or animal caps (St.ection of 250 pg of XSip1-VP16, in contrast to wild-type XSip1 and XSip1-EnR,-VP16 or XSip1-EnR RNA. LacZ RNAwas co-injected and X-gal staining waspression on the injected area in the XSip1-EnR injected embryo (arrow) and thebryos co-injected with 50 pg of XSip1 RNA or 200 pg tBR RNA together withx2. Respective inductions (A) 100%, n=28; (B) 0%, n=29; (C) 90%, n=30; (D)ownregulation of Sox2, n=40; (G) 0%, n=35; (H) 100%, n=26; (I) 0%, n=35.

36 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

2000; Papin et al., 2002; van Grunsven et al., 2006). In addition,XSip1 knockdown studies have been performed indicating that,like in the mouse, Sip1 is essential for neural differentiation(Nitta et al., 2004).

Despite the demonstrated importance of Sip1 in neural tissueformation, the underlying molecular mechanism(s) by which itcontributes to neural differentiation remain unknown. Inparticular, the in vivo significance of the CtBP interaction inSip1-neuralizing activity has not been demonstrated (vanGrunsven et al., 2003). Here we show first that XSip1 functionsas a repressor, blocking directly the transcription of the BMP4gene. Besides BMP4, XSip1 also represses the transcription ofother epidermal genes in an BMP-independent manner in thedeep and superficial layer of the ectoderm to maintain a neuralfate. Secondly, we demonstrate for the first time the significanceof CtBP interaction for Sip1 activity.

Experimental procedures

Plasmid constructs

All point mutations and deletion mutants of XSip1 were generated bystandard PCR and cloning techniques and their sequences verified. Deletionmutations of XSip1 were cloned in-frame with six Myc tags and the SV40 T-antigen nuclear localization signal (NLS) in pCS2-NLS-Myc. Full-lengthcDNAs with point mutations were cloned into pCS2-Myc. Amino acid changesor deletions are depicted in the respective figures except for XSip13xCtBPmut inwhich the 3 CtBP sites (i.e. PLNLS786–790, PLDLT816–820, and PLNLT860–864)were all mutated into AAALS or AAALT. For the XSip1SBDmut, we replaced thenucleotides coding for amino acids 467 to 488 by insertion of a linker resultingin AAA in XSip1SBDmut. For the generation of pCS2 NLS Myc XSip1-EnR,nucleotides corresponding to XSip1 (aa 205–1082) were taken from pCS2 NLSXSip1205–1082 and ligated into the StuI restriction site of pCS2 NLS EnR. pCS2Flag mCtBP2 was constructed by insertion of CtBP2, from pCDNA3 FlagmCtBP2 (van Grunsven et al., 2003), into the StuI restriction site of pCS2. pCS2VP16-mCtBP2 was generated by cloning the full open reading frame ofmCtBP2 in frame with VP16 in the pAct vector (Promega) and subsequentsubcloning the entire VP16-mCtBP2 coding region into the StuI site of pCS2.The XSip1 NZF-CtBP and NZF-GFP expression constructs were obtained bycloning the full open reading frames of mCtBP2 and eGFP (Invitrogen),respectively, in-frame with the NZF of XSip1-NZF(XSip1198–344 in Fig. 4A).Previously described expression constructs used here are: CA-Alk3 (Onichtch-ouk et al., 1999), tBR (Suzuki et al., 1994), XSip1 (van Grunsven et al., 2000),mSip1-CZFmut-CID and mCtBP2 (van Grunsven et al., 2003), and XSip1-VP16(Papin et al., 2002). For the CZFmut-CIDmut, the CID domain in the mSip1-CZFmut-CID construct was replaced by a CID domain with mutated CtBPbinding sites (van Grunsven et al., 2003).

Xenopus embryo manipulations and injections

Xenopus embryos were obtained from adult frogs by hormone inducedegg-laying and in vitro fertilization, using standard methods (Sive et al.,

Fig. 2. Effect of BMP signaling on XSip1 activity. (A–I) Animal caps derived from(A), XSip1 RNA (B, C, E–I) or XSip1ΔSBD mRNA (D) with or without CA-Alk3 RNprobes. For each marker, control non-injected animal caps are shown on the left. Nneurula stage (dorsal view, anterior right) are shown on the right. Note that CA-Alk3 band the repression of Gata2 expression (E). In contrast, CA-Alk3 does not affect XSiHya-1 and Vgl-4 (E–I). (J) Lateral views of embryos injected with XSip1 RNA aloneof CA-Alk3 does not affect XSip1 repression of epidermal keratin. Respective inductiinhibited, n=18; (C) all inhibited, n=52; (D) all inhibited, n=36; (E) none inhibited,(J) all inhibited, n=33. Arrows in panels E–J indicate the injected area.

2000), and staged according to Nieuwkoop and Faber (1997). Cappedtranscripts were synthesized in vitro by using the mMessage mMachine kit(Ambion, Austin, TX). RNA or antisense morpholino oligonucleotides (20/blastomere of CtBP MO; 5′-GCCTCTTCACTTTGTGTTTATCCAT-3′) (GeneTools LLC) were injected as described and at the indicated stage. Syntheticnuc-LacZ RNA (from 100 to 500 pg), encoding a nuclear form of β-galactosidase, was used as a lineage tracer. Animal cap explants wereprepared at late blastula (stage 9.5) and cultured in 1× Steinberg mediumsupplemented with 0.1% BSA (Steinberg–BSA) until neurula stage.Trichostatin A (33 mM) was diluted 1:66,200 in 1× Steinberg–BSA for usein the animal cap explants.

RT-PCR

The Qiagen Rneasy mini kit was used for RNA isolation from embryos atdifferent developmental stages. All RNA preparations were treated with DNAse I(Qiagen) and checked with 32 cycles of Histone H4 for DNA contamination. RT-PCRwas carried out as described in the GeneAmpRNAPCRkit (Perkin-Elmer).

The following primers were used:

Histone H4, (F) 5′-CGGGATAACATTCAGGGTATCACT-3′, (R) 5′-ATCCATGGCGGTAACTGTCTTCCT-3′; 56 °C, 25 cycles; XSip1, (F) 5′-ATGCAGCACTTAGGTGTAGGGATGG-3′, (R) 5′-GTTGATGCAATG-GAATTGGAC; CAGG-3′; 57 °C, 35 cycles; XCtBP, (F) 5′-AGTGAAGAGGCAACGTTTGG-3′, (R) 5′-CGAGAAAGTGT-GATTGTGTG-3′; 57 °C, 33 cycles; XCtBP1, (F) 5′-GTTCTCACTTGC-TAAACAAGG-3′, (R) 5′-TGAACTTCTCCAGACTTCCC-3′; 57 °C, 33cycles; BMP4, (F) 5′-GAGGAGCATTTGGAGAATCTACCAA-3′, (R) 5′CATAATTGCAGGGCTTACATCAA; 56 °C, 25 cycles; Sox2, (F) 5′-GAGGATGGACACTTATGCCCAC-3′, (R) 5′-GGACATGCTGTAGG-TAGGCGA-3′; 56 °C, 25 cycles.

PCR products were separated on 2% agarose gels and visualized by stainingwith ethidium bromide.

RNase protection assay

RNAwas prepared with the phenol/NETS method and analyzed using 32P-labeled probes as previously described (Lahaye et al., 2002). RNase protectionprobe for Ep. Keratin was generated by linearizing pGEM81B plasmid (Jonas etal., 1985) by NarI and transcribing it with Sp6. The protected fragment is 270bases long. A 230 bp Sox2 probe was obtained by linearizing a pBS SK(−)-Sox2plasmid (Mizuseki et al., 1998a,b) by NcoI and transcribing it with T7. An rFGFantisense probe (Ryan et al., 1998) was used as a loading control.

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 insitu hybridization using digoxygenin-labeled antisense RNA probes (Sive et al.,2000). To produce the XCtBP probe, the cDNA was linearized by SmaI andtranscribed with T7 (pExpress XCtBP obtained from Biocat GmBH; Genbankaccession no. BC076800). For the XCtBP1 probe the cDNAwas linearized withSmaI and transcribed with T7 (pCMV Sport6 XCtBP1 obtained from BiocatGmBH; Genbank accession no. AF091554). Plasmids used for generating theother probes for in situ hybridization were as described previously: Xbra (Smith

embryos injected at the four cell-stage (250 pg/blastomere) with Geminin RNAA (100 pg) analyzed at neurula stage by in situ hybridization with the indicatedon-injected control embryos and CA-Alk3 RNA-injected (100 pg) embryos atlocks XSip1's ability to induce the neuronal markers Sox2 and NCAM (B and C)p1's ability to block several other epidermal genes like epidermal keratin, TA-2,or together with CA-Alk3 RNA and stained for epidermal keratin. Co-expressionons/inhibitions in “+CA-Alk3 caps and embryos”: (A) all positive, n=35; (B) alln=40; (F) all inhibited, n=22; (G) all inhibited, n=25; (H) all inhibited, n=28;

37L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

38 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

et al., 1991), Ep. Keratin (Jonas et al., 1985), BMP4 (Schmidt et al., 1995),Gata2 (Walmsley et al., 1994), Sox2 (Kishi et al., 2000), Sox3 (Penzel et al.,1997), NCAM (Krieg et al., 1989), XSip1 (van Grunsven et al., 2000), Grainy-head (Tao et al., 2005), TA-2, Hayluronan synthase 1 and vestigial like 4(Chalmers et al., 2002). For sections, after completion of the whole-mountprocedure, the embryos were gelatin-embedded and vibratome-sectioned at30 μm thickness.

Cell culture, transient transfections and immunoprecipitations

For BMP4 reporter assays, HEK293T cells were cultured in DMEMsupplemented with 10% fetal clone (Invitrogen). Transient reverse transfectionswere carried out in 96-well plates (5×104 cells/well) using Fugene6 and a totalof 100 ng DNA per well. Two nanograms of a LacZ reporter construct thatcontains the cytomegalovirus promoter inserted upstream of E. coli LacZ wascotransfected for normalization. Cell extracts were prepared and assayed using aLuminoskan Ascent (Labsystems) for luciferase and β-galactosidase activityaccording to the manufacturers' protocols (Luciferase assay system (Promega)and Galscreen™ (Tropix)), and only transfections with similar β-galactosidasevalues were taken into account. The data, obtained in triplicate experiments,were then normalized by calculating the ratio of luciferase and β-galactosidaseactivities. Each experiment was repeated at least 5 times. For Western analysis,one fifth of the cell lysate was run on an 8% SDS–PAGE gel followed byWestern detection of the produced proteins by means of anti-Myc antibodies andsubsequent visualization using the Western lightning detection system (Perkin-Elmer). Immunoprecipitation of Sip1–CtBP complexes from transfectedHEK293T cells was carried out as described previously (van Grunsven et al.,2006).

Electrophoretic mobility shift assays

In vitro-translated proteins were prepared using the TNT coupledtranscription–translation system (Promega). Gel mobility shift assays werecarried out with PCR products obtained using the 2 kb XBMP4 promoter(Genbank accession no. XLAJ5076) as a template and the following primers:Probe 1 (F) 5′-GCCCCAATGGCCCTTTAA-3′, (R) 5′-CGTGAAA-CACCTGGCATGTTT-3′. Probe 2 (F) 5′-CGAAGGCTGAGGTGAGCAA-3′,(R) 5′-TGGTCGTGCCAATCCTATT CT-3′; Probe 3 (F) 5′-GGAATG-TGTGTCCAGTAGCTGATT-3′, (R) 5′-TGTAAAGATCAACAATACTGT-GAATAAATTG-3′. The PCR products were labeled using T4 DNA kinase inthe presence of [γ-32P] dATP. Protein–DNA complexes were formed byincubation of the proteins with 50,000 CPM (about 0.1 ng) of radioactivelylabeled DNA for 20 min at 24 °C in 20 μl of binding buffer (10 mM Tris–HClpH 7.5, 150 mM KCl, 1 mM DTT, 10% glycerol, 0.1% NP40, 0.2 mg BSA/mland 0.5 μg polydI–dC/ml). DNA–protein complexes were resolved byelectrophoresis on a 4% acrylamide gel in 0.5× TBE buffer. The dried gelwas exposed to X-ray film overnight at −80 °C with an intensifying screen.

Results

XSip1 acts as a repressor during neural differentiation

Previous studies showed that Sip1 can act both as a repressorand an activator of transcription (Verschueren et al., 1999; Comijnet al., 2001; Long et al., 2005; Yoshimoto et al., 2005; vanGrunsven et al., 2006). To determine the transcriptional activity ofXSip1 responsible for neural induction in Xenopus embryos, weused chimeric proteins in which XSip1 (as full-length protein or apolypeptide containing both zinc finger clusters) is fused to theherpes simplex virus VP16 activator domain (Papin et al., 2002)or the Drosophila Engrailed repressor domain (Jaynes andO'Farrell, 1991). The neural-inducing activities of the VP16activator fusion (XSip1-VP16) and the Engrailed repressor fusionprotein (XSip1-EnR) were examined by overexpression in animal

cap explants and in embryos. In animal caps, overproduction ofXSip1 and XSip1-EnR induced the expression of the early neuralmarker Sox2, whereas XSip1-VP16 did not have this effect (Figs.1A–D). Similar results were obtained using a later neural marker,NCAM (data not shown). In embryos, overproduction of XSip1-EnR, like XSip1 (van Grunsven et al., 2006), expanded Sox2staining on the injected side whereas XSip1-VP16 inhibited Sox2expression (Figs. 1E, F). Previously, XSip1-VP16 has beenshown to induce the expression of the mesodermal marker Xbrain animal caps (Papin et al., 2002) using RNAase protectionassays. Since this could explain the inhibition of Sox2 weobserved in embryos, we re-examined XSip1-VP16 injectedembryos and animal cap explants for Xbra expression. Surpris-ingly, we found that XSip1-VP16 does not induce Xbraexpression, as detected by in situ hybridization or RT-PCR.Moreover, in embryos, Xbra expression was inhibited by XSip1-VP16 (data not shown) as is the case for overexpression of thenative protein (Papin et al., 2002; van Grunsven et al., 2006).Thus, the downregulation of Sox2 expression by XSip1-VP16might not be due to conversion of ectoderm into mesoderm andmay reflect the upregulation of putative downstream targets ofXSip1 that are normally inhibited during neural tissue formation.

To assess the potential inhibitory effect of XSip1-VP16 inneural differentiation, XSip1-VP16 was co-expressed withnative XSip1 or a dominant negative BMP receptor (tBR)(Suzuki et al., 1994) and the induction of Sox2 was examined.As expected, both XSip1 and tBR induced Sox2 (Figs. 1A, H),while this response was fully inhibited by XSip1-VP16 (Figs.1G, I). Taken together, the similarity of the effects of XSip1 andXSip1-EnR and the opposite effects of the XSip1-VP16construct suggests that XSip1 is likely to function as a repressorto induce neural differentiation. Our data also extend previousobservations indicating that XSip1 has an essential role in earlyneural gene expression (Nitta et al., 2004).

XSip1 neuralizes ectoderm through BMP inhibition but alsohas a BMP independent function in the suppressionof epidermal fate

In animal caps cut from XSip1 overexpressing embryos, theneural markers Sox2, Sox3 and NCAM are induced, whereasepidermal markers and BMP4 expression are suppressed (Eisakiet al., 2000; Postigo et al., 2003; Nitta et al., 2004; vanGrunsven et al., 2006). These observations are consistent with amechanism in which downregulation of BMP4 RNA levels isthe primary means by which XSip1 induces neural tissue. If so,co-injection of CA-Alk3 RNA, encoding a constitutively activeBMP type I receptor, should block neuralization and rescueepidermalization. In our experiments, we injected 100 pg ofCA-Alk3 per blastomere into 2 cell-stage embryos, an amountthat was sufficient to induce ectopic staining within the neuralplate of all epidermal markers tested and that could relievesuppression of epidermal development induced by overexpres-sion of other epidermal inhibitors such as geminin (Kroll et al.,1998) (Fig. 2A and right panels in E–I). Higher doses were alsotested but could not be used in our assays as they were found forunknown reasons to inhibit epidermal gene expression (data not

39L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

shown). We found that co-injection of CA-Alk3 RNA withXSip1 RNA blocked XSip1's ability to induce Sox2 and NCAMin animal cap explants (Figs. 2B, C and data not shown).Interestingly, co-expression of CA-Alk3 with a XSip1 mutantunable to bind to Smads (XSip1ΔSBD) also blocked inductionof Sox2 (Fig. 2D). This suggests that the inhibition observed isnot due to direct interaction of BMP-activated phosphorylatedSmads with XSip1 but probably involves indirect convergenceof Smad signals and XSip1 on target gene promoters. Co-

Fig. 3. Repression of the BMP4 promoter by Sip1. (A) Schematic representation of thnumber XLAJ5076). Putative Sip1 binding sites are indicated by vertical stripes. Thebase pairs of exon1 of XBMP4 are represented by a dark gray box. (B) The N-termpromoter. In vitro transcribed-translated proteins were incubated with the indicatedindicate the XSip1 specific shift in migration of the labeled probes. The shift is notXBMP4 fragments, but not by XBMP4 fragments with mutated CACCT sites. (Canalyzed 40 h later for the XBMP4 promoter-driven luciferase activity. β-Galactosmanner the activity of the indicated 2 kb XBMP4 promoter-luciferase reporter constreporter construct. Control Western blots for Myc-Sip1 protein content in the transfe

injection of CA-Alk3 RNAwith XSip1 RNA affected the abilityof XSip1 to repress Gata2, a marker of both superficial anddeep layers of the ventral ectoderm, but did not block its abilityto repress other epidermal genes, such as Epidermal keratin,Grainyhead-like-1 (Tao et al., 2005) and TA-2 that areselectively expressed in the superficial layer and Vestigial like4 (Vgl-4) and Hyaluronan synthase 1 (Hya-1) (Chalmers et al.,2002), which are deep cell specific markers (Figs. 2E–I and datanot shown). Similar results were obtained in embryos (Fig. 2J

e BMP4 promoter based reporter construct used in this study (Genbank accessionprobes used in panel B are represented by numbered thick horizontal stripes. 54inal zinc finger cluster of XSip1 can bind to upstream regions of the XBMP4probes and the complexes were run on a polyacrylamide gel. The arrowheadsobserved using another protein, Su(H), and it is blocked by an excess of cold) The indicated plasmids were transfected into HEK293T cells and the cellsidase corrected luciferase values are given. Sip1 represses in a dose-dependentruct. A Sip1 double zinc finger mutant (dbl ZFmut) does not regulate the BMP4cted cells used for the luciferase assays are shown.

40 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

and data not shown). These results indicate that XSip1's abilityto repress BMP signaling is essential for its neuralizingproperties but that XSip1 also has a BMP independent functionin epidermal inhibition.

Repression of the BMP4 promoter by XSip1

BMP genes, including BMP4, are expressed in a patterncomplementary to the emerging neural plate in gastrula andneurula embryos (Hemmati-Brivanlou and Thomsen, 1995). Todetermine whether XSip1 could regulate BMP4 transcriptiondirectly, we examined the upstream regulatory region of theXenopus BMP4 gene. Several CACCT(G) type Sip1 bindingsequences were identified, including in the region that has beenshown to be essential for transcriptional activation of BMP4(Metz et al., 1998) (Fig. 3A). We first tested in electro mobilityshift assays (EMSA) whether XSip1 could bind in vitro to theCACCT(G) sequences located in the −488 to +54 bp fragmentof the promoter (the three probes used are indicated by numbersin Fig. 3A). Fig. 3B shows that a XSip1 truncated protein (withthe NZF zinc fingers) that retains high-affinity and sequencespecific DNA-binding in vitro (Remacle et al., 1999) binds toeach of these probes in a specific manner. Moreover, no bindingto the BMP4 promoter fragments is observed with anotherDNA-binding protein, Suppressor of hairless (Su(H)).

To further assess the relevance of the XSip1 binding sitesidentified in the BMP4 promoter, we tested in mammalian cellswhether a luciferase reporter gene under the control of a 2 kb-long XBMP4 promoter fragment is regulated by Sip1. For theseassays, we used mouse Sip1 expression vectors (Remacle et al.,1999; Verschueren et al., 1999; van Grunsven et al., 2003) thatare conceptually identical to the wild-type and mutant XSip1constructs used for the Xenopus studies (see below). A dose-dependent decrease of the activity of the BMP4 promoter regionwas observed when plasmids encoding Myc-tagged Sip1proteins were co-transfected with the BMP4-promoter-basedreporter plasmid. No such decrease was observed using adouble zinc finger Sip1 mutant (Fig. 3C) incapable of binding toDNA. Similar results were obtained using a shorter (−488 bp)BMP4 promoter fragment (data not shown). These data suggestthat Sip1 repression of BMP4 transcription is direct.

Fig. 4. Mutational analysis of XSip1 reveals multiple regions, including the CtBrepresentation of XSip1 wild-type protein and mutants used in this study. The amino-Smad-binding domain (SBD; black box), homeodomain-like domain (HD; dotted boxblack stripes) are shown. Embryos injected unilaterally at the two-cell stage (250 pg)hybridization for epidermal keratin expression. For animal cap assays, embryos wereanalyzed at neurula stage by in situ hybridization or RNase protection for Ep. keratkeratin expression and to induce Sox2 was evaluated based on the RNase protection dembryos were used for each condition. Note that, with the exception of the XSip1dbl Zthe epidermal keratin repression assay. In contrast, in the Sox2 induction assay, only thmutants show some Sox2 induction activity, XSip1205–1082 and XSip1CtBPmut. (overproducing Myc-tagged XSip1 wild-type and several mutant proteins. (C) RNAafrom embryos overeproducing wild-type and several XSip1 mutants. Note that, compEp. keratin repression activity. XSip1198–344 and XSip1993–1082 mutants correspondhybridization analysis of Sox2 and Ep. keratin expression in animal caps (top panelEmbryos are viewed laterally. Note that, in contrast to XSip1ΔSBD, which functions ascaps and XSip1dbl ZFmut was inactive. Both XSip1ΔSBD and XSip13xCtBPmut, as judgepanel D indicate the injected area.

XSip1 possesses multiple repression domains that are requiredfor efficient neural induction including the CtBP interactionmotif

Although XSip1 binds to Smads and CtBPs (van Grunsven etal., 2000, 2006), it has not been addressed whether their bindingaffects its ability to regulate transcription of target genes duringneural induction. A full-length Sip1 mutant that cannot bind toCtBP (through mutation of all three binding sites for CtBP) canstill repress endogenous E-cadherin expression in dog kidneycells (MDCK), suggesting that Sip1 represses target genesthrough the recruitment of other co-repressors as well (vanGrunsven et al., 2003). To identify the domains of XSip1necessary for its neural inducing capacities, we made a series ofXSip1 deletion mutants (Fig. 4A). In addition, we generatedmutants that no longer interact with Smads (XSip1 Δ466–488,the XSip1ΔSBDmut) or CtBP (XSip13xCtBPmut). The correspond-ing mouse mutant proteins have been previously characterized(Verschueren et al., 1999; van Grunsven et al., 2003) (see alsoSupplementary Fig. 1). Those tagged point mutant proteins weresynthesized at levels similar to that of the full-length wild-typeXSip1 (Fig. 4B). The XSip1ΔSBD, XSip13xCtBPmut and thedifferent XSip1 deletion mutants were tested for their ability toblock epidermal keratin expression in embryos by in situhybridization and in animal caps by RNase protection. Theirability to induce Sox2 expression was tested in animal capexplants by in situ hybridization and RNase protection. Asshown in Figs. 4A–D, binding to DNA is required for XSip1activity since the XSip1dbl ZFmut, containing mutations in thethird zinc finger of the NZF and CZF cluster (H299S/H1072S,which are mutations known to lead to loss of high-affinity DNAbinding of mouse Sip1 in vitro) (Remacle et al., 1999), failed toinduce Sox2 and to repress epidermal keratin. The zinc fingerclusters alone (XSip1198–344 and XSip1993–1082) were neithersufficient to induce Sox2, nor to block epidermal keratin,indicating that XSip1's activities do not depend merely on theoccupation of DNA sequences that can be bound by transcrip-tional activators. Removal of the N-terminal and C-terminalnon-finger sequences (XSip1205–1082) reduced XSip1's ability toinduce Sox2, indicating that one or both of those regionscontribute to XSip1 activity. Although less active than the native

P interaction domain, involved in XSip1 repression activity. (A) Schematicterminal and carboxy-terminal zinc finger domains (NZF and CZF; gray boxes),), C3H type zinc finger (light gray boxes) and the CtBP interacting domain (CID:overproducing wild-type XSip1 or mutants were tested at neurula stage by in situinjected at the four-cell stage in each cell (250 pg/blastomere). Animal caps werein and Sox2 expression. The ability of the different XSip1 mutants to block Ep.ata: (+) fully active, (+/−) partially active, and (−) not active. A minimum of 30

Fmut, and NZF or CZF alone mutants, all mutants are active, fully or partially, ine XSip1 wild-type and XSip1ΔSBD polypeptides are fully active. Only two otherB) Control Western blot with extracts prepared from injected animal capsse protection analysis of Sox2 and Ep. keratin expression in animal caps derivedared to the wild-type protein, XSip1205–1082 shows reduced Sox2 induction anding to the first and second zinc finger domains alone are not active. (D) In situs) and embryos (bottom panels) that overproduce the indicated XSip1 proteins.the native XSip1 protein, XSip13xCtBPmut induced less efficiently Sox2 in animald by in situ hybridization, retain epidermal keratin repression activity. Arrows in

41L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

XSip1 protein, this XSip1205–1082 mutant was still able to induceSox2 expression and to repress epidermal keratin, indicating thatthe inter-zinc finger domain also plays a role in XSip1 repressionactivities (Fig. 4C). Other deletion mutant proteins tested,retaining only one of the two zinc finger clusters and flankingamino or carboxy-terminal sequences, failed to neuralize ecto-derm (i.e. XSip11–333 and XSip1993–1213, XSip1469–1213 andXSip11–468) which may be due to the fact that both zinc finger

clusters might be required for high-affinity binding to bipartitetarget sequences and thus for full repressor activity (Remacleet al., 1999). Although unable to neuralize ectoderm, all thosemutants still had, at least to some extent, epidermal keratinrepression activity, suggesting that the non finger portionpresent in those mutants also contribute to XSip1 repressionactivities. Interestingly, the XSip13xCtBPmut was also less activein our Sox2 and epidermal keratin assays while the SBD

42 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

mutant showed similar activities as the wild-type protein inboth assays (Figs. 4D and 7B).

As XSip1 and several of the mutants tested could inhibitepidermal keratin expression without inducing Sox2, we askedwhether cells overexpressing XSip1, or some deletion mutantsshowing reduced repression activity (such as XSip13xCtBPmut),express neural crest or placodal markers that are known to form

Fig. 5. Comparison of XSip1, XCtBP, and XCtBP1 expression during embryogenesis.embryonic stages. Histone H4 was used as a loading control. Note that both XCtBP aaccumulate at the start of zygotic transcription. (B) Whole-mount in situ hybridizatiowith anterior toward the right. (C) XCtBP expression in a horizontal section of a stagesensorial layer of the ectoderm. Transversal sections of the neural tube at the level ofXCtBP1 have similar expression patterns during early embryonic development anexpressed earlier and at a higher level than XCtBP1. Later, XCtBP and XCtBP1 havemz, marginal zone; vz, ventricular zone; pn, pronephros; ba, branchial arches; nc,ectoderm. St. 11 embryos: dorso-vegetal view, anterior right; St. 13 embryos: dorsa

under conditions of intermediate levels of BMP inhibition(Marchant et al., 1998). We found that ectodermal cells locatedat a distance from the neural plate in embryos that were injectedwith 250 pg of XSip1 or XSip13xCtBPmut RNA per blastomeredid not adopt a placodal or neural crest identity (see Sup-plementary Fig. 2). This suggests that BMP inhibition inducedby overexpression of those two proteins was effective enough to

(A) RT-PCR analysis of XCtBP, XCtBP1, and XSip1 expression at the indicatednd XCtBP1 RNA are detected throughout development, while XSip1 transcriptsn of XCtBP, XCtBP1, and XSip1 at the indicated stages. All embryos are shown18 embryo with a blow-up showing the restricted XCtBP expression in the deepthe hindbrain of stage 31 embryos are shown at the right. Note that XCtBP andd are co-expressed with XSip1 in the developing neural tissue, XCtBP beingdistinct expression domains within the developing neural tube. Abbreviations:

neural crest; nt, neural tube; pm, posterior mesoderm; sl, sensorial layer of thel view, anterior right; St. 25 and 31 embryos: lateral view, anterior right.

43L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

prevent both epidermis and neural crest and placode cell fatesfrom forming. The nature of these XSip1 expressing cells thatare neither epidermal nor neural or neural plate border cellsremains unknown. Together, these data indicate that XSip1possesses multiple repression domains, which when combinedallow it to exceed a critical threshold of repression on targetgenes necessary for neural induction, and that CtBPs, but notSmads, contribute to its repression activity.

Temporal and spatial expression patterns of XCtBP andXCtBP1

The above results suggest that intact CtBP binding sites areimportant for XSip1 to induce Sox2. For this to occur, the genesencoding CtBP and XSip1 have to be co-expressed duringneural induction. Two CtBP genes have been identified in Xe-nopus, designated XCtBP1 (Sewalt et al., 1999) and XCtBP,which is likely to be the homologue of mouse and humanCtBP2 (Brannon et al., 1999). XCtBP is expressed throughoutXenopus development, but the XCtBP1 expression profile hasnot been mapped. We determined whether both CtBP expres-sion profiles overlap with that of XSip1 during early embryo-genesis. Using RT-PCR, we found that both XCtBP andXCtBP1 transcripts are detected maternally and after the startof zygotic transcription. At all stages examined, XCtBP isexpressed at a much higher level than XCtBP1. As previouslyreported, XSip1 transcripts accumulate after mid-blastulatransition (van Grunsven et al., 2000, 2006) (Fig. 5A). In situhybridization shows that XSip1 RNA expression in gastrula andneurula is restricted to the prospective neurectoderm whileXCtBP is broadly expressed in the entire ectoderm, thestrongest signal being detected in the anterior neural plate.XCtBP transcripts are also seen from neurula stage in theposterior mesoderm. Sections revealed that, like XSip1 (vanGrunsven et al., 2000), early expression of XCtBP within theectoderm is restricted to the deep sensorial layer (Fig. 5C, St.18). XCtBP1 transcripts are not detectable by in situ hybridiza-tion during gastrulation and are only weakly detectable in theanterior portion of the neural plate in the eye field at neurulastage (Fig. 5B, St. 11 and 13). At tailbud stage, both XCtBP andXCtBP1 are expressed, like XSip1, in the developing neuraltube, in the eye and migratory cranial neural crest cells (Fig. 5B,St. 25). At tadpole stage, sections revealed that, while XSip1 isno longer expressed in the neural tube, XCtBP and XCtBP1continue to be expressed in the neuroepithelium. While XCtBPtranscripts are located in the ventricular zone, XCtBP1transcripts are found in the lateral part of the neural tube (Fig.5C, St. 31). Thus, although both CtBP transcripts are detectableby RT-PCR during early embryonic development, only XCtBPis strongly expressed in the developing neurectoderm at thetime of neural induction. This suggests that, among the twoCtBP proteins, XCtBP may be the predominant XSip1biological partner during neural induction. Later, XCtBP andXCtBP1 have distinct expression patterns and their expressiondiffers from that of XSip1 in a number of tissues, whichsuggests that they may have unique functions that are XSip1independent.

XCtBP is required for XSip1 neuralization activity

To further investigate the importance of CtBP for XSip1neuralization activity, we first constructed a chimeric protein inwhich mCtBP2 was fused directly to the N-terminal zinc fingerregion of XSip1 (NZF-mCtBP2) and tested the ability of thisprotein to neuralize ectoderm in embryos and animal capexplants. Injection of this construct inhibited epidermal keratinexpression and efficiently induced Sox2 expression. This wasnot observed when overproducing XCtBP, NZF alone, or aXSip1-NZF-GFP fusion (Fig. 6A and data not shown). Second,we generated a chimeric protein in which the C-terminal zincfinger region of mSip1 was fused to a mouse Sip1 polypeptideencompassing the CtBP interacting domain (CZF-CID). Wefound that this fusion protein, although unable to neuralizeanimal caps (data not shown), slightly decreases the level ofexpression of epidermal keratin. Overproduction of CZF aloneor a fusion with a mutated CID (CZF-CIDmut) did not have thiseffect (Fig. 6B). Third, we examined whether loss or inhibitionof CtBPs affects XSip1 activity. To study CtBPs requirement,we used three independent and complementary approaches.First, an out-titration experiment was performed. As shown inFig. 6C, co-injection of XSip1 RNA with RNA coding for amouse Sip1 CID polypeptide fused to a mutant CZF (abolishedDNA binding) and a nuclear localization signal (CID) (vanGrunsven et al., 2003) reduced XSip1-mediated induction ofSox2. No such effect was observed upon co-injection withRNA coding for the same polypeptide but with mutated CtBPbinding sites (CIDmut). As a second independent method tointerfere with CtBP function, we generated an antimorphic formof mCtBP2 by fusing it to the VP16 activation domain (CtBP-VP16) and tested its effect on XSip1 neuralizing activity. Asshown in Fig. 6D, CtBP-VP16 strongly decreased XSip1 abilityto induce Sox2 expression in animal caps. Finally, we designeda morpholino oligonucleotide complementary to the XCtBPmRNA at the region of the initiator methionine codon (MOCtBP). This MO efficiently blocked the translation of XCtBPmRNA in vitro (Fig. 6E). RNA encoding XSip1 was injectedalone or together with XCtBPMO. Fig. 6E shows that inductionof Sox2 expression by XSip1 was reduced by co-injection ofXCtBP MO. Co-injection of wild-type mouse CtBP2 RNApartially restored Sox2 expression, indicating that the loss ofSox2 expression is specific to the reduction of CtBP activity.

One way by which CtBPs are known to mediate repression isthrough histone modification. As CtBPs are part of a complexcontaining histone deacetylases (HDACs) (Shi et al., 2003), weinvestigated whether XSip1 activity in neural induction involveshistone deacetylation. To this end, we cultured XSip1-over-producing animal caps in the presence of 400 nM of the HDACinhibitor trichostatin A (TSA). Treatment with TSA completelyblocked the ability of XSip1 to induce Sox2 (Fig. 6F) suggestingthat XSip1 represses gene function during neural tissueformation in an HDAC-dependent manner. This inhibitoryeffect is not a toxic effect as genes such as H4 and EF1α arenormally expressed in TSA treated caps (data not shown) (Kao etal., 1998; Koyano-Nakagawa et al., 1999). Together, theseresults indicate that the neuralizing activity of XSip1 is CtBP-

44 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

45L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

dependent and suggest that it occurs in an HDAC-dependentmanner.

XCtBP is required for XSip1 repression of BMP4

As the ability of Sip1 to repress BMP4 is crucial for itsneuralizing activity, we asked whether CtBP binding is requiredfor Sip1-mediated BMP4 repression. We first tested this bycomparing the activity of wild-type and CtBP mutant mouseSip1 proteins in transfected cells. We observed that co-transfection of increasing amounts of the mSip13xCtBPmut

expression plasmid, in contrast to the wild-type and the SBDmutant counterparts, had no effect on the activity of the 2-kb-long XBMP4 promoter (Fig. 7A). We next compared theactivities of wild-type and CtBP mutant XSip1 proteins onBMP4 expression in vivo. In animal caps overproducing similaramounts of wild-type and mutant proteins analyzed by RT-PCR,XSip13xCtBPmut repressed BMP4 and Ep. keratin expressionless efficiently than the wild-type protein and, in accordancewith the above in situ data, induces less efficiently Sox2 (Fig.7B). These data suggest that CtBP proteins significantlycontribute to XSip1 neuralizing activity by acting as a co-repressor of XSip1 for the regulation of BMP4 transcription.

Discussion

Previous studies have shown that Sip1 plays an essential roleduring vertebrate early neural development but the likelymultiple mechanisms by which Sip1 exerts its functions remainunknown (Eisaki et al., 2000; van Grunsven et al., 2000, 2006;Wakamatsu et al., 2001; Sheng et al., 2003; Van de Putte et al.,2003; Nitta et al., 2004). We show that XSip1 acts in neuraldifferentiation as a repressor that inhibits BMP4 transcriptionand therefore contributes to the limitation of BMP activity andsubsequent signaling. We also demonstrate an effect of XSip1on certain epidermal genes unrelated to BMP repression.Finally, we provide evidence that XSip1 possesses multiplerepression domains and that CtBP interaction is involved inSip1 in vivo activity, being required for efficient repression ofBMP4 and induction of neural differentiation.

Fig. 6. CtBP dependence of XSip1-mediated neuralizing activity. Whole-mount in skeratin. (A) Animal caps derived from embryos injected with NZF-CtBP or NZF-GFwith the same RNA analyzed for epidermal keratin expression. (B) Lateral view of Nfor Ep. keratin expression. (C) Animal caps derived from embryos injected with XSRNA (1 ng/blastomere) and analyzed for Sox2 expression. Note that mSip1 CID, butblot for Myc-XSip1 and Myc-CID content in the animal caps is shown in the right panof Sox2 expression. Non-injected controls are shown on the right and XSip1 controls aRNA (100 pg/blastomere) alone or co-injected with XCtBPMO (20 ng/blastomere) oSox2 expression. Caps expressing mCtBP2 alone and embryos analyzed for Sox2 areanalysis of in vitro transcription/translation reactions of XCtBP performed in the presinjected embryos were cultured from the time of their excision in the presence or abseSox2 expression. Note that TSA treatment inhibits the ability of XSip1 to induce Sox2none for NZF-GFP (n==25); 100% inhibited embryos with NZF-CtBP (n=30); nonefor NZF alone (n=30) and for NZF-CIDmut (n=32); (C) 90% with reduced stainingstaining (n=40) for XSip1+CtBP-VP16; no staining in control caps (n=30); (E) all sMO CtBP; 50% with strongly positive, n=30 for XSip1+CtBP MO and mCtBP2; nnone for XSip1+TSA (n=45).

By making use of chimeric XSip1 proteins containing aheterologous and acknowledged transcriptional repressor oractivator domain, we analyzed whether XSip1 functions as anactivator or a repressor during neural differentiation. XSip1-EnR produced the same phenotype as XSip1 in injectedembryos and animal caps on Sox2 and NCAM, while XSip1-VP16 produced the opposite effects, suggesting a likely role ofXSip1 as a transcriptional repressor. One exception was theearly neural marker Sox3, whose expression was activated byboth XSip1-EnR and XSip1-VP16, the strongest inductionbeing observed in the case of the XSip1-VP16 protein (data notshown). However, as Sox3 is not an exclusive neural marker,the induced expression observed by XSip1-VP16 couldcorrespond to ectodermal cells. Differential regulation of Sox2and Sox3 by Sip1 has already been previously reported inXSip1 morphant embryos and Sip1 knockout mice (Van dePutte et al., 2003; Snir et al., 2006).

XSip1-VP16 interferes with the activity of tBR and that ofthe wild-type protein suggesting that it acts in a dominantnegative fashion. In line with this idea, XSip1-VP16 has beenpreviously reported to induce Xbra expression in animal caps(Papin et al., 2002). To determine whether the absence ofinduction of definitive neural markers in caps and the inhibitionof their expression in XSip1-VP16 expressing embryos mightbe due to conversion of ectoderm into mesoderm, we analyzedXbra expression in our injected embryos and caps. In ourconditions, in caps stopped at stage 11, derived from embryosinjected at the four-cell stage using doses ranging from 100 to500 pg/blastomere, we were unable to detect any activation ofXbra by in situ hybridization or RT-PCR. In embryos,expression of Xbra was even strongly inhibited at thosedoses, as is the case for the native protein (data not shown).The reason for these contradictory results is unknown. Recentdata indicate that knockdown of XSip1, while inhibiting neuraldifferentiation, has no effect on Xbra, suggesting that additionalfactors are present in the neurectoderm that repress Xbraexpression. It has been suggested that the lack of effect of theXSip1 morpholinos used in the latter study might be due to thepresence of the second member of the Zfhx1 family, δEF1,which might cooperate with XSip1 to suppress Xbra in the

itu analysis of animal caps and embryos analyzed for expression of Sox2 or Ep.P RNA (250 pg/blastomere) analyzed for Sox2 expression and embryos injectedZF-CID, NZF, and NZF-CIDmut injected embryos (250 pg/blastomere) analyzedip1 mRNA (100 pg/blastomere) co-injected with mSip1 CID or mSip1 CIDmut

not mSip1 CIDmut, reduces XSip1's ability to induce Sox2. A control Westernel. (D) XSip1 and CtBP-VP16 co-expressing animal caps showing reduced levelre shown in panel E. (E) Animal caps derived from embryos injected with XSip1r co-injected with the XCtBPMO and mCtBP2 (200 pg/blastomere) analyzed forshown as controls. The efficiency of the XCtBPMO is shown in the Western blotence of increasing amounts of XCtBP MO. (F) Animal caps derived from XSip1nce of 400 nM TSA until stage 14 and then processed by in situ hybridization for. Respective inhibition/inductions (A) 70% positive caps, n=27 for NZF-CtBP;with NZF-GFP (n=32); (B) 60% inhibited embryos (n=25) for NZF-CID; none(n=32) for XSip1+CID; none (n=35) for XSip1+CIDmut; (D) all with reducedtrongly positive (n=38) for XSip1; all with reduced staining (n=39) for XSip1+one positive, n=35 for mCtBP2; (F) 85% positive caps for XSip1 (n=38) and

Fig. 7. CtBP-dependent repression of BMP4 transcription. (A) The indicatedplasmids were transfected into HEK293T cells and the cells analyzed 40 h laterfor the XBMP4 promoter-driven luciferase activity. β-Galactosidase correctedluciferase values are given. Unlike Sip1ΔSBD or wild-type Sip1, Sip13xCtBPmut

does not repress the 2-kb-long XBMP4 promoter. Control Western blots forMyc-Sip1 protein content in the transfected cells used for the luciferase assaysare shown. (B) Effect of XSip1 and XSip13XCtBPmut on the expression of theindicated genes analyzed by RT-PCR. Note that Xsip13XCtBPmut is less active inBMP4 repression than the wild-type protein. Bottom panel shows the equalexpression of XSip1 and XSip13XCtBPmut in this experiment evidenced byWestern blot using anti-Myc antibodies.

46 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

gastrula (Nitta et al., 2004). However, we recently observed thatXδEF1 transcription is only activated at neurula stage, whichstrongly argues against the idea that it regulates Xbra (vanGrunsven et al., 2006).

The observation that XSip1 acts as a repressor that is able toinduce the expression of neural markers in ectodermal explantssuggests that it affects these genes indirectly. This effect couldbe due to its ability to repress BMP4 (Eisaki et al., 2000; Nitta etal., 2004; Postigo et al., 2003; van Grunsven et al., 2006). We

found that co-expression of a constitutive BMP type I receptorinhibits XSip1's ability to induce neural markers such as Sox2and NCAM and to repress Gata2 demonstrating that inhibitionof BMP activity or signaling is necessary for XSip1 to induceneural tissue formation. Band shift assays showed that XSip1can bind to the BMP4 promoter while reporter assays inmammalian cells revealed a zinc finger and dose-dependentrepression by Sip1 of the XBMP4 promoter. Chromatinimmunoprecipitation experiments will be necessary to deter-mine whether endogenous XSip1, in the Xenopus embryo, isindeed bound to the BMP4 promoter and whether this accountsfor the downregulation of BMP4 within the prospectiveneurectoderm. Unfortunately, no anti-Sip1 antibodies areavailable yet that allow specific precipitation of endogenousSip1 in chromatin complexes in Xenopus. In contrast to theeffects observed on those neural markers, co-injection of CA-Alk3 RNA did not block XSip1's ability to repress several othersuperficial or deep cell specific epidermal genes, such as epi-dermal keratin, grainyhead-like-1, TA-2, vestigial like 4 andhyaluronan synthase 1, suggesting that XSip1 has multipleactivities, including BMP-independent functions in epidermalinhibition. This observation is in accordance with the observa-tions that E-cadherin and other genes coding for severalepithelial cell–cell junctions are directly repressed by Sip1(Comijn et al., 2001; Vandewalle et al., 2005). It is not knownwhether XSip1 binds to the promoters of any of those epidermalgenes. Further analysis of the promoters of these genes isrequired to determine whether they are directly regulated byXSip1.

Since the discovery of XSip1 as a Smad-binding protein, ithas been suggested that Smad–Sip1 interaction may have animpact on Sip1-dependent gene regulation (Verschueren et al.,1999). This interaction has been recently demonstrated to besignificant in the regulation of the Foxe3 promoter during lensdevelopment in the mouse (Yoshimoto et al., 2005). Ourmutational analysis of XSip1 indicates that Smad recruitment isnot required for XSip1 neuralizing activity during gastrulation,which suggests that Smad–Sip1 interaction plays a role in onlya subset of the events controlled by the multi-domain zinc fingerprotein that is Sip1. In contrast, several lines of evidencesupport that the co-repressor CtBP, which associates with, and isrequired for, the activity of several other transcriptional re-pressors such as TGIF and Evi1 (Alliston et al., 2005; Melhuishand Wotton, 2000; Van Campenhout et al., 2006) plays a role inXSip1 neuralizing properties. First, disruption of CtBP bindingmotifs of XSip1 reduced its ability to neuralize animal capexplants and abolished its capacity to repress a 2.0-kb-longXenopus BMP4 promoter in mammalian cells. Second, over-expression in Xenopus of a direct fusion of mCtBP2 to one ofthe DNA-binding domains of XSip1 gives the same effect as thewild-type protein. A chimeric protein, in which one of theDNA-binding domains of mSIP1 is fused to a polypeptideencompassing the CtBP interacting domains, also shows someepidermal keratin repression activity. Third, interference withXSip1 function by out-titration of CtBPs, by using a chimericmCtBP2-VP16 protein or XCtBP antisense morpholino oligo-nucleotides inhibits XSip1's ability to neuralize ectoderm.

47L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

Fourth, one of the two CtBP family members in Xenopus,XCtBP, is like XSip1, strongly expressed in the developingneurectoderm during neural induction. From these observations,we conclude that CtBP interaction plays an important role inXSip1 activity. This is consistent with the observation that, inCtBP knockout cells, many epithelial specific genes which arerepressed by Sip1 and δEF1 are up-regulated and can berepressed by re-expression of CtBP (Grooteclaes et al., 2003).The level of expression of CtBP has been recently shown to becritical for the action of δEF1 in human colon carcinomas (Penaet al., 2006).

CtBPs have been shown to play important roles duringDrosophila and mouse development (Bergman and Blaydes,2006). In Xenopus, injection of mRNA encoding an anti-morphic form of XCtBP, XCtBP/G4A, resulted in defects suchas loss of or reduced heads and/or eyes and a shortenedanterior–posterior axis (Brannon et al., 1999). We observed asimilar phenotype in CtBP MO injected embryos (data notshown), which further suggests that the level of CtBPs is criticalfor normal neural tissue development. In accordance with thisidea, we observed that CtBP genes in Xenopus, like in mouseand birds, are highly expressed in the developing neural tissue(Hildebrand and Soriano, 2002; Nick Van Hateren and Anne-Gaëlle Borycki, 2006). At tadpole stages, XCtBP and XCtBP1were found to have distinct RNA expression profiles in thedeveloping neural tube, by being expressed respectively in theventricular and marginal zones. XCtBP2 requirement duringearly neural development and neurogenesis is currently underinvestigation.

Our results also demonstrate that several XSip1 mutants thatare unable to induce Sox2 expression, including the CtBPmutant, still affect epidermal keratin expression. This indicatesthat XSip1 possesses additional CtBP-independent repressionactivities. This also suggests that disruption of the CtBPbinding sites only quantitatively decreases XSip1's totalrepressor output. In this hypothesis, the decrease of repressoractivity exerted by XSip13xCtBPmut may explain its lowerefficiency to neuralize ectodermal cells, which requires acomplete block of BMP signaling. However, it may still besufficiently active to affect epidermal keratin expression, whichoccurs only at very high levels of BMP signaling. Othertranscriptional zinc finger repressors that recruit CtBP such asKnirps and Ikaros also have distinct repression activities(Deconinck et al., 2000; Struffi et al., 2004; Sutrias-Grau andArnosti, 2004). Isolation of additional Sip1-interacting co-repressors (or complexes of these) will be necessary tounderstand the CtBP-independent repression mechanisms ofSip1. We also found that ventral ectodermal cells over-producing wild-type XSip1 or CtBP mutant proteins that didnot express epidermal keratin did not adopt a neural crest orplacodal fate. Similar observations have been made in the caseof overexpression of Smad6 (Delaune et al., 2005). In thefuture, it will be interesting to determine the nature of thosecells that are no longer epidermal, neural, or neural plate bordercells.

CtBPs exist as a multi-protein complex that includes thehistone-methyltransferases (HMTs) G9a and Eu-HMTase1 but

also HDAC1 and HDAC2 (Shi et al., 2003). We found thattreatment with TSA alleviates XSip1 induced expression ofSox2 suggesting that XSip1, by interacting with CtBPs, recruitsan HDAC complex, which exerts the transcriptional repressionactivity. Other studies are required to address the issue of how,in what kind of cells, and in which order, Sip1 recruits partnerslike Smads, CtBPs and other proteins to regulate target geneexpression.

Acknowledgments

We thank Horst Grunz, Richard Harland, Walter Knöchel,Doug Melton, Roger Patient, Thomas Sargent and AndrewChalmers for providing us with various constructs. This workwas supported by the Belgian program Inter-UniversitaryAttraction Poles from the Prime Minister's Office, SciencePolicy Programming (grant P5/35 and P6/20 to E.B. and D.H.),and by the Communauté Française de Belgique (grant numberARC 00/05–250 to E.B.). E.B. was also supported by the Fundfor Scientific Medical Research (FRSM 3.4555.01 and3.4556.01), by the International Brachet Stiftung (grant number01-4/l), and by the Fondation Reine Elisabeth (grant 11256).The work in the D.H. laboratory was supported by VIB (VIB7)and the Fund for Scientific Research-Flanders (FWO-VG.0279.04). L.A.v.G and J.S. were post-doctoral fellows fromFWO-Vand FNRS, respectively. V.T. and M.N. were fellows ofthe Fonds pour la Formation a la Recherche dans l'Industrie etl'Agriculture and G.V. was financed by a PhD grant of theInstitute for the Promotion of Innovation through Science andTechnology in Flanders (IWT-Vlaanderen).

Appendix A. Supplementary data

Supplementary data associated with this article can be found,in the online version, at doi:10.1016/j.ydbio.2007.02.045.

References

Alliston, T., Ko, T.C., Cao, Y., Liang, Y.-Y., Feng, X.-H., Chang, C., Derynck,R., 2005. Repression of BMP and activin-inducible transcription by Evi-1.J. Biol. Chem. 280, 24227–24237.

Bergman, L.M., Blaydes, J.P., 2006. C-terminal binding proteins: emergingroles in cell survival and tumorigenesis. Apoptosis 11, 879–888.

Brannon, M., Brown, J.D., Bates, R., Kimelman, D., Moon, R.T., 1999. XCtBPis a XTcf-3 co-repressor with roles throughout Xenopus development.Development 126, 3159–3170.

Bylund,M., Andersson, E., Novitch, B.G.,Muhr, J., 2003. Vertebrate neurogenesisis counteracted by Sox1–3 activity. Nat. Neurosci. 6, 1162–1168.

Chalmers, A.D., Welchman, D., Papalopulu, N., 2002. Intrinsic differencesbetween the superficial and deep layers of the Xenopus ectoderm controlprimary neuronal differentiation. Dev. Cell 2, 171–182.

Comijn, J., Berx, G., Vermassen, P., Verschueren, K., van Grunsven, L.,Bruyneel, E., Mareel, M., Huylebroeck, D., van Roy, F., 2001. The two-handed E box binding zinc finger protein SIP1 downregulates E-cadherinand induces invasion. Mol. Cell 7, 1267–1278.

Deconinck, A.E., Mead, P.E., Tevosian, S.G., Crispino, J.D., Katz, S.G., Zon,L.I., Orkin, S.H., 2000. FOG acts as a repressor of red blood cell devel-opment in Xenopus. Development 127, 2031–2040.

De Robertis, E.M., Kuroda, H., 2004. Dorsal-ventral patterning and neuralinduction in Xenopus embryos. Annu. Rev. Cell Dev. Biol. 20, 285–308.

48 L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

Delaune, E., Lemaire, P., Kodjabachian, L., 2005. Neural induction in Xenopusrequires early FGF signalling in addition to BMP inhibition. Development132, 299–310.

Eisaki, A., Kuroda, H., Fukui, A., Asashima, M., 2000. XSIP1, a member oftwo-handed zinc finger proteins, induced anterior neural markers in Xenopuslaevis animal cap. Biochem. Biophys. Res. Commun. 271, 151–157.

Graham, V., Khudyakov, J., Ellis, P., Pevny, L., 2003. SOX2 functions tomaintain neural progenitor identity. Neuron 39, 749–765.

Grooteclaes,M.,Deveraux, Q., Hildebrand, J., Zhang,Q., Goodman, R.H., Frisch,S.M., 2003. C-terminal-binding protein corepresses epithelial and proapopto-tic gene expression programs. Proc. Natl. Acad. Sci. 100, 4568–4573.

Hemmati-Brivanlou, A., Thomsen, G.H., 1995. Ventral mesodermal patterningin Xenopus embryos: expression patterns and activities of BMP-2 and BMP-4. Dev. Genet. 17, 78–89.

Hildebrand, J.D., Soriano, P., 2002. Overlapping and unique roles for C-terminalbinding protein 1 (CtBP1) and CtBP2 during mouse development. Mol.Cell. Biol. 22, 5296–5307.

Jaynes, J.B., O'Farrell, P.H., 1991. Active repression of transcription by theengrailed homeodomain protein. EMBO J. 10, 1427–1433.

Jonas, E., Sargent, T.D., Dawid, I.B., 1985. Epidermal keratin gene expressedin embryos of Xenopus laevis. Proc. Natl. Acad. Sci. U. S. A. 82,5413–5417.

Kao, H.Y., Ordentlich, P., Koyano-Nakagawa, N., Tang, Z., Downes, M.,Kintner, C.R., Evans, R.M., Kadesch, T., 1998. A histone deacetylasecorepressor complex regulates the Notch signal transduction pathway. GenesDev. 12, 2269–2277.

Kishi, M., Mizuseki, K., Sasai, N., Yamazaki, H., Shiota, K., Nakanishi, S.,Sasai, Y., 2000. Requirement of Sox2-mediated signaling for differentiationof early Xenopus neuroectoderm. Development 127, 791–800.

Koyano-Nakagawa, N., Wettstein, D., Kintner, C., 1999. Activation of Xenopusgenes required for lateral inhibition and neuronal differentiation duringprimary neurogenesis. Mol. Cell. Neurosci. 14, 327–339.

Krieg, P.A., Sakaguchi, D.S., Kintner, C.R., 1989. Primary structure anddevelopmental expression of a large cytoplasmic domain form of Xenopuslaevis neural cell adhesion molecule (NCAM). Nucleic Acids Res. 17,10321–10335.

Kroll, K.L., Salic, A.N., Evans, L.M., Kirschner, M.W., 1998. Geminin, aneuralizing molecule that demarcates the future neural plate at the onset ofgastrulation. Development 125, 3247–3258.

Kuo, J.S., Patel, M., Gamse, J., Merzdorf, C., Liu, X., Apekin, V., Sive, H.,1998. Opl: a zinc finger protein that regulates neural determination andpatterning in Xenopus. Development 125, 2867–2882.

Lahaye, K., Kricha, S., Bellefroid, E.J., 2002. XNAP, a conserved ankyrinrepeat-containing protein with a role in the Notch pathway during Xenopusprimary neurogenesis. Mech. Dev. 110, 113–124.

Lerchner, W., Latinkic, B.V., Remacle, J.E., Huylebroeck, D., Smith, J.C., 2000.Region-specific activation of the Xenopus brachyury promoter involvesactive repression in ectoderm and endoderm: a study using transgenic frogembryos. Development 127, 2729–2739.

Linker, C., Stern, C.D., 2004. Neural induction requires BMP inhibition only asa late step, and involves signals other than FGF and Wnt antagonists.Development 131, 5671–5681.

Long, J., Zuo, D., Park, M., 2005. Pc2-mediated SUMOylation of Smad-interacting protein 1 attenuates transcriptional repression of E-cadherin.J. Biol. Chem. 280, 35477–35489.

Marchant, L., Linker, C., Ruiz, P., Guerrero, N., Mayor, R., 1998. The inductiveproperties of mesoderm suggest that the neural crest cells are specified by aBMP gradient. Dev. Biol. 198, 319–329.

Melhuish, T.A., Wotton, D., 2000. The interaction of the carboxyl terminus-binding protein with the Smad corepressor TGIF is disrupted by aholoprosencephaly mutation in TGIF. J. Biol. Chem. 275, 39762–39766.

Metz, A., Knochel, S., Buchler, P., Koster, M., Knochel, W., 1998. Structuraland functional analysis of the BMP-4 promoter in early embryos of Xenopuslaevis. Mech. Dev. 74, 29–39.

Mizuseki, K., Kishi, M., Matsui, M., Nakanishi, S., Sasai, Y., 1998a. XenopusZic-related-1 and Sox-2, two factors induced by chordin, have distinctactivities in the initiation of neural induction. Development 125, 579–587.

Mizuseki, K., Kishi, M., Shiota, K., Nakanishi, S., Sasai, Y., 1998b. SoxD: an

essential mediator of induction of anterior neural tissues in Xenopusembryos. Neuron 21, 77–85.

Mowat, D.R., Wilson, M.J., Goossens, M., 2003. Mowat–Wilson syndrome. J.Med. Genet. 40, 305–310.

Munoz-Sanjuan, I., Bell, E., Altmann, C.R., Vonica, A., Brivanlou, A.H., 2002.Gene profiling during neural induction in Xenopus laevis: regulation ofBMP signaling by post-transcriptional mechanisms and TAB3, a novelTAK1-binding protein. Development 129, 5529–5540.

Nakata, K., Nagai, T., Aruga, J., Mikoshiba, K., 1997. Xenopus Zic3, a primaryregulator both in neural and neural crest development. Proc. Natl. Acad. Sci.U. S. A. 94, 11980–11985.

Nick Van Hateren, T.S., Anne-Gaëlle Borycki, 2006. Expression of avian C-terminal binding proteins (Ctbp1and Ctbp2) during embryonic development.Dev. Dyn. 235, 490–495.

Nieuwkoop, P.D., Faber, J., 1997. A Normal Table of Xenopus laevis (Daudin).North Holland Publishing Co, Amsterdam, The Netherlands.

Nitta, K.R., Tanegashima, K., Takahashi, S., Asashima, M., 2004. XSIP1 isessential for early neural gene expression and neural differentiation bysuppression of BMP signaling. Dev. Biol. 275, 258–267.

Onichtchouk, D., Chen, Y.G., Dosch, R., Gawantka, V., Delius, H., Massague,J., Niehrs, C., 1999. Silencing of TGF-beta signalling by the pseudoreceptorBAMBI. Nature 401, 480–485.

Papin, C., van Grunsven, L.A., Verschueren, K., Huylebroeck, D., Smith, J.C.,2002. Dynamic regulation of Brachyury expression in the amphibianembryo by XSIP1. Mech. Dev. 111, 37–46.

Pena, C., Garcia, J.M., Garcia, V., Silva, J., Dominguez, G., Rodriguez, R.,Maximiano, C., Garcia de Herreros, A., Munoz, A., Bonilla, F., 2006. Theexpression levels of the transcriptional regulators p300 and CtBP modulatethe correlations between SNAIL, ZEB1, E-cadherin and vitamin D receptorin human colon carcinomas. Int. J. Cancer 119, 2098–2104.

Penzel, R., Oschwald, R., Chen, Y., Tacke, L., Grunz, H., 1997. Characterizationand early embryonic expression of a neural specific transcription factorxSOX3 in Xenopus laevis. Int. J. Dev. Biol. 41, 667–677.

Postigo, A.A., Dean, D.C., 2000. Differential expression and function ofmembers of the zfh-1 family of zinc finger/homeodomain repressors. Proc.Natl. Acad. Sci. U. S. A. 97, 6391–6396.

Postigo, A.A., Depp, J.L., Taylor, J.J., Kroll, K.L., 2003. Regulation of Smadsignaling through a differential recruitment of coactivators and corepressorsby ZEB proteins. EMBO J. 22, 2453–2462.

Remacle, J.E., Kraft, H., Lerchner, W., Wuytens, G., Collart, C., Verschueren,K., Smith, J.C., Huylebroeck, D., 1999. New mode of DNA binding ofmulti-zinc finger transcription factors: deltaEF1 family members bind withtwo hands to two target sites. EMBO J. 18, 5073–5084.

Ryan, K., Butler, K., Bellefroid, E.J., Gurdon, J.B., 1998. Xenopus Eomeso-dermin is expressed in neural differentiation. Mech. Dev. 75, 167–170.

Schmidt, J.E., Suzuki, A., Ueno, N., Kimelman, D., 1995. Localized BMP-4mediates dorsal/ventral patterning in the early Xenopus embryo. Dev. Biol.169, 37–50.

Sewalt, R.G.A.B., Gunster,M.J., van der Vlag, J., Satijn, D.P.E., Otte, A.P., 1999.C-terminal binding protein is a transcriptional repressor that interacts with aspecific class of vertebrate polycomb proteins. Mol. Cell. Biol. 19, 777–787.

Sheng, G., dos Reis, M., Stern, C.D., 2003. Churchill, a zinc finger trans-criptional activator, regulates the transition between gastrulation andneurulation. Cell 115, 603–613.

Shi, Y., Sawada, J., Sui, G., Affar el, B., Whetstine, J.R., Lan, F., Ogawa, H.,Luke, M.P., Nakatani, Y., 2003. Coordinated histone modifications mediatedby a CtBP co-repressor complex. Nature 422, 735–738.

Sive, H., Grainger, R.M., Harland, R.M., 2000. Early Development of Xenopuslaevis. A Laboratory Manual. Cold Spring Harbor, Laboratory Press, ColdSpring Harbor, NY.

Smith, J.C., Price, B.M., Green, J.B., Weigel, D., Herrmann, B.G., 1991.Expression of a Xenopus homolog of Brachyury (T) is an immediate-earlyresponse to mesoderm induction. Cell 67, 79–87.

Snir, M., Ofir, R., Elias, S., Frank, D., 2006. Xenopus laevis POU91 protein, anOct3/4 homologue, regulates competence transitions from mesoderm toneural cell fates. EMBO J 25, 3664–3674.

Stern, C.D., 2005. Neural induction: old problem, new findings, yet morequestions. Development 132, 2007–2021.

49L.A. van Grunsven et al. / Developmental Biology 306 (2007) 34–49

Struffi, P., Corado, M., Kulkarni, M., Arnosti, D.N., 2004. Quantitativecontributions of CtBP-dependent and -independent repression activities ofKnirps. Development 131, 2419–2429.

Sutrias-Grau, M., Arnosti, D.N., 2004. CtBP contributes quantitatively toKnirps repression activity in an NAD binding-dependent manner. Mol. Cell.Biol. 24, 5953–5966.

Suzuki, A., Thies, R., Yamaji, N., Song, J., Wozney, J., Murakami, K., Ueno, N.,1994. A Truncated bone morphogenetic protein receptor affects dorsal–ventral patterning in the early Xenopus embryo. Proc. Natl. Acad. Sci. 91,10255–10259.

Tao, J., Kuliyev, E., Wang, X., Li, X., Wilanowski, T., Jane, S.M., Mead, P.E.,Cunningham, J.M., 2005. BMP4-dependent expression of Xenopus Grainy-head-like 1 is essential for epidermal differentiation. Development 132,1021–1034.

Van Campenhout, C., Nichane, M., Antoniou, A., Pendeville, H., Bronchain,O.J., Marine, J.-C., Mazabraud, A., Voz, M.L., Bellefroid, E.J.,2006. Evi1 is specifically expressed in the distal tubule and duct of theXenopus pronephros and plays a role in its formation. Dev. Biol. 294,203–219.

Van de Putte, T., Maruhashi, M., Francis, A., Nelles, L., Kondoh, H.,Huylebroeck, D., Higashi, Y., 2003. Mice lacking ZFHX1B, the gene thatcodes for Smad-interacting protein-1, reveal a role for multiple neural crestcell defects in the etiology of Hirschsprung disease-mental retardationsyndrome. Am. J. Hum. Genet. 72, 465–470.

Vandewalle, C., Comijn, J., De Craene, B., Vermassen, P., Bruyneel, E.,Andersen, H., Tulchinsky, E., Van Roy, F., Berx, G., 2005. SIP1/ZEB2induces EMT by repressing genes of different epithelial cell–cell junctions.Nucleic Acids Res. 33, 6566–6578.

van Grunsven, L.A., Papin, C., Avalosse, B., Opdecamp, K., Huylebroeck, D.,Smith, J.C., Bellefroid, E.J., 2000. XSIP1, a Xenopus zinc finger/homeodomain encoding gene highly expressed during early neuraldevelopment. Mech. Dev. 94, 189–193.

van Grunsven, L.A., Michiels, C., Van de Putte, T., Nelles, L., Wuytens, G.,Verschueren, K., Huylebroeck, D., 2003. Interaction between Smad-interact-ing protein-1 and the corepressor C-terminal binding protein is dispensable fortranscriptional repression of E-cadherin. J. Biol. Chem. 278, 26135–26145.

van Grunsven, L.A., Taelman, V., Michiels, C., Opdecamp, K., Huylebroeck,D., Bellefroid, E.J., 2006. dEF1 and SIP1 are differentially expressed andhave overlapping activities during Xenopus embryogenesis. Dev. Dyn. 235,1491–1500.

Verschueren, K., Remacle, J.E., Collart, C., Kraft, H., Baker, B.S., Tylzanowski,P., Nelles, L., Wuytens, G., Su, M.T., Bodmer, R., Smith, J.C., Huylebroeck,D., 1999. SIP1, a novel zinc finger/homeodomain repressor, interacts withSmad proteins and binds to 5′-CACCT sequences in candidate target genes.J. Biol. Chem. 274, 20489–20498.

Wakamatsu, N., Yamada, Y., Yamada, K., Ono, T., Nomura, N., Taniguchi, H.,Kitoh, H., Mutoh, N., Yamanaka, T., Mushiake, K., Kato, K., Sonta, S.,Nagaya, M., 2001. Mutations in SIP1, encoding Smad interacting protein-1,cause a form of Hirschsprung disease. Nat. Genet. 27, 369–370.

Walmsley, M.E., Guille, M.J., Bertwistle, D., Smith, J.C., Pizzey, J.A., Patient,R.K., 1994. Negative control of Xenopus GATA-2 by activin and nogginwith eventual expression in precursors of the ventral blood islands.Development 120, 2519–2529.

Yoshimoto, A., Saigou, Y., Higashi, Y., Kondoh, H., 2005. Regulation of ocularlens development by Smad-interacting protein 1 involving Foxe3 activation.Development 132, 4437–4448.

Suplemental Figure 1: —The XSip1CtBPmut does not bind to CtBP. Plasmids encoding either Myc-tagged wild type or CtBP mutant mSip1 or XSip1 (PLNLS786-790, PLDLT816-820, PLNLT860-864) were all mutated into AAALS or AAALT) and Flag-tagged CtBP proteins were transfected in 293T cells and cell lysates processed for immunoprecipitation of the complexes with the appropriate antibodies (for immunoprecipitation in both directions). Protein complexes (IP) were separated by SDS-PAGE and analyzed by Western blotting. A portion of each lysate was analyzed for expression of transfected proteins by Western blot (Lysate). Note that the CtBP-mutant Sip1 proteins cannot immunoprecipitate CtBP (left side of gel) nor can CtBP immunoprecipitate Sip1-CtBP mutants (right side of gel) while for the wild type Sip1 these interactions are easily detected.

IP: Anti-Myc Anti-Flag

Flag-hCtBP1 x x x x x x x x xMyc-mSip1 wt mut wt mutMyc-XSip1 mutwt wt mutwt wt

WB:Myc

WB:Flag

WB:Myc

WB:Flag

mSip1XSip1

mSip1XSip1

CtBP1

CtBP1

Lysate

IgG

Suplemental Figure2: Overexpression of XSip1 or XSip1ΔCtBP does not turn onneural crest or placodal markers. Slug expression marks neural crest progenitors.Six1 expression marks cephalic neurogenic placodes. (A) Embryos injectedanimally with the indicated ARN together with LacZ RNA. Noth that whileoverexpression of both XSip1 or XSip1ΔCtBP leads to suppression of epidermis, Slugand Six1 expression are not induced. (B) Animal caps explanted at stage 9derived from embryos overexpressing XSip1ΔCtBP or control uninjected embryos.Note that while overexpression of XSip1ΔCtBP inhibits epidermal keratin expression,Slug and Six1 expression are not induced.

XSip1

XSip1ΔCtBP

Ep. keratin Slug Six1A

B

uninj.

Control

XSip1ΔCtBP