Establishment and neural differentiation of neural crest‐derived stem cells from human dental pulp in serum‐free conditions

Abstract The potential of obtaining cell cultures with neural crest resemblance (neural crest‐derived stem cells [NCSCs]) from dental‐related tissues, including human dental pulp cells (hDPCs), has been discussed in the literature. However, most reports include the use of serum‐rich conditions and do not describe the potential for neural differentiation, slowing translation to the clinic. Therefore, we aimed to culture and characterize NCSCs from the human dental pulp in vitro and evaluate their ability to differentiate into neurons; we also investigated the effectiveness of the addition of BMP4 to enhance this potential. Cultures were established from a varied cohort of patient samples and grown, as monolayers, in serum, serum‐free, and also under sphere‐aggregation conditions to induce and identify a NCSC phenotype. hDPC cultures were characterized by immunocytochemistry and reverse transcription quantitative polymerase chain reaction. Monolayer cultures expressed stem cell, neural progenitor and neural crest‐related markers. Culturing hDPCs as neurospheres (hDPC‐NCSCs) resulted in an increased expression of neural crest‐related genes, while the addition of BMP4 appeared to produce better NCSC characteristics and neural differentiation. The neural‐like phenotype was evidenced by the expression of TUJ1, peripherin, NFH, TAU, SYN1, and GAP43. Our results describe the establishment of hDPC cultures from a large variety of patients in serum‐free medium, as NCSC that differentiate into neural‐like cells, as well as an important effect of BMP4 in enhancing the neural crest phenotype and differentiation of hDPCs.

ity to differentiate into neurons; we also investigated the effectiveness of the addition of BMP4 to enhance this potential. Cultures were established from a varied cohort of patient samples and grown, as monolayers, in serum, serum-free, and also under sphere-aggregation conditions to induce and identify a NCSC phenotype. However, a factor that can delay translation of OSCs into the clinic is that the isolation, culture, and characterization of these stem cell populations have been commonly performed in medium containing fetal bovine serum (FBS; recently reviewed by Luo et al 2 and Rodas-Junco et al 3 ), which is inappropriate for most applications in man. Hence, there is a need for protocols to establish cell cultures in conditions that would be more appropriate for advancing their translational use. 4,5 Thus, it is important to develop methods that would allow the establishment of cultures in serum-free and/or xeno-free conditions that could comply with clinical-grade manufacturing standards in the near future. At the same time, there is a need to describe in more detail the possibility of using these cells from samples obtained from a wide variety of patients that could realistically resemble common clinical scenarios.
Most dental-related stem cell sources are believed to derive from the neural crest, 6 and because of this there is an increasing interest in the study of their neural crest-derived stem cell (NCSC) characteristics. [7][8][9] In contrast to MSCs from non-neural crest origin, these NCSCs could prove particularly useful to restore neural cell types due to their molecular resemblance to embryonic neural crest cells. 10,11 In this regard, there is little information about the identification and isolation of a human NCSC population specifically from human dental pulp cells (hDPCs) of adult, permanent teeth, and only a small number of key direct applications have been reported. [12][13][14] Thus, it is highly relevant to further our understanding of NCSCs from hDPCs, and also to test their proposed potential in relevant models of nerve regeneration. Therefore, we present here a detailed description of the establishment and characterization of hDPCs grown in serum-free media in vitro, the characterization of their molecular NCSC signature, and the role of BMP4 in enhancing this signature. In addition, we present proof of concept that these NCSCs, generated under serum-free conditions from hDPCs, can derive neural-like cells, proposing them as candidates for neural regeneration treatments.

| Dental pulp collection
The study was conducted in accordance with ethical approval granted by the Leeds East Research Ethics Committee of the UK National Research Ethics Service (reference: 15/YH/0308; protocol: STH19019).
Human dental pulp was isolated from samples collected with written informed consent from patients at the Charles Clifford Dental Hospital, Sheffield, UK. Details of the patients' age and the condition and position of the tooth are provided in Table 1. Immediately after extraction, the samples were transported in ice-cold PBS with 1× Penicillin-Streptomycin (1× P/S; 100 IU/mL-100 μg/mL; GIBCO, Paisley, UK) to the laboratory for processing.

| Calculation of cumulative cell doublings
The number of living cells was obtained by trypan blue staining, and cell number was counted using the automated cell counter TC20 (Biorad, Hertfordshire, UK; particle size range 9-21 μm). cumulative cell doubling (CCD) of living cells was calculated using the following formula:

| Neurosphere sectioning
After fixation (see below), hDPC neurospheres were subjected to a sucrose gradient of 7.5%, 15%, 22%, and 30% sucrose (12-24 hours each; 4 C), and transferred to optimal cutting temperature (OCT) solution for 24 hours (4 C). The tissue was then fast-frozen on a dry ice/methyl-butane bath. Cryosectioning was then performed in a Bright OTF5000 cryostat in sections of 5 to 10 μm.

| Direct RT-qPCR preparation
hDPC-FBS transferred into the other culture conditions (FBS, OSCFM, and BMP4) and induced to neural differentiation (refer to Figure 6D) were analyzed using a Cells to CT kit (Thermofisher) according to the manufacturer's instructions.

| Quantitative polymerase chain reaction
Reverse transcription quantitative polymerase chain reaction (RT-qPCR) was performed with Taqman probes using the Thermofisher  The graphs shown in Figure 1A-F represent the percentage of successful cultures. The highest success rate of surviving cultures was FBS (93%, n = 15/16), followed by OSCFM (81%, n = 9/11) and BMP4 (35%, n = 5/14; Figure 1A). To consider common situations in the clinical setting, we also included mildly carious teeth in our experi-  Figure 1B).

| hDPC cultures express neural crest-related markers
We aimed to define the core molecular signature of hDPCs when cells were grown in two-dimensional, monolayer cultures in the three different conditions. A relative expression analysis by RT-qPCR was per-

| NCSC markers are enhanced when hDPCs are grown as spheres
Due to the evident changes in sub-populations occurring during expansion as monolayers ( Figure 2C,D), it was hypothesized that culturing hDPCs as 3D aggregates could drive them into a more stable, characterization.
An equal number of cells from each initial culture condition were transferred to low attachment culture plates, and sphere formation (as number of spheres per well and diameter) was assessed after 4 days in culture ( Figure 3A,B). Overall, no significant difference was observed in sphere number among our culture conditions, and a statistically significant smaller diameter was observed in spheres derived from BMP4 cultures ( Figure 3C).   Figure 4A,B). Interestingly, SOX2 was expressed by neurospheres from all culture conditions, while appeared undetected by RT-qPCR in the earlier monolayer cultures.
To support the observation made by immunocytochemistry, we also studied the expression of SNAIL1, P75, SOX10, and AP2a quantitatively by RT-qPCR. To do this, we analyzed the relative expression levels of each neurosphere sample in comparison to their own monolayer counterpart. Overall, P75 and SNAIL1 expression was significantly higher in the majority of independent neurospheres from all conditions ( Figure 4D), whereas SOX10 and AP2a were upregulated only in a few samples grown as neurospheres ( Figure 4D) , and TAU was analyzed by immunocytochemistry in NCSC induced to neural differentiation (NEU) and compared to their undifferentiated control (CTL). Neural morphology and an apparently stronger signal were observed in hDPC-NEU cells compared to controls (A-C). Only BMP4-NEU showed expression of the four analyzed markers (C). Scale bar = 100 μm (A-C). D, hDPC-FBS cells were transferred to OSCFM and BMP4 culture conditions before neurosphere formation and were subject to neural differentiation. The gene expression of neural related markers GAP43, NFH, and SYN1 were upregulated mainly in the neural differentiated cells from hDPC-FBS transferred to the BMP4 condition (FBS-BMP4-NEU). One-way ANOVA followed by Tukey's test for pairwise comparison. *P < .05, ****P < .0001. BMP4, bone morphogenic protein 4; CTL, control; FBS, fetal bovine serum; hDPC, human dental pulp cells; hDPC-FBS, hDPC grown in 20% FBS, hDPC-OSCFM, hDPC grown in OSCFM; hDPC-BMP4, hPDC grown in OSCFM supplemented with BMP4; NCSC, neural crest-derived stem cells; OSCFM, otic stem cell full medium sphere formation rather than during culture establishment ( Figure 5A).
Two hDPC-FBS cultures and one hDPC-OSCFM culture (also included in the previous analysis; Figure 5B) were supplemented with BMP4 during sphere aggregation. RT-qPCR was used to determine any particular change in the NCSC phenotype and compared to the neurosphere conditions described above, as well as the cells grown as monolayers (baseline control, CTL). As seen before, AP2a showed the most limited upregulation in which only the 53OSCFM neurospheres showed a positive regulation when supplemented with BMP4 ( Figure 5B). HNK-1 presented a significant upregulation in two of the three independent samples analyzed. Interestingly, the three hDPC cultures selected for this experiment showed a significant upregulation of SNAIL1 and P75 expression in comparison to both CTL and neurospheres without BMP4 (except P75-48FBS Sphere vs 48FBS sphere+BMP4, where the difference showed a trend but was not large enough to be significant).
All the evidence together suggests that the addition of BMP4 to the neurospheres can induce even higher relative expression levels of important neural crest-related genes.

| hDPCs-NCSCs differentiate in vitro into neural-like cells
hDPC-NCSCs (neurospheres) derived from every initial experimental condition ( Figure 3A) were induced to differentiate into neurons by culturing them for up to 2 weeks with dbcAMP and NT3. As control, equivalent hDPC-NCSCs were transferred to the neural differentiation media, without dbcAMP and NT3. Compared to the undifferentiated controls, many hDPCs-NCSCs-derived cells presented a neuronal-like morphology extending projections and a qualitatively higher intensity of the neuronal markers TUJ1 and peripherin (PER; Figure 6A-C). NFH was also expressed in the neural differentiated group colocalizing with TUJ1 in cultures from hDPC-OSCFM and hDPC-BMP4 ( Figure 6B,C), but not from hDPC-FBS cultures ( Figure 6A). Furthermore, TAU expression was only detected in neuronal differentiated cells from hDPC-BMP4 cultures ( Figure 6C). Notably, cells with non-neuronal morphology were frequently visible in neural induced cultures; additionally, cells were in some cases unable to grow out from the neurospheres or were nonviable (not shown).

| Neural differentiation from hDPC-FBS cultures can be enhanced when transferred to OSCFM and BMP4 culture conditions
To further explore the ability of BMP4 to support neurogenic cultures, hDPC-FBS cells were transferred to OSCFM and BMP4 conditions for 7 days and then induced to neurosphere formation and neural differentiation ( Figure 6D). RT-qPCR was used to analyze the neural-related genes GAP43, NFH, and SYN1 after neural differentiation ( Figure 6D).
As a control, we used the parental hDPC-FBS monolayer culture ( Figure 6D)  Figure 6D).

| DISCUSSION
In the present work, we have described the establishment of dental pulp cultures from a wide variety of tooth sample conditions and ages.
The range of samples presented allowed us to extract critical data in terms of efficiency, patient age, and sample condition, which should be considered for the translational research of hDPCs, and which are not commonly reported from such a varied sample cohort as this one.
One of the most complete reports previously available, described a set of 40 samples from patients aged 18 to 30 years, showing 78% success rate to grow in vitro (but under standard FBS conditions), 18 which is similar to the overall efficiency found in the present work.
We have accounted for the presence of caries in our samples, and have shown that it is possible to extract viable cells of similar identity from affected teeth, although at a reduced efficiency (  Notably, the characterization of dental-related human NCSCs, including those derived from gingival tissue, periodontal tissue, exfoliated deciduous teeth, and adult human dental pulp, has gained increasing interest for their potential use in neurodegenerative disorders. Nevertheless, most reports are limited to describing the molecular resemblance to neural crest cells, while their actual differentiation capacity is usually left untested, masking their actual translational potential for regenerative medicine. [7][8][9]32,33,37,44 Additionally, it would be important to evaluate a combination of strategies to facilitate the application of oral-derived NCSCs in regenerative medicine, such as their incorporation in advanced manufactured scaffolds or conduits for nerve repair and tissue engineering (eg, bone regeneration). 9,45,46 Finally, the field of dental pulp stem cells and their applications would benefit from addressing specific diseases or targeted applications from preclinically relevant established cultures if we aim to advance the use of dental pulp cells in regenerative medicine.
We have provided a detailed description of the culture of hDPC from a wide variety of teeth, as well as a robust characterization of the culture conditions. We have also shown that cells can be directly established in serum-free medium in contrast to the widely used serum-rich standard conditions. This should prove highly advantageous in a preclinical setting. Also, hDPCs present a basal neural crest-like phenotype that can be enhanced when induced to form neurospheres (hDPC-NCSC). Furthermore, BMP4 has a significant role in enhancing the neural crest-like phenotype in human dental pulp. It is worth noting, however, that there is a substantial level of heterogeneity and patient-topatient variation that should be considered for future clinical applications.
Altogether, the present work details the culture and characterization of NCSCs from a broad cohort of human dental pulp samples

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available from the corresponding author upon reasonable request.