WNT/β-catenin signaling mediates human neural crest induction via a pre-neural border intermediate

Abstract

Neural crest (NC) cells arise early in vertebrate development, migrate extensively and contribute to a diverse array of ectodermal and mesenchymal derivatives. Previous models of NC formation suggested derivation from neuralized ectoderm, via meso-ectodermal, or neural-non-neural ectoderm interactions. Recent studies using bird and amphibian embryos suggest an earlier origin of NC, independent of neural and mesodermal tissues. Here, we set out to generate a model in which to decipher signaling and tissue interactions involved in human NC induction. Our novel human embryonic stem cell (ESC)-based model yields high proportions of multipotent NC cells (expressing SOX10, PAX7 and TFAP2A) in 5 days. We demonstrate a crucial role for WNT/β-catenin signaling in launching NC development, while blocking placodal and surface ectoderm fates. We provide evidence of the delicate temporal effects of BMP and FGF signaling, and find that NC development is separable from neural and/or mesodermal contributions. We further substantiate the notion of a neural-independent origin of NC through PAX6 expression and knockdown studies. Finally, we identify a novel pre-neural border state characterized by early WNT/β-catenin signaling targets that displays distinct responses to BMP and FGF signaling from the traditional neural border genes. In summary, our work provides a fast and efficient protocol for human NC differentiation under signaling constraints similar to those identified in vivo in model organisms, and strengthens a framework for neural crest ontogeny that is separable from neural and mesodermal fates.

Keywords: Embryonic stem cells; Human; Neural border; Neural crest; β-Catenin signaling.

Fig. 1.

WNT-mediated induction of multipotent NC-like cells from hESCs. (A) Scheme for maintaining and differentiating hESCs into NC cells. Asterisk indicates the enzymatic passage and subsequent plating in differentiation medium to start differentiation. (B) SOX10 and PAX7 immunostaining (green) and DAPI nuclear staining (blue) in dissociated ESCs at day 0, day 5 control (D5–CHIR) and CHIR-treated (D5+CHIR) cultures. (C) SOX10⁺, PAX7⁺ and TFAP2A⁺ nuclear counts in D5+CHIR cultures. (D) Triple-immunostaining for SOX10 (green), PAX7 (red) and TFAP2A (violet) in D5+CHIR cultures. (E) RT-qPCR analyses of NC markers at day 5 with expression levels normalized to ESC controls (day 0). (F,G) Daily time course analyses from day 0 (D0) to day 5 (D5) for ESC/pluripotency (F) and neural border (NB) (G) markers. **P≤0.01, ****P≤0.0001. (H) RT-qPCR analyses of NC transcripts in +CHIR cultures treated with SB 431542 (SB) for 0, 3 or 5 days. (I) Differentiation of NC derivatives. Alcian Blue and Alizarin Red staining in day 30 cultures differentiated under chondrogenic and osteogenic conditions, respectively. Markers associated with peripheral sensory neurons (PRPH, HuC/D, TUJ1 and ISL1), glia (GFAP and S100β) and melanoblasts (MITF, SOX10; Mica et al., 2013) were detected after culturing +CHIR hNC with SU-5402, CHIR99021 and DAPT for an additional week (13 days total). Error bars represent s.e.m. Data were pooled from three or more independent experiments.

Fig. 2.

WNT/β-catenin signaling is required for human NC induction. (A) Time course of expression (RT-qPCR) for AXIN2 in hESCs treated with (+) or without (−) CHIR. *P≤0.05, **P≤0.01, ****P≤0.0001. At least three independent experiments performed for panel A data; error bars represent s.e.m. (B) Western blot of β-catenin and β-actin expression in control (luciferase) and CTNNB1 shRNA-treated ESCs (ladder sizes indicated). (C,D) RT-qPCR analysis of CTNNB1 and OCT4 expression in control and CTNNB1 shRNA-treated ESCs. (E) Effect of control or CTNNB1 shRNA on NC induction (D5+CHIR cultures) analyzed by SOX10 immunostaining. (F,G) RT-qPCR analyses of NC (F) and non-neural ectodermal (G) genes in D5+CHIR cultures treated with control or CTNNB1 shRNA. Results from one representative experiment (out of two independent experiments) presented for CTNNB1 knockdown; error bars represent s.d. in C,D,F,G.

Fig. 3.

WNT restricts non-neural ectoderm fate in differentiating hESCs. (A) Quantitative gene expression analysis of placodal and/or surface ectoderm associated transcripts (ECAD, FOXC1, EYA2, ISL1) in ESCs (ES), D5+CHIR or D5−CHIR cultures. (B-E) Quantitative gene expression analysis in control ESCs, −CHIR and +CHIR cultures from day 1 to day 5, for competent non-neural ectoderm (DLX5 and GATA3; B), preplacodal ectoderm (SIX1; C), epidermal progenitor (ΔNP63; C) and prospective neural border (TFAP2A and ZIC1; D,E) genes. (F) Model depicting the stage-wise in vitro differentiation of placodal/epidermal ectoderm and NC cells. EPI, epidermal progenitors; NNE, non-neural ectoderm; PPE, preplacodal ectoderm. All panels normalized to ESCs except for ECAD in panel A, which is normalized to day 5 CHIR-treated cultures. Error bars represent s.e.m. Data were pooled from three or more independent experiments. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.

Fig. 4.

BMP inhibition represses NC and promotes neural induction. (A) NC induction revealed by SOX10 immunostaining was monitored on D5+CHIR cultures exposed to the BMP inhibitor noggin (NOG, 300 ng/ml) for the indicated intervals. (B) Quantification of SOX10⁺ cells in cultures treated with NOG as indicated in A. One-way ANOVA was carried out with Tukey's multiple comparison tests. (C) Triple immunostaining for SOX10 (green), PAX7 (red) and TFAP2A (blue) of D5+CHIR control cultures (No NOG), or with one day of NOG (D0-1) treatment. DAPI staining (white) is shown in insets. (D,E) Quantitative gene expression analysis for NC (D) and neural (E) transcripts in D5+CHIR control cultures treated with NOG for the indicated intervals. One-way ANOVA was carried out with Fisher's LSD test. Asterisks above bars indicate statistical significant difference of those treatments against CHIR controls. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001. Error bars represent s.e.m. for data pooled from three or more independent experiments.

Fig. 5.

Exogenous BMP4 suppresses NC but de-represses the non-neural ectoderm. (A-C) Quantitative gene expression analyses for neural border (PAX3 and PAX7; A), neural-related (PAX6 and SOX2; B) and non-neural ectoderm (ECAD, GATA3, DLX5 and TFAP2A; C) genes in day 3 +CHIR cultures treated with BMP4 (20 ng/ml) for the indicated intervals (expression normalized to +CHIR). Error bars represent s.e.m. for data pooled from three or more independent experiments. (D) SOX10 (green), PAX7 (red) and TFAP2A (turquoise) triple labeling of D5+CHIR cultures treated with increasing amount of BMP4 administered from day 0 to day 1.

Fig. 6.

WNT-induced human NC cells arise independently of PAX6⁺ neuroepithelial cells. (A) Immunostaining of OCT4 (violet), SOX10 (green) and PAX6 (red) in day 1, day 3 and day 5 cultures treated with (+) or without (−) CHIR. Framed regions, from day 3 (Aa, Ab) and day 5 (Ac) cultures are shown as magnified images; arrows highlight the few PAX6⁺ cells in day 3 +CHIR cultures. (B) Quantification of SOX10⁺ and PAX6⁺ nuclei in day 1, 3 and 5 cultures treated with CHIR. (C) Western blot of PAX6 protein in control and PAX6 shRNA-treated cells, with ladder sizes indicated. (D,E) Quantitative gene expression analysis showing effective knockdown of PAX6 at day 3 (D), but no reduction of NC markers at day 5 in PAX6-specific shRNA-treated cells (E). *P≤0.05, ***P≤0.001, ****P≤0.0001. Error bars represent s.d.

Fig. 7.

Exogenous and endogenous FGF signaling is required for mesoderm and NC induction, respectively. (A,B) Quantitative gene expression analysis for the effects of the FGF inhibitor PD 173074 (PD17, 200 nM) on the FGF targets SPRY1 and SPRY2 (D3+CHIR cultures; A) and NC markers (D5+CHIR; B). (C) Quantification of brachyury (T) and TBX6 expression in day 3 CHIR-treated cultures alone, or supplemented with 20 ng/ml FGF2 (CHIR+FGF2) from day 0 to 3. (D) Brachyury (T) immunostaining of day 3 +CHIR cultures alone or supplemented with 20 ng/ml of FGF2 (CHIR+F2) from day 0 to day 3. Error bars represent s.d.; three independent culture experiments were performed and data from one representative experiment are shown. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.

Fig. 8.

Identification of a class of early pre-border genes and their signaling requirements for expression. (A) Double immunostaining of PAX7 (red) and OCT4 (turquoise) in day 3 +CHIR cultures. Yellow arrows denote nuclei expressing PAX7 but with no OCT4 staining. White arrowheads indicate nuclei expressing PAX7 and OCT4. (B-E) Time course of quantitative gene expression analysis of GBX2, SP5, ZIC3 and ZEB2 in ESCs (ES), control cultures (−CHIR) and with CHIR-treated (+CHIR) cultures. (F-H) Differential response of pre-neural border (pB: GBX2, SP5, ZIC3, ZEB2) and neural border genes (NB: PAX7, PAX3, MSX1, TFAP2A) genes to WNT, BMP and FGF signaling analyzed by RT-PCR in D3+CHIR cultures exposed to specific modulators. Statistical analyses were carried out using two-way ANOVA with Fisher's tests. β-catenin knockdown (CTNNB1 shRNA) leads to reduced expression of both pB and NB genes (F). BMP inhibition (NOG, from day 0 to day 1) represses NB but not pB genes (G). FGF modulation by FGF2 ligand led to enhanced or unaffected pB expression and reduced NB expression (H). (I) A step-wise induction model for human NC and a gene regulatory network of human NC with signaling inputs from canonical WNT, FGF and BMP signaling. *P≤0.05, **P≤0.01, ***P≤0.001, ****P≤0.0001.