Expression of the Wnt inhibitor Dickkopf-1 is required for the induction of neural markers in mouse embryonic stem cells differentiating in response to retinoic acid


Address correspondence and reprint requests to Daniela Melchiorri, PhD, Department of Human Physiology and Pharmacology, University of Rome ‘La Sapienza’, Piazzale Aldo Moro 5, 00185, Rome, Italy. E-mail:


Cultured mouse D3 embryonic stem (ES) cells differentiating into embryoid bodies (EBs) expressed several Wnt isoforms, nearly all isotypes of the Wnt receptor Frizzled and the Wnt/Dickkopf (Dkk) co-receptor low-density lipoprotein receptor-related protein (LRP) type 5. A 4-day treatment with retinoic acid (RA), which promoted neural differentiation of EBs, substantially increased the expression of the Wnt antagonist Dkk-1, and induced the synthesis of the Wnt/Dkk-1 co-receptor LRP6. Recombinant Dkk-1 applied to EBs behaved like RA in inducing the expression of the neural markers nestin and distal-less homeobox gene (Dlx-2). Recombinant Dkk-1 was able to inhibit the Wnt pathway, as shown by a reduction in nuclear β-catenin levels. Remarkably, the antisense- or small interfering RNA-induced knockdown of Dkk-1 largely reduced the expression of Dlx-2, and the neuronal marker β-III tubulin in EBs exposed to RA. These data suggest that induction of Dkk-1 and the ensuing inhibition of the canonical Wnt pathway is required for neural differentiation of ES cells.

Abbreviations used

antisense oligonucleotide






days in vitro




embryoid body


embryonic stem


fetal calf serum




glyceraldehyde-3-phosphate dehydrogenase


green fluorescent protein


glycogen synthase kinase


human recombinant Dkk-1


leukaemia inhibitory factor


low-density lipoprotein receptor-related protein


protected Least Significant Difference parametric


retinoic acid


sense oligonucleotide


small interfering RNA

The Wnt signalling pathway controls patterning and cell fate determination during embryonic development as well as tissue homeostasis and tumorigenesis in adult life (Moon et al. 2002). Recent studies have shown that Wnt signalling regulates the maintenance and differentiation of cultured embryonic stem (ES) cells (Sato et al. 2004; Hao et al. 2005). Wnt signals are transmitted via several pathways, such as the canonical Wnt/β-catenin pathway and the non-canonical Wnt/Ca2+ and Junk kinase (JNK) pathways (Chen et al. 2000; Kuhl et al. 2000). In the canonical pathway, binding of Wnt to its cell surface receptors Frizzled (FZD) and low-density lipoprotein receptor-related protein (LRP)5/6 activates a cascade of downstream events resulting in inhibition of glycogen synthase kinase (GSK)3β with ensuing migration of β-catenin to the nucleus. Nuclear β-catenin combines with transcription factors of the T-cell factor and lymphoid enhancer factor (TCF/LEF) family and activates Wnt target genes, such as c-myc and and cyclin D1.

Wnt signalling is inhibited by Dickkopf (Dkk)-1, a secreted glycoprotein that binds to the Wnt co-receptors LRP5 and LRP6, and to Kremen1 and Kremen2. The resulting trimolecular complex of Dkk-1, LRP5/6 and Kremen is endocytosed, resulting in the depletion of LRP5/6 co-receptor from the plasma membrane (Mao et al. 2002). A role for Dkk-1 has been extensively studied during early development. Dkk-1 neutralizing antibodies inhibit head and prechordal plate formation (Glinka et al. 1998; Kazanskaya et al. 2000), and Dkk-1 knockout mice do not develop CNS structures anterior to the midbrain (Mukhopadhyay et al. 2001). In addition, Davidson et al. (2002) have found that in vivo interaction between Kremens and Dkk-1 is required for the formation of the anterior CNS in Xenopus. In cultured ES cells, activation of the canonical Wnt pathway with either Wnt-1 or lithium ions (an inhibitor of GSK3β), or overexpression of a dominant active form of β-catenin, inhibits neural differentiation (Aubert et al. 2002; Haegele et al. 2003). So far, the role of Wnt inhibition in the induction of neural differentiation in vitro has been examined by either forced expression or application of Wnt antagonists. In cultured ES cells, overexpression of the Wnt antagonist secreted FZD-related protein-2 (SFRP-2) is sufficient to stimulate the production of neural progenitor cells (Aubert et al. 2002), and co-application of exogenous Dkk-1 with the anti-Nodal reagent Lefty A increases the proportion of cells expressing the early neuroectodermal marker Sox-1 (Watanabe et al. 2005). To what extent endogenously expressed Dkk-1 contributes to neural differentiation of ES cells is unknown. We now report that Dkk-1 is strongly induced in ES cells differentiating in response to retinoic acid (RA), and that Dkk-1 knockdown induced by either small interfering RNA (siRNA) or antisense oligonucleotides reduces the expression of the putative neural markers nestin and Dlx-2.

Materials and methods


All trans-retinoic acid (RA) was purchased from Sigma Aldrich (Milan, Italy), leukaemia inhibitory factor (LIF/ESGRO) was from Chemicon (Milan, Italy) and human recombinant Dkk-1 (hrDkk-1) was from Chemicon, CA, USA.

Cell cultures

The D3 mouse embryonic cell line derived from 129/Sv+c/+p mice (ATCC, Milan, Italy) were plated on tissue culture plates coated with 0.1% gelatin and cultured under non-differentiating conditions in Knockout Dulbecco's modified Eagle's medium supplemented with minimal essential medium (MEM) non-essential amino acid solution (100 μm; Invitrogen, Milan, Italy), penicillin (100 U/mL), streptomycin (100 µg/mL), glutamine (2 mm), 2-mercaptoethanol (55 μm), fetal bovine serum (ES-CULT FBS 15%; Stem Cell Technologies, Biospa, Milan, Italy) and LIF/ESGRO (1400 U/mL). Formation of embryoid bodies (EBs) was induced by withdrawal of LIF/ESGRO from cell cultures and by plating cells on to a non-adhesive substrate in the presence of serum. For the induction of neural differentiation, RA (1 µm) was added to cultures on day 4 of EB formation, and maintained for 4 more days (4 –/4 + day treatment). We also used PC12 cells to test the ability of hrDkk-1 to inhibit the Wnt pathway. PC12 cells were cultured in Dulbecco's modified Eagle's medium supplemented with glutamine (2 mm) and 10% fetal calf serum (FCS), and maintained at 37°C in a 10% CO2 humidified atmosphere.


Total RNA was extracted from EBs using Trizol reagent (Invitrogen) and subjected to Dnase I treatment (Promega, Milan, Italy) according to manufacturer's instructions. Two micrograms of total RNA was used for cDNA synthesis using Superscript II (BRL Life Technology, Milan, Italy) and random hexamer primers according to manufacturer's instructions. The RT product was diluted to 100 μl with sterile, distilled water and 1 μl cDNA was employed in each subsequent amplification. The following primers were used:

Wnt1, 5′-ACAGCAACCACAGTCGTCAG-3′ (reverse) and 5′-GAATCCGTCAACAGGTTCGT-3′ (forward); Wnt2, 5′-ATCTCTTCAGCTGGCGTTGT-3′ (reverse) and 5′-CCTTCCTTCCAGCTCTGTTG-3′ (forward); Wnt3, 5-GCGACTTCCTCAAGGACAAG-3′ (reverse) and 5′-AAAGTTGGGGGAGTTCTCGT-3′ (forward); Wnt3a, 5-CCCTTTCCAGTCCTGGTGTA-3′ (reverse) and 5′-CTTGAAGAAGGGGTGCAGAG-3′ (forward); Wnt4, 5′-CTGGAGAAGTGTGGCTGTGA-3′ (reverse) and 5′-CAGCCTCGTTGTTGTGAAGA-3′ (forward); Wnt5a, 5′-CAAATAGGCAGCCGAGAGAC-3′ (reverse) and 5′-CTCTAGCGTCCACGAACTCC-3′ (forward); Wnt6, 5′-CGGTAGAGCTCTCAGGATGC-3′ (reverse) and 5′-ATTCTCGAACCCCCAGTCTT-3′ (forward); Wnt7a, 5′-TGAGCACTTGTGGTCTCAGG-3′ (reverse) and 5′-GCATCTGAGTTTCACCAGCA-3′ (forward); Dkk-1, 5′-GAGGGGAAATTGAGGAAAGC-3′ (reverse) and 5′-GGTGCACACCTGACCTTCTT-3′ (forward); LRP5, 5′-CAGGTGCTTGTGTGGAGAGA-3′ (reverse) and 5′-CATGTTGGTGTCCAGGTCAG-3′ (forward); LRP6, 5′-GGTGTCAAAGAAGCCTCTGC-3′ (reverse) and 5′-ACCTCAATGCGATTTGTTCC-3′ (forward); FZD1, 5′-CACCTGGATAGGCATCTGGT-3′ (reverse) and 5′-CAGAAAGCCAGCGATGTAGG-3′ (forward); FZD2, 5′-CCGACGGCTCTATGTTCTTC-3′ (reverse) and 5′-TAGCAGCCGGACAGAAAGAT-3′ (forward); FZD3, 5′-AGCGTGCCTATAGCGAGTGT-3′ (reverse) and 5′-TCTCTGGGACACCAAAAACC-3′ (forward); FZD4, 5′-CTGCAGCATGCCTAATGAGA-3′ (reverse) and 5′-CGTCTGCCTAGATGCAATCA-3′ (forward); FZD5, 5′-TCTTGTCTGCGTGCTACCTG-3′ (reverse) and 5′-GGCCATGCCAAAGAAATAGA-3′ (forward); FZD6, 5′-TGTTGGGCTGTCTCTCCTCT-3′ (reverse) and 5′-TCTCCCAGGTGATCCTGTTC-3′ (forward); FZD7, 5′-CCATCCTCTTCATGGTGCTT-3′ (reverse) and 5′-TGGCCAAAATGGTGATTGT-3′ (forward); FZD8, 5′-CTGTTCCGAATCCGTTCAGT-3′ (reverse) and 5- CGGTTGTGCTGCTCATAGAA-3′ (forward); FZD9, 5′-AGTTTCCTCCTGACCGGTTT-3′ (reverse) and 5′-TCCATGTTCAGGCGTTCATA-3′ (forward); FZD10, 5′-AGTTTCCTCCTGACCGGTTT-3′ (reverse) and 5′-TCCATGTTGAGGCGTTCATA-3′ (forward).

For β-actin amplification, we used the following primers that spanned an intron and yielded products of different sizes depending on whether cDNA or genomic DNA was used as template (400 bp and 600 bp respectively): 5′-GCTCATAGCTCTTCTCCAGGG-3′ (reverse) and 5′-TGAACCCTAAGGCCAACCGTG-3′ (forward).

Real-time PCR

Real-time quantitative PCR was performed using a 2 × Supermix cocktail (Bio-Rad, Hercules, CA, USA) containing the double-stranded DNA-binding fluorescent probe Sybr Green and all necessary components except primers. Quantitative PCR conditions included an initial denaturation step of 94°C for 10 min followed by 40 cycles of 94°C for 15 s, and 55°C for 15 s. Standards, samples and negative controls (no template) were analysed in triplicate. Concentrations of mRNA were calculated from serially diluted standard curves simultaneously amplified with unknown samples and normalized with respect to β-actin mRNA levels.

Western blot analysis

Western blot analysis was performed as described previously (Iacovelli et al. 2004). Briefly, EBs were either cultured for 8 days in the absence of any pharmacological treatment (controls) or subjected to a 4 –/4 + day treatment with RA or human recombinant Dkk-1. The reaction was stopped by washing twice with ice-cold phosphate-buffered saline and cells were lysed for 10 min at 4°C in Triton X-100 lysis buffer (10 mm Tris-HCl, pH 7.4, 150 mm NaCl, 1% Triton X-100, 1 mm EDTA, 10% glycerol, 1 mm phenylmethylsulphonyl fluoride, 10 µg/mL leupeptin, 10 µg/mL aprotinin, 1 mm sodium orthovanadate, 50 mm sodium fluoride, 10 mmβ-glycerophosphate). For the detection of β-catenin in cell nuclei, nuclear fractions were isolated by differential centrifugation using a commercially available kit (Pierce, Rockford, IL, USA). The purity of the fraction was verified using an antibody against the nuclear protein poly-ADP ribose polymerase (Santa Cruz Biotechnology, Santa Cruz, CA, USA). Cell lysates were cleared by centrifugation (10 000 g for 10 min), and 40–80 µg protein was separated by sodium dodecyl sulfate–polyacrylamide gel electrophoresis, blotted on to nitrocellulose and probed using different commercial antibodies. Membranes were saturated for 1 h with Tris-buffered saline (100 mm Tris, 0.9% NaCl) containing 0.05% Tween 20 and 5% non-fat dry milk, and then incubated overnight with primary antibodies. Antibodies were used at the following dilutions: rabbit Dlx-2 polyclonal antibodies (1 : 200; Chemicon); rabbit β-catenin polyclonal antibodies (1 : 1000; Cell Signaling Technology, Beverly, MA, USA,); mouse β-III tubulin monoclonal antibody (TuJ1, 1 : 1000; Covance, DBA, Milan, Italy), mouse β-actin monoclonal antibody (1 : 5000; Sigma Aldrich) and rabbit glyceraldehyde-3-phosphate dehydrogenase (GAPDH) antibody (1 : 1000; Abcam, Cambridge, UK). Because EB differentiation was associated with changes in all housekeeping proteins we tested (β-actin, GAPDH), Dlx-2, β-III tubulin and β-catenin signals were normalized with respect to the amount of protein loaded in the gel stained with Pelican indian ink (1 µL/mL in 0.2% Tween-20 in Tris-buffered saline).


EBs were fixed in 4% paraformaldehyde for 20 min. After blocking with bovine serum albumin for 1 h, incubation with mouse monoclonal nestin antibodies (1 : 100; Chemicon), rabbit polyclonal Dlx-2 antibodies (1 : 100) or mouse monoclonal β-III tubulin (1 : 500) was performed overnight in the presence of digitonin. Soon after, cells were incubated with fluoresceinated anti-mouse or anti-rabbit secondary antibodies (dilution 1 : 100; both from Vector, Burlingame, CA, USA) in the presence of digitonin and observed with a laser scanner microscope (LSM510; Zeiss, Oberkochen, Germany).


Construction of plasmids was as follows. For Wnt7a expression construct, Wnt7a cDNA was amplified from human adult brain cDNA by using primers carrying EcoRV and XbaI restriction sites at the flanking ends. The amplified cDNA was sequenced and subcloned into the EcoRV and XbaI restriction enzyme sites of the eukaryotic expression pCIN4 vector (Rees et al. 1996; Caricasole et al. 2002).

Transient transfection assays for reporter studies

Transfection and reporter assays were carried out essentially as described by Caricasole et al. (2002). Transient transfections of PC12 cells were carried out in triplicate, using LipofectAMINE 2000 (Invitrogen) according to the manufacturer's instructions. PC12 cells (6 × 105/well) were plated 1 day before transfection in collagen-coated 96-well culture plates. A total of 0.64 µg DNA was transfected into each well, including luciferase reporter plasmid (150 ng), expression construct (200 ng of each expression construct), Renilla luciferase (phRL-TK) vector-driven internal reporter (100 ng; Promega), and carrier plasmid DNA (Bluescript, 640 ng; Promega) as appropriate. The luciferase reporter plasmid was p4TCF, comprising four copies of a TCF-responsive element upstream of a TATA element luciferase coding sequence transcriptional unit (Bettini et al. 2002). Luciferase activity was measured using Promega dual luciferase assay reagent and read using a Berthold (Bad Wildbad, Germany) LUMAT LB3907 tube luminometer. Readings were from triplicate transfections and were automatically normalized relative to the internal standard (Renilla luciferase).

Addition of antisense oligonucleotides to coltures

Cultures were treated with the following ‘end-capped’ phosphorothioate oligonucleotides directed against mouse Dkk-1: 5′-(phosphate) TAcAGATCTTGGACcAgA3′ (antisense) and 5′-TCtGGTCCAAGATCtGtA3′ (sense) (Cappuccio et al. 2005). Oligonucleotides (1.8 µm) were applied to EBs twice at day 4 and 6 after plating. Cultures were stopped at day 8 and processed for western blot analysis.

siRNA studies

ES cells were cultured for 2 days under non-differentiating conditions. On day 2, cells were electroporated using Nucleofactor technology (Amaxa Biosystem; (Cologne, Germany) Leclere et al. 2005). Briefly, following trypsinization, 2 × 106 ES cells were resuspended in 100 µL ES cell Nucleofactor solution (Amaxa Biosystem) at room temperature (25°C) followed by addition of 1.5 µg of two different siRNAs targeting the transcript of mouse Dkk-1 (Quiagen, Milan, Italy). The mixture was transferred to a 2-mm electroporation cuvette and transfected using program A24. Immediately after transfection, ES cells were resuspended and plated in the absence of LIF on to a non-adhesive substrate for the formation of EBs. Internal controls of both transfection and effective silencing were accomplished by electroporating ES cells with 1 µg pmaxGFP plasmid (Amaxa Biosystem) encoding the transcript for green fluorescent protein (GFP) with or without 1.5 µg of a validated siRNA targeting GFP protein (Amaxa Biosystem). Negative controls were transfected with 1.5 µg siRNA targeting GFP protein but not with pmaxGFP plasmid. The following validated siRNA duplexes were used: siRNAa, ACUUAAGCCAAAUAAUUAA)dTdT (sense) and UUAAUUAUUUGGCUUAAGU)dTdG (antisense); siRNAb, UCAUUAAGUUAGAAGUAUA)dTdT (sense) and UAUACUUCUAACUUAAUGA)dTdT (antisense).


Induction of Dkk-1 in EBs differentiating in response to RA

Mouse D3 ES cells were cultured for 2 days under non-differentiating conditions (in medium containing 10% FCS and 1400 U/mL LIF), and then re-plated on to a non-adhesive substrate in the absence of LIF. This resulted in the formation of EBs, which showed typical floating cell aggregates. For the induction of neural differentiation, cultured EBs were treated with 1 µm RA [applied at 4 and 6 days in vitro (DIV) according to the classical −4/+4 protocol]. Expression of Dkk-1 was initially assessed by immunoblotting. The polyclonal Dkk-1 antibody we used labelled a single band at 35 kDa. The identity of the band was confirmed using human recombinant Dkk-1 as a positive control. Expression of Dkk-1 was low in control cultures. Exposure to RA increased the expression of Dkk-1 at both 6 and 8 DIV, i.e. 48 h after the first and second addition of RA respectively (Fig. 1a). To examine the nature of this increase we measured the levels of Dkk-1 mRNA by real-time PCR. Treatment with RA induced substantial increases in Dkk-1 mRNA levels (from 4- to 6-fold) at both 6 and 8 DIV (Fig. 1b), indicating that RA induces the expression of the Dkk-1 gene. We also assessed by RT–PCR the expression of other transcripts related to the Wnt pathway in control cultures at 8 DIV and in the corresponding cultures treated with RA. Cultures constitutively expressed the transcripts of numerous isoforms of Wnt glycoproteins (Wnt1, Wnt3, Wnt3a, Wnt4 and Wnt6), all 10 FZD receptors except FZD7, and LRP5, the Wnt/Dkk-1 co-receptor. Treatment with RA induced the expression of the transcripts encoding Wnt7a and, remarkably, the Wnt/Dkk-1 co-receptor, LRP6. No expression of FZD5 mRNA was detectable in cultures treated with RA (Fig. 1c).

Figure 1.

  (a) Western blot analysis of Dkk-1 in EBs at 6 or 8 days, with RA (1 µm) applied or not either once at day 4 (6D) or twice at day 4 and 6 (8D). The identity of the band was confirmed by the use of hrDkk-1 as a control. (b) Quantitative PCR analysis of Dkk-1 mRNA in EBs at 6 or 8 days treated or not with RA. Values were normalized with respect to the amount of β-actin mRNA. The increase in Dkk-1 mRNA induced by RA was substantial although RA also induced a 2-fold increase in β-actin mRNA levels (not shown). Values are mean + SEM of three determinations (each from triplicates). *p < 0.05 versus respective values obtained with EBs not treated with RA (one-way anova + Fisher's PLSD (protected Least Significant Difference Parametric) (c) RT–PCR analysis of Wnts, FZD receptors and LRP5and LRP 6 in EBs treated or not with RA (1 µm). The mouse embryo at day 15 (E15) is shown as a control. The presence of a single β-actin product excluded contamination by genomic DNA (see Methods).

To examine whether induction of Dkk-1 by RA was associated with inhibition of the canonical Wnt pathway, we assessed β-catenin levels by immunoblotting. Activation of the canonical Wnt pathway increases the stability and nuclear translocation of β-catenin (see Introduction and references therein). A 4-day exposure of ES cells to RA substantially reduced β-catenin levels in isolated nuclear fractions (Fig. 2), suggesting that the differentiation programme induced by RA is associated with inhibition of the canonical Wnt pathway in our cell system model.

Figure 2.

 Reduction of nuclear β-catenin levels in EBs treated with RA or recombinant human Dkk-1. Values were normalized with respect to the amount of proteins loaded on the gel and are the mean + SEM of six determinations. *p < 0.05 versus control EBs (CTR) (one-way anova + Fisher's PLSD).

Exogenous hrDkk-1 induces the expression of nestin and Dlx-2 in EBs

We compared the effect of RA and hrDkk-1 (Chemicon, CA, USA) on the expression of two established markers of neural differentiation, the neurofilament protein nestin and the transcription factor Dlx-2. Our batch of hrDkk-1 was active in inhibiting the Wnt pathway in a recombinant assay, i.e. in undifferentiated PC12 cells co-transfected with a construct encoding a reporter gene controlled by Wnt-responsive elements (see Methods) and a second construct encoding Wnt7a. In this assay, hrDkk-1 (100 ng/mL) reduced Wnt7a-stimulated gene expression by 55–60% (not shown). Human recombinant Dkk-1 (100 ng/mL) was applied to EBs twice at day 4 and 6. This treatment led to a reduction in nuclear β-catenin levels comparable to that seen in response to RA (Fig. 2), indicating that hrDkk-1 was able to inhibit the canonical Wnt pathway in EBs.

Expression of Dlx-2 was assessed by immunoblotting. Dlx-2 levels were normalized with respect to the overall amount of loaded proteins because the putative housekeeping proteins tested (β-actin and GAPDH) changed their expression in EBs differentiating in response to RA (not shown). As expected, a 4-day exposure to RA substantially increased the expression of Dlx-2 in differentiating EBs (Fig. 3a). The action of RA was mimicked by hrDkk-1, which increased Dlx-2 expression in a concentration-dependent manner (Fig. 3b). Nestin expression was assessed by fluorescent immunostaining and confocal microscopy. Figure 4 shows that both RA and hrDkk-1 substantially increased the expression of nestin in EBs, and that the action of hrDkk-1 was concentration-dependent in the range 1–100 ng/mL.

Figure 3.

 Treatment with either RA (a, b) or human recombinant Dkk-1 (b) increased Dlx-2 levels in EBs. (a) EBs were exposed to RA for 2 or 4 days (D); (b) EBs were exposed to RA or Dkk-1 for 4 days. Values were normalized with respect to the amount of protein loaded on the gel by staining with indian ink, and are mean + SEM of five and six determinations in (a) and (b) respectively. *p < 0.05 versus control EBs (CTR) (one-way anova + Fisher's PLSD).

Figure 4.

 Confocal analysis of nestin in EBs treated for 4 days with RA or human recombinant Dkk-1. Objective 30 ×.

Induction of Dkk-1 is required for the increase in Dlx-2 and β-III tubulin expression induced by RA in EBs

To examine whether expression of endogenous Dkk-1 contributes to the induction of Dlx-2 in differentiating EBs, we used either Dkk-1 antisense oligonucleotides or two different validated siRNAs targeting Dkk-1 mRNA. Antisense oligonucleotides and the corresponding sense oligonucleotides (both at 1.8 µm) were applied twice to cultured EBs at 4 and 6 days in combination with RA. The two siRNAs were transfected before the differentiation of ES cells into EBs. Cultured undifferentiated ES cells at 2 DIV were transfected with 1.5 µg/mL of either of the two siRNAs targeting Dkk-1 and cells were then replated in the absence of LIF for the induction of EBs. The efficiency of this method was validated by transfecting an anti-GFP siRNA in ES cells enforced to express GFP. This treatment markedly reduced the expression of GFP in EBs (assessed by confocal microscopy; not shown). Dkk-1 antisenses and siRNAs were tested only in EBs exposed to RA because the constitutive Dkk-1 expression in EBs was very low. Treatment with Dkk-1 antisense oligonucleotides or siRNAs markedly reduced the expression of Dkk-1 and Dlx-2 induced by RA (Figs 5 and 6). The sense oligonucleotide did not produce any of these effects (Figs 5a and b). The anti-GFP siRNA, transfected in mock ES cells, was used as a negative control for the silencing experiments (data not shown). Taken together, these data indicate that endogenous Dkk-1 contributes to the expression of Dlx-2 in EBs differentiating in response to RA.

Figure 5.

 Dkk-1 antisense oligonucleotides (AS, 1.8 µm, applied twice at day 4 and 6) prevented the increase in Dkk-1 and Dlx-2 induced by RA in EBs. S, sense oligonucleotides. Values are mean + SEM of five determinations. *p < 0.05 versus control EBs (CTR) (one-way anova + Fisher's PLSD).

Figure 6.

 Dkk-1 siRNAa and siRNAb applied to ES cells before their differentiation into EBs prevented the increase in Dkk-1 and Dlx-2 induced by RA in EBs. Values are mean + SEM of five determinations. *p < 0.05 versus all other values. (one-way anova + Fisher's PLSD).

As Dlx-2 is also expressed by non-neural tissues (Qiu et al. 1995, 1997; Thomas et al. 1997), we extended the study to the selective neuronal marker β-III tubulin in EBs exposed to RA after treatment with Dkk-1 siRNA. Treatment with RA in control cultures substantially increased the expression of β-III tubulin, as shown by immunocytochemistry (Fig. 7a) and western blot analysis (Fig. 7b). This effect was no longer observed in cultures pretreated with the two Dkk-1 siRNAs (a and b), which were proven to markedly reduce the expression of Dkk-1 (Fig. 7).

Figure 7.

 Dkk-1 siRNAa and siRNAb applied to ES cells before their differentiation into EBs prevented the increase in β-III tubulin induced by RA in EBs. (a) Immunohistochemical analysis of control EBs (CTR), EBs challenged with RA, and EBs challenged with RA after being pretreated with Dkk-1 siRNA b. (b) Immunoblot analysis of β-III tubulin in EBs treated with RA alone or with RA combined with Dkk-1 siRNAa or siRNAb. Values are mean ± SEM of four determinations. *p < 0.05 versus all other values (one-way anova + Fisher's PLSD).


ES cells are pluripotent, self-renewing cells that can differentiate into all cell lineages in response to inductive cues (Evans and Kaufman 1981; Martin 1981). The molecular events that regulate the fate of ES cells are beginning to be elucidated. A growing body of evidence suggests that the Wnt/β catenin pathway has a critical role in processes that determine the fate of ES cells. The Wnt pathway converges on LIF in supporting pluripotency and preventing differentiation of cultured mouse ES cells (Hao et al. 2005; Ogawa et al. 2006). Activation of the Wnt pathway up-regulates the transcript for Stat3, which mediates the inductive role of LIF receptor on pro-pluripotency genes (Hao et al. 2005), and activation of Wnt signalling by loss-of-function mutations of the adenomatosus polyposis coli protein or by a dominant-active form of β-catenin stimulates the expression of Bone morphogenetic protein BMP-4 and -7, thereby inhibiting neural differentiation in mouse ES cells (Haegele et al. 2003). In addition, lithium chloride, a drug that inhibits GSK3β and mimics the activation of the canonical Wnt pathway (i) sustains the expression of the pluripotent state-specific transcription factors Oct3/4, Rex-1 and Nanog in both mouse and human ES cells (Sato et al. 2004), and (ii) maintains phosphorylation of SMAD2/3, thus converging on transforming growth factor-β in supporting self-renewal of human ES cells (James et al. 2005). Wnt3a, a member of the Wnt family that signals by increasing nuclear translocation of β-catenin, sustains self-renewal of ES cells, whereas Wnt11, which signals through β-catenin-independent pathways, does not (Singla et al. 2006). In apparent contrast to these findings, increased β-catenin signalling is sufficient to induce neurogenesis in high-density ES cell culture, and RA fails to induce neurogenesis in the absence of efficient β-catenin signalling (Otero et al. 2004). In addition, Wnt has a pro-neurogenetic activity on neural stem cells in the subgranular zone of the hippocampal dentate gyrus both in vitro and in vivo (Lie et al. 2005), and overexpression of Wnt7a promotes differentiation of neural progenitor cells in the developing mouse neocortex (Hirabayashi et al. 2004). Thus, the role of the canonical Wnt pathway in the biology of stem/progenitor cells remains controversial and appears to be strictly cell and context dependent.

Here, we have shown that the Wnt antagonist Dkk-1 is induced by RA in mouse ES cells, and that its induction is required for ES cell differentiation towards a neural phenotype. Remarkably, exogenously applied hrDkk-1 induced the expression of nestin and Dlx-2 even in the absence of RA, whereas Dkk-1 knockdown largely reduced the expression of Dlx-2 induced by RA. We also showed that Dkk-1 knockdown suppressed the induction of the selective neuronal marker β-III tubulin, found in response to RA. This gives specificity to our data because Dlx-2 is also expressed by non-neural tissues, such as branchial arch ectomesenchyme (Qiu et al. 1995, 1997; Thomas et al. 1997), where Wnts and Dkks are expressed and functional (Brault et al. 2001; Cobourne and Sharpe 2003; Nie 2005; Fjeld et al. 2005). An increase in Dkk-1 gene expression is also found in undifferentiated ES cells grown in adhesive cultures and exposed for 2 days to RA or treated with siRNAs directed against the pluripotency-sustaining factors Oct4 and Nanong (Loh et al. 2006). Dkk-1 and an additional Wnt antagonist, Disabled-2, are also induced in differentiating F9 teratocarcinoma cells (Shibamoto et al. 2004). Thus, there is a convergence of data showing that Dkk-1 induction is associated with neural differentiation. In addition, our data indicate that induction of Dkk-1 and the ensuing inhibition of the Wnt pathway are obligatory components of the differentiation programme triggered by RA.

Finally, it is intriguing that blockade of Wnt signalling by Dkk-1 contributes to neural differentiation of mouse ES cells (present data) and also causes degeneration of differentiated neurones under ischaemic conditions or in response to excitotoxins or β-amyloid (Caricasole et al. 2004; Cappuccio et al. 2005). This supports the idea that developmental pathways that regulate the fate of stem/progenitor cells are recapitulated during processes of degeneration in mature neurones.