Non-proliferative neurogenesis in human periodontal ligament stem cells

Understanding the sequence of events from undifferentiated stem cells to neuron is not only important for the basic knowledge of stem cell biology, but also for therapeutic applications. In this study we examined the sequence of biological events during neural differentiation of human periodontal ligament stem cells (hPDLSCs). Here, we show that hPDLSCs-derived neural-like cells display a sequence of morphologic development highly similar to those reported before in primary neuronal cultures derived from rodent brains. We observed that cell proliferation is not present through neurogenesis from hPDLSCs. Futhermore, we may have discovered micronuclei movement and transient cell nuclei lobulation coincident to in vitro neurogenesis. Morphological analysis also reveals that neurogenic niches in the adult mouse brain contain cells with nuclear shapes highly similar to those observed during in vitro neurogenesis from hPDLSCs. Our results provide additional evidence that it is possible to differentiate hPDLSCs to neuron-like cells and suggest the possibility that the sequence of events from stem cell to neuron does not necessarily requires cell division from stem cell.

Neurons are highly polarized cells exhibiting functionally and structurally distinct processes called axons and dendrites 1,2 . Understanding the mechanisms that underlie the temporal and spatial control of neuronal polarization is not only important for neurobiology 3 , but also for nerve regeneration in injured nervous systems 4 .
The process of neuronal polarization has been studied for decades using dissociated rodent postnatal cerebellar granule neurons and embryonic hippocampal pyramidal neurons in culture 1,2 . The establishment of polarity in cultured rodent neurons are divided into different stages 5,6 .
Upon isolation, dissociated pyramidal neurons retract their processes, so their neuronal development in vitro start as rounded spheres that spread lamellipodia (stage 1). These spheres appear symmetrical, extending and retracting several immature neurites of a similar length (stage 2). Elongation of a single process, that which presumably becomes the axon, breaks this symmetry (stage 3). The next step involves the remaining short neurites morphologically developing into dendrites (stage 4). The last step (stage 5) in neuronal polarization from dissociated pyramidal neurons in culture is the functional polarization of axon and dendrites, including dendritic spine formation and axon branch formation 5 . Dissociated granule neurons also present a lamellipodia after attaching to the substratum (stage 1). These spheres extend a unipolar process at a single site on the plasma membrane (stage 2) followed by extension of a second process from the opposite side of the cell body, resulting in a bipolar morphology (stage 3). One of the two axon elongates futher and start branching (stage 4), and shorter dendritic processes develop around the cell body (stage 5) 6 .
Although much progress has been made in the knowledge of how rodent neurons establish their polarity [1][2][3]5,6 , less is known about the process of neuronal polarization in human cells 7,8 .
The major barrier to studying human neurons is the inaccessibility of living tissue, therefore an enormous effort has been made in this study to derive neurons from human stem cells [9][10][11] .
Neural crest stem cells (NCSCs) are a migratory cell population that generate numerous cell lineages during development, including neurons and glia 12,13 . NCSCs can be isolated not only from embryonic neural crest, but also from fetal and adult neural crest-derived tissues 14 .
The periodontal ligament (PDL) is a connective tissue surrounding the tooth root that contains a source of human NCSCs which can be accessed with minimal technical requirements and little inconvenience to the donor 15 . Isolation and characterization of multipotent stem cells from the human PDL have been previously described 16,17 .
In previous publication 18 , we showed that human adult periodontal ligament (hPDL) tissue and hPDL-derived cells express marker genes of stem cells and neural crest cells. In vitro, hPDL-derived cells differentiate into neural-like cells based on cellular morphology and neural marker expression. In vivo, hPDL-derived cells survive, migrate and expressed neural markers after being grafted to the adult mouse brain. Futhermore, some hPDL-derived cells graft into neural stem cell (NSC) niches such as the ventricular-subventricular zone (V-SVZ) of the anterolateral ventricle wall and the subgranular zone (SGZ) of the dentate gyrus in the hippocampus. It is important to mention that the hPDLSCs located in the NSC niches show neural stem morphology.
Here, we show that hPDLSCs-derived neural-like cells display a sequence of morphologic development highly similar to those reported before in primary neuronal cultures derived from rodent brains during neurogenesis, providing additional evidence that it is possible to differentiate hPDLSCs to neuron-like cells.
We observed that cell proliferation is not present through neurogenesis from hPDLSCs. In fact, the cell shape of hPDLSCs is reset and start their neuronal development as round spheres. Futhermore we may have discovered a transient cell nuclei lobulation coincident to in vitro neurogenesis, without being related to cell proliferation. We observed that small DNA containing structures may move within the cell to specific directions and temporarily form lobed nuclei.
Morphological analysis also reveals that the V-SVZ of the anterolateral ventricle wall and the SGZ of the hippocampal dentate gyrus in the adult mouse brain contains cells with nuclear shapes highly similar to those observed during in vitro neurogenesis from hPDLSCs. We suggest the possibility that neuronal differentiation from NSCs may also occur during in vivo adult mammalian neurogenesis without being related to cell proliferation.

Results
hPDLSCs cultured in basal media. Under proliferation conditions, hPDLSCs displayed a fibroblast-like morphology with low-density microvilli on the cell surface (Fig. 1a) and actin microfilaments and β-III tubulin microtubules oriented parallel to the longitudinal axis of the cell (Fig. 1b). The cytoskeletal protein class III beta-tubulin isotype is widely regarded as a neuronal marker in developmental neurobiology and stem cell research 25 . Dental and oral-derived stem cells displayed spontaneous expression of neural marker β-III tubulin, even without having been subjected to neural induction 26 . Western blot analysis verified the expression of β-III tubulin in hPDLSCs (Fig. 1c). During interphase, undifferentiated hPDLSCs displayed a flattened, ellipsoidal nucleus, often located in the center of the cell and with a nuclear volume around 925′356 ± 52′6184 μm 3 (Fig. 1d).
During mitosis, β-III tubulin is present in the mitotic spindle and it is detectable in all phases of mitosis (Fig. 1e). The cytoskeletal protein class III beta-tubulin isotype is a component of the mitotic spindle in multiple cell types 27 . At the end of mitosis, division of the cytoplasm by cytokinesis is observed (Fig. 1f). Mitosis and cytokinesis last no more than 2 hours (Fig. 1g).
hPDLSCs cultured in neural induction media. After 14 days of neural differentiation conditions, the hPDLSCs displayed different morphologies, including round cells with small phase-bright cell bodies and short processes; highly irregulary-shaped cells; and, also, unipolar, bipolar and multipolar-shaped cells with small phase-bright cell bodies and multiple branched processes (Fig. 2a). In addition, cells of different size were also observed ( Fig. 2b). Futhermore, microscopic analysis revealed that some hPDLSCs have different nuclear shapes, including lobed nuclei connected by an internuclear bridge (Fig. 2c).
We examined the sequence of developmental events during the first day in culture. We observed that different-shaped cells are stages of development of hPDLSCs during neural differentiation (Fig. 2d). The results may indicate that the cell culture simultaneously contains hPDLSCs at different stages of neurogenesis and neuronal polatization. We acknowledge that the definitive sequence of in vitro neurogenesis and neuronal polarization from hPDLSCs will be provided only by time-lapse microscopy of a single cell, but in our experimental conditions, several pieces of data suggest how these steps may occur.
In vitro neurogenesis from hPDLSCs. After neural induction, hPDLSCs undergo a dramatic change in shape and size, first adopting highly irregular forms and then gradually contracting into round cells with small phase-bright cell bodies (Fig. 3a). Cytoskeletal remodeling is observed during the morphological changes that occurred when the hPDLSCs round up to a near-spherical shape. Actin microfilament not longer surround the nucleus and became cortical. Unlike actin, β-III tubulin seems to accumulate around the nucleus (Fig. 3b). Actin microfilament and β-III tubulin microtubule network are almost lost in the rounded cells (Fig. 3c).
Scanning electron micrographs show that hPDLSCs also experience dramatic changes in cell surface features. Under proliferation conditions, hPDLSCs remain very flat, presenting low-density microvilli on their surface (Fig. 1a), but there is a marked increase in the number of microvilli as the cells round up to near-spherical shape (Fig. 3d). We also observed that the surface of the rounded cells is almost devoid of microvilli (Fig. 3e).
These results are consistent with previous studies reporting that microvilli increase in number during mitosis but become greatly disminished in number after cytokinesis is completed and the cells enter interphase 28,29 . Our results may indicatate that the dramatic change in shape and size observed during neurogenesis from hPDLSCs is completed.
Morphological and immunocytochemical analysis revealed that rounded cells do not represent dividing cells due to the absence of cytokinesis, mitotic chromosomes and mitotic spindle during the described of in vitro www.nature.com/scientificreports www.nature.com/scientificreports/ Neuronal polarization of hPDLSCs-derived neural-like cells. Morphological analysis revealed that hPDLSCs-derived neural-like cells display a sequence of morphologic development highly similar to those reported before in dissociated-cell cultures prepared from rodent brain 1,2,5,6 (Figs. 4-6). hPDLSCs-derived www.nature.com/scientificreports www.nature.com/scientificreports/ neural-like cells also start their development as rounded spheres that initiated neurite outgrowth at a single site on the plasma membrane 6 , first becoming unipolar, stages 1-2 (Fig. 4a). We did not observe the development of lamellipodia around the circumference of the cell body. These unipolar cells, later transformed into cells containing several short neurites, developed around the cell body, stage 3 (Fig. 4b). An analysis of the cytoskeletal organization during spherical stages of hPDLSCs-derived neural-like cells showed that the β-III tubulin microtubules and actin microfilament network is reorganized. Cytoskeletal protein β-III tubulin was densely accumulated under the cellular membrane of the cell body and in cell neurites (Fig. 4a,b) while actin microfilaments were mainly found in cell neurites (Fig. 4c).
We observed that hPDLSCs-derived neural-like cells produce neurites that showed growth cone formations at their tips ( Fig. 4c-e). The central domain of the growth cone contains β-III tubulin microtubules and the   www.nature.com/scientificreports www.nature.com/scientificreports/ peripheral domain is composed of radial F-actin bundles (Fig. 4d), similar to the typical spatial organization described in neurons 30,31 . Scanning electron micrographs also showed that the growth cone of hPDLSCs-derived neural-like cells contained filopodia and vesicles on the cell surface (Fig. 4e). These finding are consistent with a www.nature.com/scientificreports www.nature.com/scientificreports/ previous study reporting that membrane addition and extension in growth cones is mediated by diverse mechanism, including exocytosis of vesicular components 32 . Microtubule-associated proteins Tau and MAP2 were also found in hPDLSCs-derived neural-like cells (Fig. 4f). At later stages of differentiation, the hPDLSCs-derived neural-like cells gradually adopted a complex morphology by forming several processes, stage 4 ( Fig. 4g) that grew and arborized, adquiring dendritic-like and axonal-like identities, giving rise to a variety of neuron-like morphologies (Fig. 4h). The next step, stage 5, in neuronal polarization from rodent neurons in culture is the functional polarization of axon and dendrites, including dendritic spine formation and axon branch formation 1,2,5,6 . Dendritic spines are micron-sized dendrite membrane protrusions 33 . Depending on the relative sizes of the spine head and neck, they can be subdivided into different categories, including filopodium, mushroom, thin, stubby, and branched spines 34 . Dendritic spines are actin-rich compartments that protrude from the microtubule-rich dendritic shafts of principal neurons 35 . Based on morphology, complexity, and function, axon branching is grouped into different categories, including arborization, bifurcation, and collateral formation 36 .
Our morphological analysis revealed that hPDLSCs-derived neural-like cells developed well-differentiated axonal-like and dendritic-like domains. These types of processes differ from each other in morphology (Figs. 4i and 5). Cytoskeletal protein β-III tubulin and F-actin staining showed that the hPDLSCs-derived neural-like cells comprised multiple branched dendrite-like processes with dendritic spines-like structures (Fig. 4i). Scanning electron micrographs showed that the hPDLSCs-derived neural-like cells also contained multiple branched dendrite-like processes with variously shaped spine-like protusions, highly similar to filopodium, mushroom, thin, stubby, and branched dendritic spines shapes (Fig. 5a). Futhermore, hPDLSCs-derived neural-like cells also displayed different types of axonal branch-like structures, including bifurcation (Fig. 5b), arborization (Fig. 5c), and collateral formation (Fig. 5d).
The last step in neuronal polarization from rodent neurons in culture is synapse formation 1,2,5,6 . The most frequent types of synaptic communication include axodendritic, axosomatic, axoaxonic and dendrodendritic synapses. Morphological analysis revealed that the hPDLSCs-derived neural-like cells connected to one another (Fig. 6a) through different types of synapse-like interactions, including dendrodendritic-like, axoaxonic-like and axodendritic-like synapses (Fig. 6b). Synapse-associated proteins Cx43, Synaptophysin and Synapsin1 were found accumulated in the cell surface of neurites (Fig. 6c).
Nuclear remodeling. Nuclear morphology was examined in hPDLSCs under proliferation and neural differentiation conditions. The dynamic localization of the nucleoli was analyzed by immunostaining for fibrillarin, the main component of the active transcription centers 37 and the dynamic localization of the nuclear lamina was analyzed by immunostaining for lamin A/C, a nuclear lamina component 38 .
First, we analyzed the nuclear morphology in proliferative hPDLSCs. As noted above, during interphase, hPDLSCs displayed a flattened, ellipsoidal nucleus, often located in the center of the cell, and with a nuclear volume around 925.356 ± 52.6184 μm 3 (Fig. 1d). The nuclei of hPDLSCs contained two or more nucleoli and the inside surface of the nuclear envelope is lined with the nuclear lamina (Fig. S1a). Previous studies have shown that the nuclear lamina and nucleolus are reversibly disassembled during mitosis 39,40 . Microscopic analysis of hPDLSCs revealed that the dynamic localization of fibrillarin and lamin A/C proteins during mitosis are similar to those observed in previous studies (Fig. S1b). Futhermore, β-III tubulin is present in the mitotic spindle and it is detectable in all phases of mitosis (Fig. S1b). In addition, mitotic chromosomes and cytokinesis were also observed (Figs. 1e-g and S1b).
Morphological analysis also revealed that nuclear remodeling occurred during in vitro neurogenesis from hPDLSCs (Fig. 7). We acknowledge that the definitive sequence of nuclear remodeling when hPDLSCs round up to near-spherical shape will only be provided by time-lapse microscopy, but our accumulated data suggests how these steps may occur.
Small DNA containing structures arise from the main nuclei (Fig. 7a) and start to move (Fig. 7b) towards specific positions within the cell (Fig. 7c) and temporarily form lobed nuclei (Fig. 7d). Later, these lobed nuclei connected to one another through small DNA containing structures (Fig. 7e) forming nucleoplasmic bridges (Fig. 7f). Finally, lobed nuclei connected by an internuclear bridge (Fig. 7g) join in a single nucleus with an eccentric position within hPDLSCs-derived neural-like cells (Fig. 7h).
These small DNA containing structures displayed a spherical or ovoid shape (Fig. S2a), and it seems that some of them are connected to the main body of the nucleus by thin strands of nuclear material (Fig. S2b). Fibrillarin and lamin A/C proteins were detected in these small DNA containing structures (Fig. S2c).
We also observed that the nuclear lamina (Fig. S3a) and nucleolus (Fig. S3b) are not disassembled during in vitro neurogenesis from hPDLSCs, contrary to what happens during mitosis 39,40 . Moreover, mitotic chromosomes, mitotic spindle and cytokinesis were not observed during the described of in vitro neurogenesis processes or neuronal polarization from hPDLSCs (Figs. 2-7 and S2-S5).
In addition, ultrastructural morphological characteristics of hPDLSCs were examined under neural differentiations conditions. Transmission electron microscopy (TEM) is considered the gold standard to confirm apoptosis 41 . Apoptotic cell contains certain ultrastructural morphological characteristics, including electron-dense nucleus, disorganized cytoplasmic organelles, large clear vacuoles, nuclear fragmentation and apoptotic bodies. TEM analysis revealed that hPDLSCs under neural differentiations conditions do not meet the criteria described above (Fig. S4). Therefore, hPDLSCs with lobed nuclei do not represent apoptotic cells.
During neuronal polarization, no lobed nuclei were observed as hPDLSCs-derived neural-like cells gradually acquired a more mature neuronal-like morphology (Fig. S5a). We also found that as the cells round up to a near-spherical shape the nuclear volume of the hPDLSCs decreases to an approximate volume of 279.589 ± 38.8905 μm 3 (Fig. S5b).
Interestingly, the morphological analysis revealed that the adult rodent V-SVZ of the anterolateral ventricle wall (Fig. 8a) and the SGZ of the hippocampal dentate gyrus (Fig. 8b), where adult neurogenesis has been clearly demonstrated, contained cells with nuclear shapes highly similar to those observed during in vitro neurogenesis from hPDLSCs.
It is important to mention that nuclear morphology of hPDLSCs observed during in vitro neurogenesis from hPDLSCs (Fig. 7) are highly similar to nuclear morphology of V-SVZ B cells in the human fetal brain 42 and V-SVZ B cells in adult mice brain [43][44][45] . Although it has been suggested that lobed nuclei connected by an internuclear bridge are associated with quiescence in aNSCs 45 , we observed that this kind of nuclei may be associated to nuclear movement within the cell during initial phases of neurogenesis, without being related to cell proliferation.

Discussion
In this study we show that hPDLSCs-derived neural-like cells display stages of development highly similar to those reported before in primary neuronal cultures derived from rodent brains 1,2,5,6 . The hPDLSCs-derived neural-like cells gradually adopted a complex morphology by forming several processes, that grew and arborized, adquiring dendritic-like and axonal-like identities, giving rise to a variety of neuron-like morphologies.
During neuronal polarization in vivo, the stages of development observed in mouse immature neurons are not necessarily the same as those observed in vitro, since they depend on the stage of development and the brain region 1,2 . In a previous publication, we showed that hPDLSCs-derived neural-like cells integrate and differentiate after implantation into the adult mammalian brain 18 . It is important to mention that the hPDLSCs-derived neural-like cells located in the NSC niches show neural stem morphology, which suggests hPDLSCs-derived neural-like cells may achieve a correct polarization in vivo. www.nature.com/scientificreports www.nature.com/scientificreports/ Therefore, our results provide additional evidence that it is possible to differentiate hPDLSCs to neuron-like cells, as suggested by their neural-crest origin, stem cell characteristics and neural differentiation potential both in vitro and in vivo [13][14][15][16][17][18][19][20][21][22][23][24] . However, it is important to mention that future research is required to resolve the potential limitations in their medical application 10,14,46 , and also to optimize the diversity of in vitro neural induction protocols that have been designed for dental stem cells 22 . In addition, future analysis will be necessary to study neural linages derived from hPDLSCs including neuron-like, astrocyte-like, and oligodendrocyte-like cells.
In this study, we also show that cell proliferation is not present through neurogenesis from hPDLSCs. The undifferentiated polygonal and fusiform cell shapes are reset and start their neuronal development as round spheres. As noted above, morphological characteristics of the hPDLSCs, including cytoskeleton and nuclear morphology were examined in cells under proliferation and neural differentiation conditions. During mitosis, the nucleolus and the nuclear lamina are reversibly disassembled and β-III tubulin protein is present in the mitotic spindle and it is detectable in all phases of mitosis. Moreover, mitotic chromosomes and cytokinesis were observed. Mitosis and cytokinesis last no more than 2 hours. During neurogenesis, the nucleolus and the nuclear lamina are not disassembled, contrary to what happens during mitosis 39,40 . Moreover, mitotic chromosomes, mitotic spindle and cytokinesis were not observed. In addition, the duration of the morphological changes that occurred when the hPDLSCs round up to a near-spherical shape lasts much longer than mitosis and cytokinesis. Collectively, these results revealed that round cells do not represent dividing cells.
In this study, we may have discovered a transient cell nuclei lobulation coincident to in vitro neurogenesis. Ultrastructural analysis with transmission electron microscopy revealed that hPDLSCs with lobed nuclei do not represent apoptotic cells due to the absence of features of cells undergoing apoptotis 41 . Interestingly, the morphological analysis revealed that the adult rodent V-SVZ of the anterolateral ventricle wall and the SGZ of the hippocampal dentate gyrus contained cells with nuclear shapes highly similar to those observed during in vitro neurogenesis from hPDLSCs. It is important to mention that nuclear morphology of hPDLSCs observed during in vitro neurogenesis from hPDLSCs are highly similar to nuclear morphology of V-SVZ B cells in the human fetal brain 42 and V-SVZ B cells in adult mice brain [43][44][45] .
The presence of neural stem cells in the adult mammalian brain (aNSCs) have been described in two neurogenic niches, the V-SVZ of the anterolateral ventricle wall and the subgranular zone SGZ of the hippocampal dentate gyrus 43,47-53 . The study of the cell composition of neurogenic niches and the use of methods for detecting proliferating cells, suggest that neurogenesis occurs progressively through sequential phases of proliferation and the neuronal differentiation of aNSCs.
In the V-SVZ, putative aNSCs (type B cells) divide to give rise to intermediate progenitor cells (type C cells), which divide a few times before becoming neuroblasts (type A cells). The neuroblast then migrate into the olfactory bulb and differentiate into distinct types of neurons 43,48,49 . In the SGZ, putative aNSCs divide to give rise to intermediate progenitor cells which exhibit limited rounds of proliferation before generating polarized neuroblast [49][50][51][52][53] . Neuroblast, as polarized cells, then migrate, guided by the leading process, along SGZ and differentiate into dentate granule neurons 54,55 .
Previous ultrastructure and immunocytochemistry studies show that the V-SVZ stem cell niche contains cells with different morphologies and irregular nuclei [42][43][44][45]47,48,[56][57][58] . Type-B cells have irregular nuclei that frecuently contain invaginations and irregular contours of the plasma membrane. Type-C cells nuclei contained deep invaginations and these cells are more spherical. Type-A cells have elongated cell body with one or two processes and the nuclei are occasionally invaginated 43 . Importantly, some studies have shown that murine and human V-SVZ have segmented nuclei connected by an internuclear bridge 42,44,45 . Although it has been suggested that lobed nuclei connected by an internuclear bridge are associated with quiescence in aNSCs 45 , we observed that this kind of nuclei may be associated to nuclear movement within the cell during initial phases of neurogenesis, without being related to cell proliferation.
In addition, previous reports also shown irregular shaped nuclei in the adult SGZ [59][60][61][62][63][64][65][66] . Adult SGZ NSCs (type 1 cells) have irregular contours of the plasma membrane, and differences in heterochromatin aggregation has been also observed 53 . Futhermore, adult SGZ NSCs (type 2 cells) had an irregularly shaped nucleus 55,67 . Importantly, one study also found that many cultured hippocampal neurons have irregular nuclei or even consisted of two or more lobes connected by an internuclear bridge 68 .
It has commonly been assumed that adult neurogenesis occurs progressively through sequential phases of proliferation 54,55 . Despite the advantages for the detection of adult neurogenesis using exogenous nucleotide analog administration or endogenous cell cycle markers, in addition to retroviral transduction, cell stage and lineage commitment markers, recent findings indicate that some observations interpreted as cell division could be false-positive signals [69][70][71][72] . The main method used to labed new neurons has been the incorporation of the thymidine analogs into the genome of dividing cells during S-phase of the cell cycle, but nevertheless thymidine analogs such as tritiated thymidine, BrdU, CldU and IdU may also be incorporated during DNA turnover or DNA repair [69][70][71][72] . Infection with retrovirus is another method used to label new neurons however, retroviral vectors not specifically infect dividing cells 71 and also, it has even been observed that postmitotic pyramidal neurons can also be labeled by fused infected microglia 73 . Although the expression of endogenous cell cycle proteins is also used to label new neurons, recent findings indicate that cell cycle proteins expression is not necessarily related to cell division 72 . Proliferating cell nuclear antigen is also invoved in DNA repair 74 . Positivity of the proliferation marker KI-67 in noncycling cells has also been observed 75 .
These findings indicate that there is a lack of a reliable definitive method to label new neurons. In addition, it is important to note that almost none of these labeled new born neurons show mitotic chromosomes or mitotic spindle to really confirm that adult neurogenesis occurs progressively through sequential phases of proliferation 43,[47][48][49][50][51][52][53] . Importantly, the existence of non-proliferative neuronal precursors in several brain areas has also been observed 76,77 . Moreover, the self-renewal and multipotent properties demonstrated by NSC in vitro 78 have not been clearly demonstrated in vivo 54,67,79 .
Collectively, these results suggest the possibility that the sequence of events from aNSCs to neuron may also occur without being related to cell proliferation. It would therefore be interesting to examine whether SVZ and SGZ intermediate progenitor cells represent different stages of neurogenesis without being related to cell proliferation.
It is known for many decades that adult cells can change their indentity throught spontaneous dedifferentiation, differentiation and transdiferentiation in vitro and in vivo [80][81][82] . Future analysis will be necessary to study if hPDLSCs convert directly to neural-like cells or if they go through neural progenitor stage first. Our results suggest the possibility that hPDLSCs could be use to advance knowledge of the cellular plasticity.
Beyond the central nervous system, the presence of lobed nuclei has been reported in most blood and immune cells, but the functional significance of multilobed nuclear structures is not yet known [83][84][85][86] . We observed that the nuclei of hPDLSCs during initial phases of neurogenesis are highly similar to those reported in immune cells. Thus, we suggest the possibility that multilobed nuclear structures may be associated to nuclear movement within the cell. Therefore, it would also be interesting to examine whether these putative madure cells also represent different stages of haematopoietic stem cell differentiation without being related to cell proliferation.
One of the most important discoveries in this work is the observation that small DNA containing structures may move within the cell to specific directions and temporarily form lobed nuclei. These small DNA containing structures displayed a spherical or ovoid shape, and it seems that some of them are connected to the main body of the nucleus by thin strands of nuclear material. It is important to note that some DNA containing structures are highly similar to nuclear envelope-limited chromatin sheets (ELCS) 45,87,88 . Fibrillarin and lamin A/C proteins were detected in these small DNA containing structures.
It is known for many decades that chromatin particles can appear in the cellular cytoplasm and they are referred to as micronuclei, nucleoplasmic bridge and nuclear bud 89 . Although these nuclear anomalies have been associated with chromosomal instability events during mitosis [89][90][91][92] , recent reports showed generation of micronuclei during interphase [93][94][95] . These findings call into question that micronuclei, nucleoplasmic bridge and nuclear bud does necessarily generated during mitosis 96 . Moreover, a high frequency of human mesenquimal stem cells with nuclear bud, micronuclei and nucleoplasmic bridge was detected under normal in vitro culture 97 . Therefore, the mechanisms that lead to extra-nuclear bodies formation and their biological relevance are still far from been understood 96 www.nature.com/scientificreports www.nature.com/scientificreports/ In this study, we show that there can be a relationship in the formation of the nuclear bud, micronuclei, nucleoplasmic bridge and nuclear envelope-limited chromatin sheets. Collectively, these results suggest the possibility that the interphase cell nucleus may can reversibly disassembled into functional subunits that may moved independently within the cell, if necessary.
Multinuclear cells are commonly found in various human organs including heart, liver, salivary glands, muscle and endometrium, but their functional advantage remains unknown 100,101 . In addition, alterations in nuclear morphologies are closely associated with a wide range of human diseases, including cancer 102 . hPDLSCs could facilitate an understanding of the mechanisms regulating nuclear morphology in response to cell shape changes and their functional relevance 103,104 . Cell culture. Dissociated cell cultures of hPDL tissue were prepared as previously described 18 . Human premolars were extracted and collected from three different healthy adult donors undergoing orthodontic therapy in Murcia dental hospital (Spain). hPDL was scraped from the middle third region of the root surface. After washing the extracted PDL with Ca and Mg-free Hank's balance salt solution (HBSS; Gibco), hPDL was digested with 3 mg/ml type I collagenase (Worthington Biochemical Corporation) and 4 mg/ml dispase II (Gibco) in alpha modification minimum essential medium eagle (α-MEM) (α-MEM; Sigma-Aldrich) for 1 h at 37 °C. The reaction was stopped by the addition of α-MEM. PDL 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 in serum-containing media (designated as the basal media), composed of α-MEM supplemented with 15% calf serum (Sigma), 100 units/ml penicillin-streptomycin (Sigma) and 2 mM l-glutamine (Sigma). The cell suspension was plated into six-well multiwell plates (BD Falcon) and incubated at 37 °C in 5% CO 2 . The cells at passage 3-4 were then used for the following experiments. 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.

Methods
To induce neural differentiation, cells were cultured in serum-free media (designated as the neural induction media), consisting in Dulbecco's modified Eagle's medium/F12 (DMEM/F12, Gibco) supplemented with bFGF (20 ng/ml, R&D Systems), EGF (20 ng/ml, R&D Systems), glucose (0.8 mg/ml, Sigma), N2-supplement (Gibco), 2 mM l-glutamine (Sigma), and 100 units/ml penicillin-streptomycin (Sigma). Cells at passage 3-4 were allowed to adhere to the plates overnight. Media was removed the following day and neural induction media was added directly to the cells. Neural induction media were changed every 3 days until the end of the experiment (2 weeks).
In previous publication 18 , we showed that several stem cell and neural crest cell markers are expressed in human adult periodontal ligament (hPDL) tissue and hPDL-derived cells.

Immunocytochemistry.
A standard immunocytochemical protocol was used as previously described 18 .

Immunohistochemistry.
A standard immunohistochemistry protocol was used as previously described 18 .
Experiments were carried out according to the guidelines of the European Community (Directive 86/609/ ECC) and in accordance with the Society for Neuroscience recommendations. Animals used in this study were 12-week-old immune-suppressed mouse (Hsd:Athymic Nude-Foxn1 nu/nu; Harlan Laboratories Models, S.L), housed in a temperature and humidity controlled room, under a 12 h light/dark cycles, with ad libitum access to food and water. The animals were anesthetized and intracardially perfused with freshly prepared, buffered 4% PFA (in 0.1 M PB, pH 7.4). Brains were removed, post-fixed for 12 hr in the same fixative at 4 °C and dehydrated in 30% sucrose solution at 4 °C until sunk. 30 µm thick coronal sections were collected using a freezing microtome. Serial sections were used for DAPI staining. Free-floating sections were incubated and mounted onto Superfrost Plus glass slides (Thermo Scientific). The slides were dried O/N and coverslipped with mowiol-NPG (Calbiochem).
Images and data analyses. Analyses and photography of visible and fluorescent stained samples were carried out in an inverted Leica DM IRB microscope equipped with a digital camera Leica DFC350FX (Nussloch) or in confocal laser scanning microscope Leica TCS-SP8. Digitized images were analyzed using LASX Leica confocal software. Z-stacks of confocal fluorescent images were also analyzed to calculate the nuclear volume by using ImageJ software.
We used Photoshop software to improve the visibility of fluorescence images without altering the underlying data. The fluorescent photographs presented in this study are original fluorescence images and inverted images (negatives) that are produced from the original fluorescence images.
Scanning electron microscopy. Cells were plated onto collagen IV (Sigma) coated glass coverslips and maintained in basal media or neural induction media. Cells were treated with fixative for 20 minutes. Coverslips were postfixed in 1% osmium tetroxide for 1 hour and dehydrated in graded ethanol washes. The coverslips were allowed to dry at a conventional critical point and were then coated with gold-palladium sputter coated. Coverslips were view on a Jeol 6100 scanning electron microscope.
Transmission electron microscopy. hPDL-derived cells were harvested using trypsin/EDTA (Gibco) and were treated with fixative for 60 minutes. Cells were postfixed in osmium tetroxide solution, dehydrated embedded in resin. Ultrathin sections (70-90 nm) were cut, stained with lead citrate, and examined under Jeol 1011 and Philips Tecnai 12 transmission electron microscopes.