Fine‐tuning of dual‐SMAD inhibition to differentiate human pluripotent stem cells into neural crest stem cells

Abstract Objectives The derivation of neural crest stem cells (NCSCs) from human pluripotent stem cells (hPSCs) has been commonly induced by WNT activation in combination with dual‐SMAD inhibition. In this study, by fine‐tuning BMP signalling in the conventional dual‐SMAD inhibition, we sought to generate large numbers of NCSCs without WNT activation. Materials and methods In the absence of WNT activation, we modulated the level of BMP signalling in the dual‐SMAD inhibition system to identify conditions that efficiently drove the differentiation of hPSCs into NCSCs. We isolated two NCSC populations separately and characterized them in terms of global gene expression profiles and differentiation ability. Results Our modified dual‐SMAD inhibition containing a lower dose of BMP inhibitor than that of the conventional dual‐SMAD inhibition drove hPSCs into mainly NCSCs, which consisted of HNK+p75high and HNK+p75low cell populations. We showed that the p75high population formed spherical cell clumps, while the p75low cell population generated a 2D monolayer. We detected substantial differences in gene expression profiles between the two cell groups and showed that both p75high and p75low cells differentiated into mesenchymal stem cells (MSCs), while only p75high cells had the ability to become peripheral neurons. Conclusions This study will provide a framework for the generation and isolation of NCSC populations for effective cell therapy for peripheral neuropathies and MSC‐based cell therapy.

requires the efficient generation and acquisition of homogenous neural crest stem cells (NCSCs). NCSCs are multipotent progenitors that originate from the neural tube in vertebrate embryos. 2,3 They migrate extensively and generate various cell types of the peripheral nervous system, smooth muscle, bone and connective tissue.
Previous studies have demonstrated that specification of the neuroectoderm from hPSCs requires the BMP, WNT and FGF pathways. [4][5][6][7][8][9][10][11][12][13][14] Since first being reported in 2009, the dual-SMAD inhibition method has been frequently used for the generation of neural precursor cells (NPCs) from hPSCs. 15,16 Dual-SMAD inhibition often uses SB431542 and dorsomorphin (DM), which function by blocking the TGFβ/Activin A/Nodal-SMAD2/3 and BMP-SMAD1/5/8 signalling pathways, respectively. Conventional SMAD inhibition drives hPSCs into neural rosettes, which largely consist of SOX1 + and PAX6 + NPCs; NCSCs are also found as a small population and can be isolated by cell-sorting techniques. 17 In this study, we describe the efficient generation of NCSCs from hPSCs by fine-tuning BMP signalling of the conventional dual-SMAD inhibition system commonly used for NPC generation. We detected two populations of p75-positive NCSCs, p75high and p75low cells and showed that their differentiation abilities were different.
Furthermore, we introduced a convenient method to isolate p75high cells that were free from p75low cells.
This study offers insight into the generation of NCSC populations for effective cell therapy for peripheral neuropathies and mesenchymal stem cell (MSC)-based cell therapy.

| EB formation and induction of NCSCs
hPSCs were detached with 2 mg/mL collagenase, type IV (Worthington Biochemical Corporation, Lakewood, NJ, USA), for 30 minutes at 37°C, followed by centrifugation at 300 g for 5 minutes, and the suspension was transferred to a fresh bacteriological petri dish. For NCSC generation, EBs were cultured for 4 days in DMEM/F12 containing 20% knockout serum replacement, 1% non-essential amino acids and 55 μmol/L β-mercaptoethanol (all from Invitrogen) supplemented with 10 μmol/L SB431542 (Tocris Bioscience, Bristol, UK) and 0.5-5 μmol/L DMH1 (or DM) (Merck Millipore, Burlington, MA, USA). During differentiation, the medium was changed daily. On day 4, EBs were attached to Matrigelcoated dishes in NCSC differentiation medium containing 1% N2 supplement (Invitrogen), 20 ng/mL bFGF (CHA Biotech, Pangyo, Korea) and 25 μg/mL human insulin solution (Sigma-Aldrich) and continued to differentiate for 5 more days with the medium changed every day. Isotype control antibodies (Miltenyi Biotec) were used as negative controls. p75high and p75low cells were separately sorted using BD FACSAria™ III Cell Sorter (BD Biosciences).

| Quantitative reverse transcriptase-polymerase chain reaction (qRT-PCR)
The total RNA samples were purified using the NucleoSpin RNA II kit (MACHEREY-NAGEL, Duren, Germany) following the manufacturer's instructions. One microgram of total RNA was reversetranscribed with the ReverTra Ace qPCR RT Kit (Toyobo, Osaka, Japan) according to the manufacturer's instructions. Real-time quantitative PCR was performed with SYBR™ Select Master Mix (Applied Biosystems, Foster City, CA, USA) and analysed by a StepOnePlus™ Real-Time PCR System (Applied Biosystems). The qRT-PCR conditions used in this study were as follows: (1) denaturation at 95°C for 15 seconds, (2) annealing at 60°C for 30 seconds and (3) extension at 72°C for 30 seconds. These steps were repeated for 40 cycles, followed by a final extension of 10 minutes at 72°C. The primer sequences used for PCR analysis are listed in Table S1. Glyceraldehyde 3-phosphate dehydrogenase (GAPDH) was used as a reference gene for normalization.

| Microarray analysis
Total RNA was purified using the NucleoSpin RNA II kit (MACHEREY-NAGEL) according to the manufacturer's suggestions. One microgram of total RNA was utilized for global gene expression profiling using the Illumina HumanHT-12 v4 Expression BeadChip (Illumina, San Diego, CA, USA).

| Immunocytochemistry
Cells were fixed in 4% paraformaldehyde in PBS for 10 minutes and permeabilized with 0.2% Triton X-100 (Sigma-Aldrich) for 10 minutes. Cells were blocked with 5% normal goat serum, 1% BSA and 0.1% Tween-20 (Sigma-Aldrich) in PBS for 1 hour at RT. The cells were then incubated with primary antibodies for 1 hour at 37°C or overnight at 4°C. The primary antibodies used were for p75 (Santa Cruz

| Mesodermal differentiation of NCSCs
For MSC generation, NCSCs were plated onto tissue culture dishes at a density of 5 X 10 4 cell/cm 2 in α-MEM containing 10% FBS (all from Invitrogen). The medium was changed every 2 days, and the cells were passaged every 4-5 days for 2 weeks.

| Colony-forming unit assay of MSCs
MSCs at passage 3 were seeded into 100 mm dish at a density of 3 × 10 2 cells in α-MEM containing 10% FBS (all from Invitrogen). The medium was changed every 2 days. After 14 days, the cells were washed with PBS, were fixed in 4% paraformaldehyde/PBS for 10 minutes and then were stained with 3% crystal violet in 100% methanol for 5 minutes at RT.

| Electrophysiology
Neurons cultured on individual coverslips were transferred to the recording chamber mounted on the stage of an upright microscope (Olympus, Tokyo, Japan) fitted with a 40× water-immersion objective lens. Cells were visualized either directly via the optical microscope or indirectly via a high-resolution CCD camera system

| Statistical analysis
Statistical analysis was performed using GraphPad Prism (GraphPad software, Inc, La Jolla, CA). Student's t test was used to compare 2 groups, while one-way ANOVA with Tukey's test was used to compare the multiple groups. Only P-values less than 0.05 were considered statistically significant.

| Determination of the fate of hPSCs into NCSCs by fine-tuning the BMP signalling pathway
The currently used dual-SMAD inhibition approach was shown to drive hPSCs into the neuroectoderm lineage, mostly NPCs 15,16 ; dual-SMAD inhibition occurs by blocking TGFβ and BMP pathways using In contrast, at 5 μmol/L DM (DMH1)/10 μmol/L SB431542, most of the attached EBs turned into neural rosette structures, with some NCSCs migrating away from the rosettes, eventually forming a cell F I G U R E 1 Differentiation of hPSCs into NCSCs using modified dual-SMAD inhibition. (A) A schematic diagram summarizing the differentiation of H9 hESCs into NPCs and NCSCs. (B) Chemical structures of DM and DMH1. (C) EBs generated in suspension for 4 days in EB medium supplemented with SB431542 (10 μmol/L) and DMH1 (or DM) (0.5, 2.5 and 5 μmol/L) were then attached to dishes for another 5-d culture. Spherical cell clumps and neural rosettes are marked by white and yellow arrows, respectively. Scale bar: 100 μm. (D, E) After 9 d of differentiation, cells were analysed with flow cytometry using antibodies against p75 (D) and SOX1 (E). (F) Immunostaining of the cells was performed with antibodies against p75 and SOX1 after differentiation with 10 μmol/L SB431542 and 0.5 μmol/L DMH1. Scale bar: 200 μm F I G U R E 2 Flow cytometric analysis of the cell populations present at day 9 after differentiation. H9 hESCs were differentiated into NCSCs via EB formation using modified dual-SMAD inhibition (0.5 μmol/L DMH1/10 μmol/L SB431542). Three different areas/structures (ie, spherical cell clumps ('1'), non-spherical long clumps with comet tails ('2') and monolayer cells ('3')) were dissected mechanically and subjected to flow cytometry using an antibody against p75. Independently, whole cells were also analysed by flow cytometry. Scale bar: 100 μm monolayer ( Figure 1C right panels, Figure S1 hESCs were shown to retain normal karyotypes ( Figure S3).
Intriguingly, two populations of p75 + cells, p75high and p75low cells were detected at 0.5 μmol/L DM (DMH1), while only the p75low cell population was evident at 5 μmol/L DM (DMH1) ( Figure 1D, Figure S4A,B). Both p75high and p75low cell populations were also HNK1 + , another typical marker of NCSCs, suggesting that both p75 + cell populations were NCSCs ( Figure S4A,B). In summary, our results showed that dual-SMAD inhibition drives hPSCs into mostly HNK + p75 + NCSCs or SOX1 + NPCs when the BMP level is high or low, respectively.

F I G U R E 3
Sorting and characterization of p75high and p75low populations. (A) Two p75 + populations, p75high and p75low cells, were separately isolated by FACS. The cell morphology of the two cell groups is shown (right panels). (B) Expression levels of representative neural crest genes, p75, AP2, SOX10 and SNAIL2, were analysed using quantitative RT-PCR. The glyceraldehyde 3-phosphate dehydrogenase (GAPDH) gene was used as a normalization control. The expression level of each gene in H9 hESCs was arbitrarily set to 1. *P < .05, **P < .01

| Spherical cell clumps consisted of p75high NCSCs
To examine the characteristics of the cells comprising the spherical cell clumps, we mechanically isolated the spheres and dissociated them into single cells for flow cytometric analysis. Our results showed that the spherical cell clumps mostly consisted of p75high cells ('1' in Figure 2).

| Gene expression profiles of the p75high and p75low cell populations
To examine the p75high and p75low cell populations, we isolated each cell population via fluorescence-activated cell sorting (FACS) using an antibody against p75 ( Figure 3A). No significant difference was detected in cell morphology between the two groups ( Figure 3A, right panels). Quantitative RT-PCR analysis showed that typical markers for NCSCs, p75, AP2 and SOX10 were expressed more abundantly in the p75high population than in the p75low population ( Figure 3B). Noticeably, Snail2, another neural crest marker that is implicated in epithelial-mesenchymal transitions, was expressed more significantly in the p75low cell population than in the p75high cell population ( Figure 3B).
To compare the two p75 + populations more carefully, we compared the global gene expression profiles of the H9 hESCs, NPCs, p75high cell group and p75low cell group (Figure 4). Heatmap analysis showed that substantial differences in global gene expression were evident between the two p75 + cell groups ( Figure 4A). The scatter plot indicated gene probes that were more than 4-fold different among the NPCs, p75high and p75low cell groups ( Figure 4B).
Intriguingly, neural crest-or Schwann cell-related genes were more abundantly expressed in the p75high population than in the p75low population, while no dramatic change in the expression of neural stem cell-and melanocyte-related genes was detected between the two groups ( Figure 4C). The list of genes that were differentially expressed by more than 4-fold between p75high and p75low cells are shown in Table S2: neural crest-, peripheral nervous system-or (transcription factor AP-2 gamma), were much more highly expressed in p75high cells than in p75low cells.
Taken together, these results showed that p75high cells expressed more neural crest-related genes than p75low cells.

| Differentiation of p75low and p75high cells into mesenchymal stem cells
NCSCs are multipotent and can differentiate into a variety of cell types from peripheral nervous system and mesodermal tissue lineages. 18 To examine the differentiation capability of p75high and p75low cells into MSCs, cells were cultured in DMEM/F12 growth medium containing 10% foetal bovine serum ( Figure 5A). No significant difference was detected in cell morphology between the two groups at 2 weeks post-differentiation ( Figure 5A). On day 14 post-differentiation, both flow cytometric and quantitative RT-PCR analyses showed similar expression levels of MSC markers, such as CD44, CD73 and CD105, in both p75 + cell population-derived MSCs ( Figure 5B,C). The NCSC-derived MSCs displayed efficient self-renewal ability as shown by high colony-forming efficiency ( Figure S6).
We next confirmed that MSCs derived from both p75high and p75low cell populations could be converted into adipocytes, osteocytes and chondrocytes ( Figure 5D).
Together, the results showed that both the p75high and p75low populations retained the capability to become MSCs.

| p75high cells, but not p75low cells, could be differentiated into peripheral neurons
Because gene expression profile analysis showed that p75high cells expressed neural crest-related genes more significantly than These results showed that the p75high, but not p75low, cell population retained the capability to become peripheral neurons.
Our results suggested that isolation of the p75high cell population would be required to develop an efficient cell therapy for peripheral neuropathies.

| D ISCUSS I ON
Early studies have reported the differentiation of hPSCs into NCSCs by coculture with the mouse stromal cell lines PA6 and MS5. 17,19 Later, a dual-SMAD inhibition method was developed to drive the differentiation of hPSCs into the neuroectoderm lineage. 15,16 However, dual-SMAD inhibition generates mostly neural rosettes, which mainly consist of NPCs with some p75 + NCSCs in the periphery of the neural rosettes. 15,20 To direct the differentiation of hPSCs towards NCSCs, WNT signalling activation was performed along with dual-SMAD inhibition. [21][22][23] In contrast, some studies used inhibition of only TGFβ/Activin A/Nodal signalling, not BMP signalling, in combination with WNT activation. 24,25 Most studies used glycogen synthase kinase 3 (GSK3) inhibitors, such as CHIR99021 and 6-bromoindirubin-3′-oxime (BIO), to activate Wnt pathway (21)(22)(23)(24)(25). Instead of GSK3 inhibitors, the addition of Wnt3a protein also induced NCSC specification from hPSCs, suggesting that Wnt3a was one of the Wnt family proteins involved in this process. 22,24 Although the implication of Wnt pathway in NCSC specification was well established, the role of BMP signalling modulation in NCSC derivation from hPSCs has yet to be clarified.
The presence of p75high and p75low populations has been previously reported, although the differences between these two populations have not been investigated thoroughly. [23][24][25] Strikingly, we noticed the formation of spherical cell clumps that consisted of p75high cells.
Although these spherical cell clumps have been reported previously in one study (named 'dense aggregates'), no detailed characterization of these spheres has been performed, especially in relation to p75high and p75low cell populations. 28 Our study first reported that spherical cell clumps were composed of p75high cells only.
The percentage of p75 + NCSCs among total cells resulting from our differentiation procedure was variable from experiment to experiment but was approximately 62%-87%. The percentage of the p75high cell population among the p75 + NCSCs was also variable and was approximately 30%-70% depending on experiments. The most remarkable advantage of our differentiation condition was that the p75high and p75low cell populations can be easily isolated by mechanically dissecting spherical cell clumps and monolayer cells, respectively ( Figure 2).
It would be interesting to unveil the characteristics of p75high and p75low cells, especially in terms of differentiation potential.
Intriguingly, both p75high and p75low cell populations could be differentiated into MSCs efficiently. No significant differences in MSC marker expression, proliferation or differentiation capability were detected between p75high NCSC-and p75low NCSC-derived MSCs. In contrast, strikingly, only p75high cells were able to be differentiated into peripheral neurons, while p75low cells died during neuronal differentiation. In support of this observation, global gene expression analysis revealed that p75high cells expressed more neural crest-and peripheral neural-related genes than p75high cells. To the best of our knowledge, this is the first study to demonstrate differences between p75high and p75low cell populations.
In summary, this study described a new differentiation system for the efficient induction of NCSCs from hPSCs by fine-tuning BMP signalling in a conventional dual-SMAD inhibition system used for NPC generation. Our results showed that at low concentrations of BMP inhibitor (0.5 μmol/L) during dual-SMAD inhibition, neuroectodermal differentiation was geared to p75 + NCSC generation instead of NPC generation. Notably, our NCSC specification protocol did not use GSK3 inhibitors or Wnt3a protein. Unveiling the underlying mechanisms and potential connection between BMP and Wnt pathways in NCSC specification from hPSCs is of interest and would be the future research goal.
Furthermore, the HNK + p75 + NCSCs consisted of two subpopulations, p75high and p75low cells: p75high cells can be differentiated into mesenchymal lineage and peripheral neurons, but p75low cells can be differentiated into mesenchymal lineage only.
This study provides insight into the early differentiation of NCSCs from hPSCs and offers a useful platform for the isolation of a p75high cell population to treat many neural crest-related diseases.

ACK N OWLED G EM ENTS
This work was funded by grants from the Ministry of Health and Welfare, Korea (HI16C1559).

CO N FLI C T S O F I NTE R E S T
All authors of this paper declare no potential conflicts of interest.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from the corresponding author upon reasonable request.