Differentiation of human adult-derived stem cells towards a neural lineage involves a dedifferentiation event prior to differentiation to neural phenotypes

Although it has been reported that mesenchymal stem cells isolated from adult tissues can be induced to overcome their mesenchymal fate and transdifferentiate into neural cells, the findings and their interpretation have been challenged. The main argument against this process is that the cells rapidly adopt neuron-like morphologies through retraction of the cytoplasm rather than active neurite extension. In this study, we examined the sequence of biological events during neural differentiation of human periodontal ligament-derived stem cells (hPDLSCs), human bone marrow-derived stem cells (hBMSCs) and human dental pulp-derived stem cells (hDPSCs) by time-lapse microscopy. We have demonstrated that hPDLSCs, hBMSCs and hDPSCs can directly differentiate into neuron-like cells without passing through a mitotic stage and that they shrink dramatically and change their morphology to that of neuron-like cells through active neurite extension. Furthermore, we observed micronuclei movement and transient cell nuclei lobulation concurrent to in vitro neurogenesis from hBMSCs and hDPSCs. Our results demonstrate that the differentiation of hPDLSCs, hBMSCs and hDPSCs towards a neural lineage occurs through a dedifferentiation step followed by differentiation to neural phenotypes, and therefore we definitively confirm that the rapid acquisition of the neural phenotype is via a differentiation trait.

In the classical view of cell development, embryonic stem cells differentiate into progressively more precursor cells with an increasingly restricted lineage potential, until the final mature, specialised cell types are generated and functionally integrated into their respective tissues. The differentiated state of a cell was believed to be terminal and irreversible 1,2 . While asymmetric cell division is considered to be the mechanism by which the asymmetric inheritance of cellular components during mitosis defines the distinct fate of each daughter cell 3 .
However, there is increasing evidence that the rules of irreversibility and germ-layer restriction can be broken 4 . It has long been accepted that cells can change their identity, both in vitro and in vivo, a phenomenon known as cellular plasticity 5,6 . Cellular conversion falls into two major categories: dedifferentiation and transdifferentiation [4][5][6][7][8] . A definitive, well-accepted criterion for dedifferentiation and transdifferentiation has yet to be established 9 . In its simplest form, dedifferentiation is the conversion from one differentiated cell stage to a less differentiated stage. Transdifferentiation is the conversion of one differentiated cell type into another. It has been suggested that transdifferentiation may involve a dedifferentiation step before cells differentiate to a new mature phenotype, or it may occur directly in a process that bypasses such immature phenotypes 10 .
Literature published over the past 20 years have shown that mesenchymal stem cells (MSCs) isolated from adult tissues can be induced to overcome their mesenchymal fate and transdifferentiate into neural cells [11][12][13][14][15][16][17][18] . Although the transdifferentiation of human MSCs into neurons has aroused considerable interest, as it would have immense clinical potential in cell replacement therapy and regenerative medicine 19,20 , the findings and their interpretation have been challenged 21,22 . The main argument against this process is that the cells rapidly adopt neuron-like morphologies through retraction of the cytoplasm rather than by active neurite extensión 23,24 .
In a recent study 25 , we showed that human periodontal ligament-derived stem cells (hPDLSCs) can directly differentiate into neuron-like cells without passing through any mitotic stages. When hPDLSCs were exposed to a neural induction medium, we found that they rapidly underwent a dramatic change in shape and size; they initially adopted highly irregular forms before gradually contracting into round cells (neurogenesis). These round cells then slowly adopted a complex morphology and finally gave rise to a variety of neuron-like morphologies (neuronal polarisation). Furthermore, we also reported nuclear remodelling concurrent to in vitro neurogenesis from hPDLSCs.
The present study is designed to confirm that human stem cells isolated from adult tissues in the early stages of exposure to a neural induction medium adopt a neuron-like morphology because of a differentiation trait rather than an artefact. We performed time-lapse phase-contrast microscopy to record the changes in the cell morphology of hPDLSCs, human bone marrow-derived stem cells (hBMSCs) and human dental pulp-derived stem cells (hDPSCs).
We found that hPDLSCs, hBMSCs and hDPSCs can directly differentiate into neuron-like cells without passing through a mitotic stage. All three types of stem cell shrunk dramatically and changed their morphology to that of neuron-like cells with active neurite extension. Additionally, we may have evidenced micronuclei movement and transient cell nuclei lobulation in parallel with in vitro neurogenesis from hBMSCs and hDPSCs. Our results clearly indicate that the differentiation of hPDLSCs, hBMSCs and hDPSCs towards a neural lineage occurs through a dedifferentiation step before the cells differentiate to neural phenotypes. These findings definitively establish that the rapid acquisition of a neuron-like morphology during neural differentiation is due to a differentiation trait as opposed to an artefact.

Results
Time-lapse microscopy of hPDLSCs cultured in neural induction media. To conclusively affirm that hPDLSCs adopt a neuron-like morphology in the early stages of exposure to a neural induction medium is due to active neurite extension rather than retraction of the cytoplasm, we performed time-lapse phase-contrast microscopy and immunocytochemical analysis within the first 24 h of neural differentiation.
Time-lapse imaging revealed that, after neural induction, the hPDLSCs underwent a rapid change in shape and size, first adopting highly irregular forms and then gradually contracting into round cells. Subsequently, we observed the growth of new neurites from the cell body of round cells (Fig. 1, black arrows). The round cells were also noted to gradually adopt a complex morphology, acquiring dendrite-like (Fig. 2a, white arrows) and axon-like identities (Fig. 2a, red arrows), giving rise to a variety of neuron-like morphologies (Figs. 1, 2).
In addition, cytoskeletal protein β-III tubulin and F-actin staining showed that hPDLSC-derived neuron-like cells developed distinct dendrite-like and axon-like domains (Fig. 2b). Time-lapse imaging revealed that the hPDLSC-derived neuron-like cells connected to one another through different types of interactions (Fig. 2c, red arrows), including axodendritic-like contacts (Fig. 2c, yellow arrows).
hPDLSCs did not differentiate at the same time, so the cell culture simultaneously contained hPDLSCs at different stages of neurogenesis and neuronal polarisation. The results demonstrate that hPDLSCs can directly differentiate into neuron-like cells without passing through a mitotic stage, thus definitively confirming that the rapid acquisition of a neuron-like morphology during neural differentiation is by means of a differentiation trait rather than merely an artefact.
Time-lapse microscopy of hBMSCs and hDPSCs cultured in neural induction media. Isolation, characterisation and neural differentiation of multipotent stem cells from human bone marrow 11,19,20,26,27 and dental pulp [28][29][30][31] have been described previously. Under proliferation conditions, hBMSCs and hDPSCs displayed a fibroblast-like morphology with actin microfilaments and β-III tubulin microtubules oriented parallel to the cell's longitudinal axis (Fig. S1). During interphase, undifferentiated hBMSCs and hDPSCs had an ellipsoidal nucleus, often located in the centre of the cell (Fig. S1). These results are consistent with previous studies that found hBMSCs and dental-derived stem cells exhibited spontaneous expression of neural marker β-III tubulin even without neural induction 32 .
To determine whether the same rapid morphological changes could also be induced in other human adultderived stem cells, we tested hBMSCs and hDPSCs under the same neural induction conditions. As with hPDLSCs, time-lapse imaging revealed that after neural induction hBMSCs and hDPSCs underwent a dramatic change in shape and size, first adopting highly irregular forms, before gradually contracting into round cells. Subsequently, we noted the growth of new neurites from the cell body of hBMSC-derived (Fig. 3a, black arrows) and hDPSC-derived round cells (Fig. 3b, black arrows).
We also observed that hBMSC-derived (Fig. 4a) and hDPSC-derived neuron-like cells (Fig. 4b) gradually adopted a complex morphology, acquiring dendrite-like (Fig. 4, white arrows) and axon-like identities (Fig. 4, red arrows), and giving rise to a variety of neuron-like morphologies (Figs. 3,4,5). Cytoskeletal protein β-III tubulin and F-actin staining showed that hBMSC-derived (Fig. 5a) and hDPSC-derived neuron-like cells (Fig. 5b) developed distinct dendrite-like and axon-like domains. Morphological analysis also revealed that hBMSCderived neuron-like cells were interconnected by different types of interactions, including dendrodendritic-like contacts (Fig. 5c, red arrows) and axoaxonic-like contacts (Fig. 5c,  www.nature.com/scientificreports/ The results indicate that there was no cell proliferation during neurogenesis from hBMSCs and hDPSCs. The undifferentiated spindle-shaped cells were reset and started their neuronal development as round spheres. We observed that the hBMSCs and hDPSCs did not differentiate at the same time and therefore the cell culture simultaneously contained hBMSCs and hDPSCs at different stages of neurogenesis and neuronal polarisation.
These findings indicate that hPDLSCs, hBMSCs and hDPSCs have similar morphological neural development sequences in vitro. The differentiation of hPDLSCs, hBMSCs and hDPSCs towards a neural lineage occurs through a dedifferentiation step before the cells differentiate to neural phenotypes.
Nuclear remodelling. In a recent study 25 , we showed that nuclear remodelling occurred during in vitro neurogenesis from hPDLSCs. We discovered that many hPDLSCs exhibit unusual nuclear structures and extranuclear bodies in the cellular cytoplasm when hPDLSCs approach a spherical morphology (neurogenesis). In addition, no unusual nuclear structures were observed as hPDLSC-derived neuron-like cells gradually acquired a more mature neuron-like morphology (neuronal polarisation).
While we acknowledge that the definitive nuclear remodelling sequence, when hPDLSCs become nearspherical, can only be determined by detecting fluorescent labelled cell nuclei in time-lapse microscopy, our accumulated data suggest how these steps may occur 25 . Chromatin-containing bodies arise from the main nuclei and start to move towards specific positions within the cell, temporarily forming lobed nuclei. Subsequently, these lobed nuclei connect together to form nucleoplasmic bridges and finally the internuclear bridges facilitate the merger into a single nucleus with an eccentric position within the cell.
To determine whether nuclear remodelling also occurs when hBMSCs and hDPSCs take on a near-spherical shape, we differentiated hPDLSCs, hBMSCs and hDPSCs in the same neural induction medium (Figs. 6,7,8).
Within 2 h of exposure to the neural induction medium we observed DNA-containing structures arise from the main nuclei of hPDLSCs (Fig. 6a), hBMSCs (Fig. 7a) and hDPSCs (Fig. 8a). We also observed DNA-containing structures moving away from the main nuclei of and towards specific positions within hPDLSCs (Fig. 6b, c, respectively), hBMSCs (Fig. 7b, c) and hDPSCs (Fig. 8b, c), where they temporarily formed lobed nuclei.
The results indicate that nuclear remodelling also occurred during in vitro neurogenesis from hBMSCs and hDPSCs. hBMSCs and hDPSCs also exhibited unusual nuclear structures and extranuclear bodies in the cellular cytoplasm when they approach a near-spherical shape. These DNA-containing structures displayed a spherical or ovoid shape (Fig. S2a) and some of them appeared to be connected to the main body of the nucleus by thin strands of nuclear material (Fig. S2b, yellow arrows). The hPDLSCs, hBMSCs and hDPSCs had very similar nuclear morphologies during in vitro neurogenesis. These results confirm our previous data 25,33 and further support the idea that hBMSCs, hPDLSCs and hDPSCs follow similar morphological neural development sequences in vitro.

Discussion
In this study, we have shown that hPDLSCs, hBMSCs and hDPSCs can directly differentiate into neuron-like cells without passing through a mitotic stage. In fact, the spindle-shaped cells (differentiated cell stage) were reset and started their neuronal development as round spheres (less differentiated cell stage). It would, therefore, be interesting to examine whether cell rounding is a common feature of cellular plasticity.
Our results indicatate that the differentiation of hPDLSCs, hBMSCs and hDPSCs towards a neural lineage occurs through a dedifferentiation step (neurogenesis) before the cells differentiate to neural phenotypes (neuronal polarisation).
In the present work, we have demonstrated that hPDLSC-derived, hBMSC-derived and hDPSC-derived neuron-like cells produce neurites that gradually adopt a complex morphology, acquiring dendrite-like and www.nature.com/scientificreports/ axon-like identities. Therefore, our results definitively establish that the rapid acquisition of a neuron-like morphology during neural differentiation occurs though a differentiation trait and is not just an artefact. Although further research is required to confirm the successful differentiation of hPDLSCs, hBMSCs and hDPSCs into a neural lineage, and finally facilitate the production of autologous cells for cell replacement in a diseased central nervous system, our results provide additional evidence that cells derived from human adult tissues can be converted into neuronal cells without genetic manipulation 34 .
In a previous publication 25 , we reported that nuclear remodelling occurred during in vitro neurogenesis from hPDLSCs. We observed that many hPDLSCs had an unusual nuclear structure and chromatin-containing bodies in the cellular cytoplasm when they took on a near-spherical shape. In the present study, we have shown that nuclear remodelling also occurred during in vitro neurogenesis from hBMSCs and hDPSCs. We observed that hBMSCs and hDPSCs exhibited unusual nuclear structures and chromatin-containing bodies in the cellular cytoplasm when they became near-spherical in shape. It is important to mention that the hPDLSCs, hBMSCs www.nature.com/scientificreports/ and hDPSCs had a very similar nuclear morphology during in vitro neurogenesis. These results confirm our previous findings 25,33 and further support that hPDLSCs, hBMSCs and hDPSCs follow similar morphological neural development sequences in vitro. These unusual nuclear structures and extranuclear bodies were described independently in mammals (including in most blood and immune cells 35 ), vertebrates, invertebrates, plants and protozoa [35][36][37] . However, the exact role of these extranuclear bodies and their possible correlation with unusual nuclear structures is not yet known 38 . Our results suggest these unusual nuclear structures and extranuclear bodies may be associated with nuclear movement within the cell. Although future analysis involving live-cell nucleus fluorescence labelling and timelapse microscopy is necessary to determine whether chromatin-containing bodies move within the cell, and if there is any correlation between the formation of these extranuclear bodies and the unusual nuclear structures, www.nature.com/scientificreports/ a recent study has demonstrated that chromatin-containing bodies arise from the main nuclei, move within the cell and are ultimately loaded in exosomes 39 . Importantly, the nuclear morphology of hPDLSCs, hBMSCs and hDPSCs observed during the dedifferentiation step (neurogenesis) bears great similarities to the nuclear morphology of neural stem cells (NSCs) located in the ventricular-subventricular zone of the anterolateral ventricle wall in the human foetal brain 40 and adult mouse brain [41][42][43] ; their nuclear morphology is also very similar to that of many cultured hippocampal neurons 44 . Although it has been suggested that these unusual nuclear structures are associated with quiescence in adult NSCs 43 , our results suggest that they may be associated with nuclear movement within the cell during the initial phases of neurogenesis, but without any relation to cell division.
It has generally been believed that adult neurogenesis occurs progressively through sequential phases of proliferation and neuronal differentiation of adult NSCs 45 . However, the current models of adult NSC maintenance www.nature.com/scientificreports/ and differentiation are still contentious because the approaches used to probe stem cell division and differentiation are inherently limited 46 .
It is important to note that very few reports describing newborn neurons in the adult brain have found mitotic chromosomes [47][48][49][50][51][52][53][54][55] , which would confirm that adult neurogenesis occurs progressively through sequential phases of proliferation. What is more, several studies indicate that some observations interpreted as cell division could be false positives [56][57][58][59][60] . The findings mentioned above and the similarities between the nuclear morphology of hPDLSCs, hBMSCs and hDPSCs during in vitro neurogenesis compared to that of adult NSCs located in neurogenic niches and cultured hippocampal neurons suggest that adult NSCs may also convert into neurons without involving cell division. Future research is necessary to determine the likelihood of this conjecture.
As noted above, there is increasing evidence to suggest that adult cells can assume new fates without asymmetric cell division through dedifferentiation and transdifferentiation processes. These findings call into question the rules of irreversibility and germ-layer restriction as well as the classical definition of a stem cell 4-10,61 . www.nature.com/scientificreports/ Our results demonstrate that hPDLSCs, hBMSCs and hDPSCs retain plasticity. Therefore, these cells could also help increase our understanding of the mechanisms underlying changes in cell fate and shed light on whether differentiation, dedifferentiation and transdifferentiation occur in different contexts or are different terms for the same cellular process.

Methods
Ethical conduct of research. Isolation and culture of hPDLSCs and hDPSCs. To obtain hPDLSCs, human premolars were extracted and collected from three different healthy adult donors undergoing orthodontic therapy in Murcia dental hospital (Spain). Human periodontal ligament (hPDL) was scraped from the middle third region of the root surface.
To obtain hPDSCs, human third molars were extracted and collected from three different healthy adult donors undergoing orthodontic therapy in Murcia dental hospital (Spain). Human dental pulp (hDP) was harvested from the teeth. Dissociated cell cultures of hPDL and hDP tissues were prepared as previously described 25,62 . After washing the extracted hPDL and hDP with Ca and Mg-free Hank's balance salt solution (HBSS), hPDL and hDP was digested with type I collagenase (3 mg/ml) and dispase II (4 mg/ml) in alpha modification minimum essential medium eagle (α-MEM) for 1 h at 37 °C. The reaction was stopped by the addition of α-MEM. hPDL and hDP derived from different subjects were pooled together to obtain single cell suspensions by passing the cells through a 70 μm strainer (BD Falcon). Cells were centrifuged, and the pellet was resuspended in serum-containing media (designated as the basal media), composed of DMEM low glucose medium supplemented with fetal bovine serum (10%), non-essential amino acid solution, penicillin-streptomycin (100 units/ml) and l-glutamine (2 mM). The cell suspension was plated into six-well multiwell plates (BD Falcon) and incubated at 37 °C in 5% CO2. All studies were performed using hPDLSCs and hDPSCs expanded within culture passages 3-4. Isolation and culture of hBMSCs. Bone marrow aspirates were obtained by percutaneous direct aspiration from the iliac crest of 5 healthy volunteers at University Hospital Virgen de la Arrixaca (Murcia, Spain). Bone marrow was collected with 20 U/ml sodium heparin, followed by a Ficoll density gradient-based separation by centrifugation at 540 g for 20 min. After, mononuclear cell fraction was collected, washed twice with Ca 2+ /Mg 2+ -free phosphate buffered saline (PBS) and seeded into 175-cm2 culture flasks (Nunc, Thermo Fisher Scientific) at a cell density 1.5 × 10 5 cells/cm 2 in DMEM low glucose medium supplemented with fetal bovine serum (10%), GlutaMAX (1%) and penicillin/streptomycin (1%). After 3 days of culture at 37ºC and 5% CO 2 , non-attached cells were removed and fresh complete medium was added. Culture media were renewed every 2 days, and the isolated hBMSCs were passaged when cultures were 70-80% confluent. All studies were performed using hBMSCs expanded within culture passages 3-4.
Time-lapse microscopy of hPDLSCs, hPDSCs and hBMSCs cultured in neural induction media. We used μ-Dish 35 mm, high Grid-500 (Ibidi) for live cell imaging. Numeric marks on the bottom of each dish allow users to identify the location of cells. Cells were plated onto collagen IV coated plastic or glass coverslips, and maintained in basal media or neural induction media.
To induce neural differentiation, cells at passage 3-4 were allowed to adhere to the plates overnight. Media was removed the following day and the cells were cultured for 2 days in serum-free media (designated as the neural basal media) consisting in Dulbecco's modified Eagle's medium/F12 (DMEM/F12 Glutamax, Gibco) supplemented with N2-supplement, glucose (0.6%), HEPES (5 mM), human serum albumin (0.5%), heparin (0.0002%), non-essential amino acid solution and penicillin-streptomycin (100 units/ml). On day 3, cells were cultured in neural induction media, consisting in the neural basal media supplemented with retinoic acid (500 nM), dibutyryl cAMP (1 mM) and growth factors like BDNF (10 ng/ml; Peprotech), GDNF (10 ng/ml; Peprotech) and IGF-1 (10 ng/ml; R&D systems). We perform time-lapse phase-contrast microscopy within the first 24 h after neural induction media was added directly to the cells.

Immunocytochemistry.
A standard immunocytochemical protocol was used as previously described 25,62 .
Cells were plated onto collagen IV coated plastic or glass coverslips, and maintained in basal media or neural induction media. Cells were rinsed with PBS and fixed in freshly prepared paraformaldehyde (4%). Fixed cells were blocked for 2 h in PBS containing normal horse serum (10%) and Triton X-100 (0.25%) and incubated overnight at 4 °C with antibodies against β-III-tubulin (TUJ1; 1:500, Covance) in PBS containing normal horse serum (1%) and Triton X-100 (0.25%). On the next day, cells were rinsed and incubated with the secondary antibody: Alexa Fluor® 594 (anti-mouse; 1:500, Molecular Probes). Cell nuclei were counterstained with DAPI (0.2 mg/ml in PBS, Molecular Probes). Alexa Fluor 488® phalloidin (Molecular Probes) was used to selectively stains F-actin. Data are representative of ten independent experiments per condition.
Images and data analyses. Analyses and photography of visible and fluorescent stained samples were carried out as previously described 25 . We used an inverted Leica DM IRB microscope equipped with a digital camera Leica DFC350FX (Nussloch) and confocal laser scanning microscope Leica TCS-SP8. Digitized images www.nature.com/scientificreports/ were analyzed using LASX Leica confocal software. We used Photoshop software to improve the visibility of fluorescence images without altering the underlying data.