Differences in DPSC proliferation and differentiation capacities in vitro
and in vivo
as described in previous reports indicate heterogeneity within DPSC populations 
. In turn, such heterogeneity may be attributed to distinct developmental origins. The hypothesis that DPSCs are of neural crest origin has been suggested and indirectly related by previous reports 
. Accordingly, human dental pulp contains self-renewing DPSCs capable of giving rise to mesenchymal-lineages and non-mesenchymal cells such as neuron-like, smooth muscle-like, and melanocyte-like cells, indicating that these cells might be derived from neural crest 
. Herein we explored several criteria to determine if stem cells in dental pulp originate from neural crest. First we investigated the proliferative capacity of murine DPSCs cultured in stem cell selective media, as well as, the expression of stem cell and neural crest genes. Next, we determined in vitro
and in vivo
multi-differentiation of DPSCs into neural crest-lineage. Lastly, we used the Wnt1-Cre/R26R-LacZ
transgenic mouse to conclusively determine that DPSCs generated in our cultures are of neural crest origin.
Unlike human, mouse DPSCs have not been well described possibly due to the difficult nature of isolating dental pulp from murine teeth 
. A recent study characterized progenitors and stem cells in dental pulps of mouse molars in uneruption and eruption states in vitro
and revealed a small population of mesenchymal multipotent cells in erupted molars but not in the dental pulp of the unerupted molars. Instead, a majority of dental pulp cells of the unerupted molar were osteo-dentogenic progenitor cells 
. In our studies, we determined the existence of stem cells in mouse dental pulp from unerupted molars which exhibit high proliferation and multi-differentiation into mesenchymal and non-mesenchymal lineages. DPSCs sufficiently self-perpetuated more than 14 passages (~3 months) without morphological changes. DPSCs chosen for all differentiation experiments were harvested near passage 7, the time which Nanog
expression was highest (Figure S5
). In this study we used the optimal conditions for long-term cultivation of stem cells without any signs of degeneration or spontaneous differentiation; near physiological oxygen concentration, and a low concentration of serum supplemented with platelet derived growth factor-BB (PDGF-BB), epidermal growth factor (EGF), insulin transferrin selenium (ITS), dexamethasone, and leukemia inhibitory factor (LIF) 
. This combination differs from previously described culture condition used to grow human and mouse DPSCs 
. Therefore, a direct comparison of culture conditions must be undertaken to determine if DPSC described here are equivalent to previously described populations.
Interestingly the expression of embryonic pluripotent genes suggests the existence of primitive stem cells within the DPSC population that share a similar genetic program to embryonic and inducible pluripotent stem cells 
. Q-RT-PCR results showed down-regulation of Klf4
after differentiation of DPSCs, suggesting that both transcription factors play a similar role and function for maintenance of DPSC multi-potentiality and undifferentiated state. Consistent with our hypothesis of a neural crest origin, DPSCs expressed multiple neural crest markers whereas they did not show mesodermal developmental markers. To determine that DPSCs fulfill the stem cell criteria, in addition to self-renewal, we also explored their multi-differentiation capacity with emphasis on neural crest lineages.
Explants of cranial and trunk neural crest cells give rise to chondrocytes, glia, neuronal cells, smooth muscle cells, and melanin-forming cells, which are all neural crest-lineage derived during embryogenesis 
. In our differentiation conditions, treated DPSCs expressed markers for osteoblast, chondrocyte, neuron, smooth muscle and adipocyte differentiation, indicating that non-clonal DPSCs contains multipotent stem cells with capacity to differentiate into mesenchymal and non-mesenchymal lineages of neural crest cells. Consistently human DPSCs can in vitro
differentiate into non-mesenchymal lineages 
DPSC clones, initiated from early passages of non-clonal cultures, were sub-cultured for 1 month without changing cell morphology. Like non-clonal DPSCs, all clones expressed Klf4, Nanog, and neural crest-related genes, suggesting that clones maintain neural-crest multipotential. Our observations showed DPSC clones gave rise to odontoblast-like, chondrocyte-like, adipocyte-like, smooth muscle-like, and neuronal-like cells.
To confirm the neural crest origin of DPSCs, we used a genetic lineage tracing model of neural crest derivatives. Although the Wnt1-Cre/R26R-LacZ
mouse has been widely used to study neural crest development, to date the origin of DPSCs has not been examined 
. Staining results of Wnt1-Cre/R26R-LacZ
tooth sections showed the existence of neural crest-derived DPSCs in sub-odontoblastic and perivascular regions. Those two areas have been reported as stem cell niches in dental pulp tissue 
. Under our culture conditions, early dental pulp cells isolated from the Wnt1-Cre/R26R-LacZ
revealed the majority of cells were of neural crest origin. This is consistent with the study by Chai et al (2000) showing that dental pulp consists of cells from neural crest and non-neural crest origins 
. In addition, Wnt1-Cre/R26R-LacZ
-derived BMSCs, previously known as non-neural crest-derived used as a control showed negative X-gal staining 
The X-gal staining of both DPSCs and BMSCs indicates that the Wnt1-Cre/R26R-LacZ is a good strategy to study the developmental origin of DPSCs. This model definitely confirms the neural crest origin of DPSCs. In spite of a mixed cell population of DPSC in early culture, after 6-week-in vitro expansion, all cells were positive by X-gal staining indicating only neural crest-derived cells remained and were present in late cultures. The Wnt1-Cre/R26R-LacZ-derived DPSC clones also expressed strongly pluripotent stem cell and neural crest-related genes, and gave rise to neural crest lineages. In addition to the expression of neural crest genes and differentiation capacity into neural crest-lineage, this lineage tracing data undeniably proves that in our culture condition DPSCs are neural crest-derived stem cells.
To further assess DPSC plasticity, we transplanted cultured DPSCs to examine their response in vivo
in two different models: subcutaneous (SC) and intramuscular (IM) transplantations. Interestingly, each transplantation method resulted in different cell fates indicating that DPSCs are not under a fixed program as environmental factors influence their response. Short subcutaneous transplantation with HAp/TCP resulted in the production of a disorganized dentin-like structure, mimicking a structure generated in reparative dentin formation after tooth injury 
. Reparative dentin exhibits disorganized arrangement due to lack of signaling from inner enamel epithelium and basement membrane which results in multiple inducing molecules secreted from dentin matrix 
. Lack of epithelium induction may explain why DPSC clones generated a disorganized structure. Like the renal capsule or subcutaneous regions, we used the intramuscular model to transplant DPSCs since we hypothesized that a high vascularized system in muscle may enhance the function of DPSCs 
. In turn, IM transplantation of DPSCs generated well-organized but immature collagen-forming matrix. These results may be explained by several possibilities. First, each transplant environment has a distinct influence on differentiation of DPSCs. Secondly, the IM transplanted structure was more organized possibly because muscle fibers function as scaffold to stabilize and directly instruct transplanted cells toward the proper arrangement and orientation. Thirdly, the length of transplantation in the IM injections was only 2 weeks, 3 weeks shorter than SC transplantation, and thus, longer IM transplantation may result in more mature matrix formation.
To determine if longer transplantations are required for formation of more mature matrices, we performed IM and SC transplantations for 12 weeks. Surprisingly, longer in vivo exposure did not result in more mature matrix formation in the intramuscular transplants. However, longer subcutaneous transplantations with HAp/TCP resulted in more mature matrix formation in both DPSC and BMSC transplants but the morphology and matrix compositions differed.
First, we observed many abundant small blood vessels in close proximity to odontoblast-like cells in DPSC transplants. Many of these microvessels were fenestrated and stained positive for VEGFR-3. We observed pericyte-like cells adjacent to these microvessels; some of which were donor DPSC-derived. This observation illustrates the perivascular nature of DPSCs and their potential to function as pericytes 
. Although BMSC transplants also contained abundant microvessels, bone condensations were significantly distant to vessels. Interestingly, in BMSC transplantations we observed a large number of microvessels in areas rich in adipocytes. These cells may be from BMSC-derived or derived from host cells which were signaled by BMSCs to differentiate into adipocytes.
Secondly, long-term DPSC transplantation resulted in a more organized mineralized matrix formed by elongated and polarized cells resembling odontoblast-like morphology which were in close proximity to microvessels. These matrices represent dentin-like structure with strong positive staining for dentin matrix proteins, DMP1, DSP, and BSP. In contrast, BMSC transplantation showed mineralized matrices with osteocyte-like cells in lacunae. Consistently, the matrices produced by BMSCs were positive for the bone protein, BSP, but not positive for the dentin protein, DSP. However, some studies reported that certain states of bone or tooth development can express low levels of dentin or bone proteins, respectively; therefore, we cannot distinguish bone or dentin structures by only protein expression 
. As mentioned above, the combination of cell/tissue morphology and matrix protein expression is important to characterize the structure created by DPSCs and BMSCs 
Thirdly, as described previously in human DPSC in vivo
transplantation, murine DPSCs can create collagen fibers arranging perpendicularly to the scaffold/matrix surface whereas BMSCs created collagen fibers running parallel to the scaffold/matrix surface 
. Lastly, the distribution and association of odontoblasts with microvessels resembles dentinogenesis whereas the formation of bone condensation in avascular areas resembles early osteogenesis 
In summary, we showed dental pulp cells in mouse neonates with neural crest-related stem cell properties. Although some studies showed that both neural crest and non-neural crest cells can be found in dental pulp during tooth development, our stem cell conditions select for only the outgrowth of neural crest-derived stem cells (). Thus, future studies should aim at comparing the differentiation capacity, contribution and cross-talk of neural crest- and non-neural crest-derived cells from dental pulp. Upon transplantation, DPSCs formed dentin-like matrix composed of odontoblast-like cells and pericytes associated with microvessels in dental pulp-like tissue recapitulating dentinogenesis ().
Summary of results and working models of neural crest-derived stem cells isolated from murine dental pulp tissue.
The mechanisms that control the differentiation fate of dental pulp stem cells are not completely elucidated but interactions with their niche are important for maintenance of stemness and induction/guidance for differentiation. In our transplantation model, interactions with microvessels seem pivotal for the differentiation of DPSCs into pericytes and odontoblasts. Since the dental pulp tissue is highly vascularized, it is probable that dental pulp stem cells reside in a vascular niche. To date multiple reports have revealed the presence of various adult stem cells residing within the vascular niches 
In addition, several recent studies have shown the existence of neural crest stem cells in various postnatal tissues 
. Thus, future studies should also aim at understanding the interactions of neural crest-derived stem cells with their respective niches from multiple adult tissues. Understanding the mechanisms that control organogenesis and adult tissue homeostasis will be instrumental for tissue engineering and tissue regeneration. Thus, studying how neural crest-derived stem cells from multiple organs interact within each niche to remain either stem cells or differentiate may help elucidate the mechanisms for maintenance of tissue-specific stem cells and tissue- specific differentiation. This will broaden the potential clinical applications of DPSCs as a readily accessible source for tissue regeneration not only of pulp regeneration but possibly other neural crest derivatives.