Up‐regulation of semaphorin 4A expression in human retinal pigment epithelial cells by PACAP released from cocultured neural cells

Development and homeostasis of multicellular organisms require interactions between neighbouring cells. We recently established an in vitro model of cell–cell interaction based on a collagen vitrigel membrane. We have now examined the role of neural cells in retinal homeostasis by coculture of human retinal pigment epithelial (RPE) cells and neural cells on opposite sides of such a membrane. The neural cells (differentiated PC12 cells) induced up‐regulation of semaphorin 4A (Sema4A), a member of the semaphorin family of neural guidance proteins, in RPE (ARPE19) cells. This effect of the neural cells was mimicked by the neuropeptide pituitary adenylate cyclase–activating polypeptide (PACAP) and was abolished by the PACAP antagonist PACAP(6–38). Coculture with neural cells or stimulation with PACAP also induced the phosphorylation of extracellular‐signal‐regulated kinase in ARPE19 cells, and this effect of the neural cells was inhibited by PACAP(6–38). Finally, among various cytokines examined, only the amount of interleukin‐6 released by cocultures of ARPE19 and neural cells differed from that released by ARPE19 cells cultured alone. Interleukin‐6 was not detected in culture supernatants of neural cells, and the reduction in the amount of interleukin‐6 released by the cocultures compared with that released by ARPE19 cells alone was prevented by PACAP(6–38). Our findings suggest that PACAP released from retinal neural cells (photoreceptors or optic nerve cells) may regulate Sema4A expression in RPE cells and thereby contribute to the maintenance of retinal structure and function. Development and homeostasis of multicellular organisms require interactions between neighbouring cells. With the use of a coculture system based on a collagen vitrigel membrane, we have now shown that neural cells induce up‐regulation of the neural guidance protein Sema4A in RPE cells. This effect of neural cells appears to be mediated by the neuropeptide PACAP. PACAP released from retinal neural cells (photoreceptors or optic nerve cells) may thus regulate Sema4A expression in RPE cells and thereby contribute to the maintenance of retinal structure and function. Copyright © 2014 John Wiley & Sons, Ltd.     

A morphological and experimental study of the mesencephalic neural crest cells in the mouse embryo using wheat germ agglutinin-gold conjugate as the cell marker

The distribution of the mesencephalic neural crest cells in the mouse embryo was studied by mapping the colonization pattern of WGA-gold labelled cells following specific labelling of the neuroectoderm and grafting of presumptive neural crest cells to orthotopic and heterotopic sites. The result showed that (1) there were concomitant changes in the morphology of the neural crest epithelium during the formation of neural crest cells, in the 4- to 7-somite-stage embryos, (2) the neural crest cells were initially confined to the lateral subectodermal region of the cranial mesenchyme and there was minimal mixing with the paraxial mesoderm underneath the neural plate, (3) labelled cells from the presumptive crest region colonized the lateral cranio-facial mesenchyme, the developing trigeminal ganglion and the pharyngeal arch, (4) the formation of neural crest cells was facilitated by the focal disruption of the basal lamina and the cell-cell interaction specific to the neural crest site and (5) the trigeminal ganglion was colonized not only by neural crest cells but also by cells from the ectodermal placode.     

Mapping of neural crest pathways in Xenopus laevis using inter- and intra-specific cell markers

This study examines the pathways of migration followed by neural crest cells in Xenopus embryos using two recently described cell marking techniques. The first is an interspecific chimera created by grafting Xenopus borealis cells into Xenopus laevis hosts. The cells of these closely related species can be distinguished by their nuclear dimorphism. The second type of marker is created by microinjection of lysinated dextrans into fertilized eggs which can then be used for intraspecific grafting. These recently developed fluorescent dyes are fixable and identifiable in both living and fixed embryos. After grafting labeled donor neural tubes into unlabeled host embryos, the distribution of neural crest cells at various stages after grafting was used to define the pathways of neural crest migration. To control for possible grafting artifacts, fluorescent lysinated dextran was injected into a single blastomere which gives rise to a large number of neural crest cells, thereby labeling the neural crest without grafting. By all three techniques, Xenopus neural crest cells were observed along two predominant pathways in the trunk. The majority of neural crest cells were observed along a "ventral" route, between the neural tube and somite, the notochord and somite, and along the dorsal mesentery. A second group of neural crest cells was observed "dorsally" where they populated the dorsal fin. A third minor "lateral" pathway was observed primarily in borealis/laevis chimerae and in blastomere-injected embryos; some neural crest cells were observed underneath the ectoderm lateral to the neural tube. Along the rostrocaudal axis, neural crest cells were not continuously distributed but were primarily located across from the caudal two-thirds of the somite. Fewer than 3% of the neural crest cells were observed across from the rostral third of each somite. When grafted to ventral locations, neural crest cells were not able to migrate dorsally but migrated laterally along the dorsal mesentery. Labeled neural crest cells gave rise to cells of the spinal, sympathetic, and enteric ganglia as well as to adrenal chromaffin cells, Schwann cells, pigment cells, mesenchymal cells of the dorsal fin, and some cells in the integuments and in the region of the pronephros. These results show that the neural crest migratory pathways in Xenopus differ from those in the avian embryo. In avians NC cells migrate as a closely associated sheet of cells while in Xenopus they migrate as individual cells. Both species exhibit a metamerism in the neural crest cell distribution pattern along the rostrocaudal axis.(ABSTRACT TRUNCATED AT 400 WORDS)     

Origins and Developmental Potential of the Neural Crest

Neural crest cells are a migratory population that forms most of the peripheral nervous system, facial skeleton, and numerous other derivatives. These cells arise from the neural ectoderm and are first recognizable as discrete cells after neural tube closure. In this review, I summarize the results of studies from our laboratory on neural crest cell lineage and origin. Our recent experiments demonstrate that interactions between the presumptive neural plate and the nonneural ectoderm are likely to be instrumental in the induction of the avian neural crest. Juxtaposition of these tissues at early stages results in the formation of neural crest cells at the interface. However, neural crest cells do not appear to be segregated from other neuroepithelial cells; cell lineage studies have demonstrated that individual precursor cells within the neural tube can give rise to both neural crest and neural tube derivatives as diverse as sensory, commissural, and motor neurons. This suggests that individual neuroectodermal cells are multipotent, such that a precursor within the neural tube has the ability to form both neural tube (central nervous system) and neural crest (peripheral nervous system and other) derivatives. Further support for flexibility in the developmental program of neuroepithelial cells comes from experiments in which the cranial neural folds are ablated; this results in regulation by the remaining ventral neural tube cells to form neural crest cells after the endogenous neural crest is removed. At later stage of development, this regulative capacity is lost. Following their emigration from the neural tube, neural crest cells become progressively restricted to defined embryonic states. Taken together, these experiments demonstrate that: (1) the neural crest is an induced population that arises by interactions within the ectoderm; (2) initially, progenitor cells are multipotent, having the potential to form multiple neural crest and neural tube derivatives; and (3) with time, the precursors become progressively restricted to form neural crest derivatives and eventually to individual phenotypes.     

Connexin 43 expression reflects neural crest patterns during cardiovascular development

We used transgenic mice in which the promoter sequence for connexin 43 linked to a lacZ reporter was expressed in neural crest but not myocardial cells to document the pattern of cardiac neural crest cells in the caudal pharyngeal arches and cardiac outflow tract. Expression of lacZ was strikingly similar to that of cardiac neural crest cells in quail-chick chimeras. By using this transgenic mouse line to compare cardiac neural crest involvement in cardiac outflow septation and aortic arch artery development in mouse and chick, we were able to note differences and similarities in their cardiovascular development. Similar to neural crest cells in the chick, lacZ-positive cells formed a sheath around the persisting aortic arch arteries, comprised the aorticopulmonary septation complex, were located at the site of final fusion of the conal cushions, and populated the cardiac ganglia. In quail-chick chimeras generated for this study, neural crest cells entered the outflow tract by two pathways, submyocardially and subendocardially. In the mouse only the subendocardial population of lacZ-positive cells could be seen as the cells entered the outflow tract. In addition lacZ-positive cells completely surrounded the aortic sac prior to septation, while in the chick, neural crest cells were scattered around the aortic sac with the bulk of cells distributed in the bridging portion of the aorticopulmonary septation complex. In the chick, submyocardial populations of neural crest cells assembled on opposite sides of the aortic sac and entered the conotruncal ridges. Even though the aortic sac in the mouse was initially surrounded by lacZ-positive cells, the two outflow vessels that resulted from its septation showed differential lacZ expression. The ascending aorta was invested by lacZ-positive cells while the pulmonary trunk was devoid of lacZ staining. In the chick, both of these vessels were invested by neural crest cells, but the cells arrived secondarily by displacement from the aortic arch arteries during vessel elongation. This may indicate a difference in derivation of the pulmonary trunk in the mouse or a difference in distribution of cardiac neural crest cells. An independent mouse neural crest marker is needed to confirm whether the differences are indeed due to species differences in cardiovascular and/or neural crest development. Nevertheless, with the differences noted, we believe that this mouse model faithfully represents the location of cardiac neural crest cells. The similarities in location of lacZ-expressing cells in the mouse to that of cardiac neural crest cells in the chick suggest that this mouse is a good model for studying mammalian cardiac neural crest and that the mammalian cardiac neural crest performs functions similar to those shown for chick.     

Differentiation state determines neural effects on microvascular endothelial cells

Growing evidence indicates that nerves and capillaries interact paracrinely in uninjured skin and cutaneous wounds. Although mature neurons are the predominant neural cell in the skin, neural progenitor cells have also been detected in uninjured adult skin. The aim of this study was to characterize differential paracrine effects of neural progenitor cells and mature sensory neurons on dermal microvascular endothelial cells. Our results suggest that neural progenitor cells and mature sensory neurons have unique secretory profiles and distinct effects on dermal microvascular endothelial cell proliferation, migration, and nitric oxide production. Neural progenitor cells and dorsal root ganglion neurons secrete different proteins related to angiogenesis. Specific to neural progenitor cells were dipeptidyl peptidase-4, IGFBP-2, pentraxin-3, serpin f1, TIMP-1, TIMP-4 and VEGF. In contrast, endostatin, FGF-1, MCP-1 and thrombospondin-2 were specific to dorsal root ganglion neurons. Microvascular endothelial cell proliferation was inhibited by dorsal root ganglion neurons but unaffected by neural progenitor cells. In contrast, microvascular endothelial cell migration in a scratch wound assay was inhibited by neural progenitor cells and unaffected by dorsal root ganglion neurons. In addition, nitric oxide production by microvascular endothelial cells was increased by dorsal root ganglion neurons but unaffected by neural progenitor cells.     

The developmental potentials of the caudalmost part of the neural crest are restricted to melanocytes and glia

The avian spinal cord is characterized by an absence of motor nerves and sensory nerves and ganglia at its caudalmost part. Since peripheral sensory neurons derive from neural crest cells, three basic mechanisms could account for this feature: (i) the caudalmost neural tube does not generate any neural crest cells; (ii) neural crest cells originating from the caudal part of the neural tube cannot give rise to dorsal root ganglia or (iii) the caudal environment is not permissive for the formation of dorsal root ganglia. To solve this problem, we have first studied the pattern of expression of ventral (HNF3beta) and dorsal (slug) marker genes in the caudal region of the neural tube; in a second approach, we have recorded the emergence of neural crest cells using the HNK1 monoclonal antibody; and finally, we have analyzed the developmental potentials of neural crest cells arising from the caudalmost part of the neural tube in avian embryo in in vitro culture and by means of heterotopic transplantations in vivo. We show here that neural crest cells arising from the neural tube located at the level of somites 47-53 can differentiate both in vitro and in vivo into melanocytes and Schwann cells but not into neurons. Furthermore, the neural tube located caudally to the last pair of somites (i.e. the 53rd pair) does not give rise to neural crest cells in any of the situations tested. The specific anatomical aspect of the avian spinal cord can thus be accounted for by limited developmental potentials of neural crest cells arising from the most caudal part of the neural tube.     

Direct neural fate specification from embryonic stem cells: a primitive mammalian neural stem cell stage acquired through a default mechanism

Little is known about how neural stem cells are formed initially during development. We investigated whether a default mechanism of neural specification could regulate acquisition of neural stem cell identity directly from embryonic stem (ES) cells. ES cells cultured in defined, low-density conditions readily acquire a neural identity. We characterize a novel primitive neural stem cell as a component of neural lineage specification that is negatively regulated by TGFbeta-related signaling. Primitive neural stem cells have distinct growth factor requirements, express neural precursor markers, generate neurons and glia in vitro, and have neural and non-neural lineage potential in vivo. These results are consistent with a default mechanism for neural fate specification and support a model whereby definitive neural stem cell formation is preceded by a primitive neural stem cell stage during neural lineage commitment.     

The origins of neural crest cells in the axolotl

We address the question of whether neural crest cells originate from the neural plate, from the epidermis, or from both of these tissues. Our past studies revealed that a neural fold and neural crest cells could arise at any boundary created between epidermis and neural plate. To examine further the formation of neural crest cells at newly created boundaries in embryos of a urodele (Ambystoma mexicanum), we replace a portion of the neural folds of an albino host with either epidermis or neural plate from a normally pigmented donor. We then look for cells that contain pigment granules in the neural crest and its derivatives in intact and sectioned host embryos. By tracing cells in this manner, we find that cells from neural plate transplants give rise to melanocytes and (in one case) become part of a spinal ganglion, and we find that epidermal transplants contribute cells to the spinal and cranial ganglia. Thus neural crest cells arise from both the neural plate and the epidermis. These results also indicate that neural crest induction is (at least partially) governed by local reciprocal interactions between epidermis and neural plate at their common boundary.     

Behavior of neural crest cells on embryonic basal laminae

Neural crest cells separate from the neural epithelium in a region devoid of a basal lamina and migrate along pathways bordered by intact basal laminae. The distribution of basal laminae suggests that they might have an important role in the morphogenesis of the neural crest by acting as a barrier to migration. The experiments reported here have tested directly whether neural crest cells can penetrate a basal lamina. Isolated neural tubes, neural crest cells cultured for 24 hr, or pigmented neural crest cells were explanted onto human placental amnions from which the epithelium had been removed to expose the basal lamina. In no case did neural crest cells or crest derivatives penetrate the basal lamina to invade the underlying stroma. If crest cells were grown on the stromal side of the amnion, they invaded the connective tissue. Pigmented neural crest derivative and [3H]thymidine-labeled nonpigmented crest cells were also confronted with chick embryonic basal laminae by grafting the cells into the lumen of the neural tube at the axial levels where host crest migration had commenced. Most of the grafted cells invaded the neural epithelium and accumulated after 24 hr at the basal surface of the neural tube. A few crest cells escaped through the dorsal surface of the neural tube and entered the overlying ectoderm, presumably through the wound created during the grafting procedure. Some of these grafted cells, located initially by light microscopy, were examined at the higher magnification and resolution offered by the transmission electron microscope to determine the relationship of the grafted cells to the basal lamina. In 50% (14 total) of the cases, the crest cells never reached the basal lamina of the neural tube, but were trapped by cell junctions between the neural epithelial cells. Of the remaining grafted cells that were relocated in the TEM (50%, total 15) all were spread on the basal lamina, but were not seen penetrating it. Likewise, in the three cases where crest cells were found in the epidermal ectoderm, all were in contact with the basal lamina of the ectoderm but did not have any processes extending through it. In three cases, at the level of the light microscope, crest cells were found to extend through the basal surface of the neural tube. In all these instances, the cells followed the dorsal root nerve exiting through a region of the neural tube that is devoid of a basal lamina.(ABSTRACT TRUNCATED AT 400 WORDS)     

Expression of Schwann cell markers by mammalian neural crest cells in vitro

During embryonic development, neural crest cells differentiate into a wide variety of cell types including Schwann cells of the peripheral nervous system. In order to establish when neural crest cells first start to express a Schwann cell phenotype immunocytochemical techniques were used to examine rat premigratory neural crest cell cultures for the presence of Schwann cell markers. Cultures were fixed for immunocytochemistry after culture periods ranging from 1 to 24 days. Neural crest cells were identified by their morphology and any neural tube cells remaining in the cultures were identified by their epithelial morphology and immunocytochemically. As early as 1 to 2 days in culture, approximately one third of the neural crest cells stained with m217c, a monoclonal antibody that appears to recognize the same antigen as rat neural antigen-1 (RAN-1). A similar proportion of cells were immunoreactive in cultures stained with 192-IgG, a monoclonal antibody that recognizes the rat nerve growth factor receptor. The number of immunoreactive cells increased with time in culture. After 16 days in culture, nests of cells, many of which had a bipolar morphology, were present in the area previously occupied by neural crest cells. The cells in the nests were often associated with neurons and were immunoreactive for m217c, 192-IgG and antibody to S-100 protein and laminin, indicating that the cells were Schwann cells. At all culture periods examined, neural crest cells did not express glial fibrillary acidic protein. These results demonstrate that cultured premigratory neural crest cells express early Schwann cell markers and that some of these cells differentiate into Schwann cells. These observations suggest that some neural crest cells in vivo may be committed to forming Schwann cells and will do so provided that they then proceed to encounter the correct environmental cues during embryonic development.     

Acetylcholinesterase activity in neural crest cells of the early chick embryo

The appearance and distribution of AChE activity in the neural crest cells of the chick embryo were histochemically investigated. Prior to closure of the neural tube, neural crests were not demonstrated and most of the cells constituting the neural plate and the more lateral ectoderm were AChE-negative. With the closure of the neural tube, the neural crests assumed the form of a cell mass in its mid-dorsal portion and AChE activity was demonstrated in some elements of both tube and crests. The neural crest cells beginning to migrate ventrally or laterally were AChE-positive, and some showed intense enzymatic activity. Electron microscopically, the neural crest cells and the cells migrating from the neural crest displayed AChE activity in the cisternae of the nuclear envelope and in a few r-ER profiles, but were morphologically undifferentiated. As assessed by 3H-thymidine autoradiography, these cells possessed the potential to proliferate. These findings indicate that with the formation of the neural tube and neural crest, cells constituting these structures begin to differentiate with respect to AChE activity and that the enzyme appears in the neural crest cells before the onset of neuronal differentiation.     

Analysis of neural crest cell lineage and migration.

In this review, we describe the results of recent experiments designed to investigate various aspects of neural crest cell lineage and migration. We have analyzed the lineage of individual premigratory neural crest cells by injecting a fluorescent lineage tracer dye, lysinated fluorescein dextran, into cells within the dorsal neural tube. Individual clones contained cells that were located in very diverse sites consistent with their being sensory neurons, prepigment cells, Schwann cells, adrenergic cells, and neural tube cells. These results suggest that some neural crest cells in the trunk and cranial regions are multipotent prior to their emigration from the neural tube. The environment through which neural crest cells move influences both the pattern and direction of their migration. We have shown that the sclerotomal portion of the somites are responsible for the rostrocaudal pattern of trunk neural crest cell movement, whereas the neural tube appears to govern the dorsoventral position of neural crest-derived ganglia. In addition, the notochord inhibits the movement of neural crest cells. In order to understand necessary cell-matrix interactions in neural crest migration, we have performed perturbation experiments, in which antibodies directed against cell surface or extracellular matrix molecules were introduced along neural crest pathways. We find that integrins, fibronectin, laminin, and tenascin all play some role in cranial neural crest emigration. Thus, multiple factors may be involved in controlling neural crest cell migration, and different factors may be important for migration in different regions of the embryo.     

Model systems for the study of heart development and disease. Cardiac neural crest and conotruncal malformations

Neural crest cells are multipotential cells that delaminate from the dorsal neural tube and migrate widely throughout the body. A subregion of the cranial neural crest originating between the otocyst and somite 3 has been called "cardiac neural crest" because of the importance of these cells in heart development. Much of what we know about the contribution and function of the cardiac neural crest in cardiovascular development has been learned in the chick embryo using quail-chick chimeras to study neural crest migration and derivatives as well as using ablation of premigratory neural crest cells to study their function. These studies show that cardiac neural crest cells are absolutely required to form the aorticopulmonary septum dividing the cardiac arterial pole into systemic and pulmonary circulations. They support the normal development and patterning of derivatives of the caudal pharyngeal arches and pouches, including the great arteries and the thymus, thyroid and parathyroids. Recently, cardiac neural crest cells have been shown to modulate signaling in the pharynx during the lengthening of the outflow tract by the secondary heart field. Most of the genes associated with cardiac neural crest function have been identified using mouse models. These studies show that the neural crest cells may not be the direct cause of abnormal cardiovascular development but they are a major component in the complex tissue interactions in the caudal pharynx and outflow tract. Since, cardiac neural crest cells span from the caudal pharynx into the outflow tract, they are especially susceptible to any perturbation in or by other cells in these regions. Thus, understanding congenital cardiac outflow malformations in human sequences of malformations as represented by the DiGeorge syndrome will necessarily require understanding development of the cardiac neural crest.     

Appearance of nerve growth factor receptors on cultured neural crest cells

Light microscopic radioautography of differentiating quail neural crest cultures (1 to 2 weeks after explanation) incubated with Iodine-125-labeled nerve growth factor (125I-NGF) revealed that approximately 35% of the cells bound NGF. The binding was specific and saturable; it was blocked by an excess of nonradioactive NGF, and was not detected following incubation with biologically inactive 125I-NGF. In addition, the binding did not appear to be blocked or diminished by insulin. Cell cultures prepared from somites or notochord showed no specific binding of 125I-NGF. Melanocytes comprised approximately 10% of the cell population in these cultures and appeared to be unlabeled. The subpopulation of cells with NGF receptors that were morphologically similar to other non-melanocyte unlabeled cells present in the neural crest cultures are probably the targets of the factor during differentiation and development. In contrast, there was no evidence of 125I-NGF binding by premigratory neural crest (adherent to the isolated neural tube) or by early migratory neural crest cells (24 hr after explantation). Both of these types of neural crest cells are relatively undifferentiated. The cells of the neural tube were also unlabeled. The binding of 125I-NGF to differentiating neural crest cells was not noticeably diminished by a brief pretreatment with trypsin or Dispase, enzymes used in the isolation of neural tubes. Hence, the absence of NGF receptors on premigratory neural crest and early migratory neural crest cultures was not due to enzymatic alterations of the receptor. It seems, therefore, that receptors for NGF appear on neural crest cells during the time when these cells are acquiring their phenotypic characteristics.     

Alterations in neural crest migration by a monoclonal antibody that affects cell adhesion

The possible role of a 140-kD cell surface complex in neural crest adhesion and migration was examined using a monoclonal antibody JG22, first described by Greve and Gottlieb (1982, J. Cell. Biochem. 18:221-229). The addition of JG22 to neural crest cells in vitro caused a rapid change in morphology of cells plated on either fibronectin or laminin substrates. The cells became round and phase bright, often detaching from the dish or forming aggregates of rounded cells. Other tissues such as somites, notochords, and neural tubes were unaffected by the antibody in vitro even though the JG22 antigen is detectable in embryonic tissue sections on the surface of the myotome, neural tube, and notochord. The effects of the JG22 on neural crest migration in vivo were examined by a new perturbation approach in which both the antibody and the hybridoma cells were microinjected onto neural crest pathways. Hybridoma cells were labeled with a fluorescent cell marker that is nondeleterious and that is preserved after fixation and tissue sectioning. The JG22 antibody and hybridoma cells caused a marked reduction in cranial neural crest migration, a build-up of neural crest cells within the lumen of the neural tube, and some migration along aberrant pathways. Neural crest migration in the trunk was affected to a much lesser extent. In both cranial and trunk regions, a cell free zone of one or more cell diameters was generally observed between neural crest cells and the JG22 hybridoma cells. Two other monoclonal antibodies, 1-B and 1-N, were used as controls. Both 1-B and 1-N bind to bands of the 140-kD complex precipitated by JG22. Neither control antibody affected neural crest adhesion in vitro or neural crest migration in situ. This suggests that the observed alterations in neural crest migration are due to a functional block of the 140-kD complex.     

Neural stem cell differentiation is mediated by integrin beta4 in vitro

Neural stem cells are capable of differentiating into three major neural cell types, but the underlying molecular mechanisms remain unclear. Here, we investigated the mechanism by which integrin beta4 modulates mouse neural stem cell differentiation in vitro. Inhibition of endogenous integrin beta4 by RNA interference inhibited the cell differentiation and the expression of fibroblast growth factor receptor 2 but not fibroblast growth factor receptor 1 or fibroblast growth factor receptor 3. Overexpression of integrin beta4 in neural stem cells promoted neural stem cell differentiation. Furthermore, integrin beta4-induced differentiation of neural stem cells was attenuated by SU5402, the inhibitor of fibroblast growth factor receptors. Finally, we investigated the role of integrin beta4 in neural stem cell survival: knockdown of integrin beta4 did not affect survival or apoptosis of neural stem cells. These data provide evidence that integrin beta4 promotes differentiation of mouse neural stem cells in vitro possibly through fibroblast growth factor receptor 2.