Thrombin, a multifunctional serine protease, plays central functions in haemostasis but also promotes a wide range of cellular responses via an interaction with specific seven-transmembrane domains receptors . Thrombin and perhaps other coagulant proteases, mediates proliferative responses triggered by tissue damage [18–20]. In vitro, thrombin is the most potent activator of blood platelets [21, 22], is chemotactic for monocytes [21, 23], and is a potent mitogenic factor for vascular smooth-muscle cells, fibroblasts [21, 25] and vascular endothelial cells [26, 27]. Thrombin signalling is mediated by a family of G protein-coupled protease-activated receptors (PARs), for which PAR1 is the prototype . Thrombin activates PAR1 through binding to and subsequent cleavage of the N-terminal domain of the receptor to expose a new amino-terminus that then acts as a tethered ligand to initiate intracellular signalling by the receptor . We found on both radial glial cells and rat MSCs a membranous expression of PAR1 suggesting that thrombin may act on these cells physiology. We also found that PAR1 is expressed in the nucleus and cytoplasm. The intracellular staining for PAR1 can be explained by the presence of intracellular store of PAR1 in the cells .
There is increasing evidence suggesting that members of the serine protease family, including thrombin, chymotrypsin, urokinase plasminogen activator, and kallikrein, may play a role in normal development and/or pathology of the nervous system. Thrombin-like proteases have been shown to exert deleterious effects on different neuronal and non-neuronal cell populations in vitro, including neurite retraction and death [29–31]. Moreover, the proteolytic activity of thrombin has been shown to inhibit long-term morphological differentiation in serum-free cultures of several cell types, including spinal cord and brain neurons, neuroblastoma cells, astroglial and neuroepithelial cells [32–35].
Nestin expression in the adult nervous system has been detected in a number of pathological conditions including cerebral ischemia [36–38]., and traumatic brain injury . Nakamura et al. have demonstrated that intracerebral injections of low doses of thrombin that induces neuroprotection without causing detectable brain injury resulted in a marked increase in nestin expression indicating that this could constitute a protective mechanism induced in astrocytes by levels of stress that do not produce a definable lesion. Nestin expression could, therefore, represent an embryonic reversion of the mature cytoskeleton that may aid in the response to and recovery from a wide variety of cerebral injuries, but might also perhaps aid in damage prevention.
In this study, we demonstrate that nestin expression is regulated by thrombin in opposite manner in RG and MSCs although the proliferation of both cell types is stimulated by this protease. Thrombin effect is thus associated in both cell types with a proliferating, undifferentiated state but in RG cells this involves the induction of nestin expression, a marker of immaturity for neural progenitors. In MSCs however, nestin expression corresponds to the first step toward acquisition of a neural fate and thus corresponds to a progression from the mesenchymal "undifferentiated", proliferating phenotype [3–5]. Maintenance of MSCs proliferation in the presence of thrombin would indeed be associated with an inhibition of nestin expression by MSCs as it constitutes one feature of their undifferentiated state. A possible selection of one population type over the other can be discounted as an explanation for the decrease of nestin expression by MSCs in the presence of thrombin because this effect still occurs in cells that cannot proliferate when treated with FUdR. Furthermore, the absence of apoptotic cells in the cultures treated with thrombin as well as the similar proliferation rates for both nestin-negative and nestin-positive MSCs which are induced in the presence of serum or thrombin in the growth medium demonstrate that thrombin represses directly the expression of nestin by MSCs and does not act by selecting the population of nestin-negative cells.
Thus unlike radial glial cells, MSCs in vitro respond to thrombin by a decrease of nestin expression. The nestin gene contains two tissue-specific transcriptional elements : a cis-element in the first intron drives nestin expression in somitic muscle precursors, while an enhancer in the second intron directs expression to CNS precursor cells  and thus to radial glial cells. Which of these two regulatory elements controls nestin expression by MSCs remains to be determined.
Finally, we demonstrate that nestin expression by MSCs decreases at a high cell density, as observed by immunocytofluorescence and Western blot assay. Tropepe et al. have found that embryonic stem cells express nestin when they differentiate into neural cells. This expression and the subsequent neural differentiation of embryonic stem cells are inhibited by a high cell density suggesting a similar behaviour for both MSCs and embryonic stem cells. It has also been demonstrated that a high cell density could accelerate the differentiation of human bone marrow MSCs into chondrocyte in a chondrogenic differentiation medium . However, this chondrogenic differentiation is mediated by soluble factors while the decrease of nestin expression observed in our experiments is mediated by a direct cell-to-cell contact. Since our experiments were made with MSCs obtained after more than 10 passages, we cannot exclude that these apparently different regulatory processes for the high cell density effect on MSCs mesodermal fate result from the use of MSCs of early and late passages.
There are however indications that MSCs of late passage secrete a soluble factor that can act on other cell types. Indeed Wislet-Gendebien et al. have already demonstrated that nestin-positive MSCs secrete BMP4, a member of TGFβ family, which stimulates the astrocytic differentiation of co-cultured neural stem cells, a maturation process that involves a loss of nestin expression by these cells. However, our co-culture experiments did not show any significant decrease in the proportion of nestin-positive MSCs when cells at a permissive density were grown with other MSCs at higher densities in a setup that allows no direct contact, suggesting rather a requirement for a direct cell-to-cell contact for nestin repression. Alternatively, high cellular density by increasing the direct cell-to-cell contact in the culture dish could inhibit specifically the proliferation of nestin-positive MSCs which then could explain their decrease in number as the overall cellular density increases.
During the last few years, a number of studies have addressed the phenotypic plasticity of MSCs. Most of these studies were performed in vivo and demonstrated that environmental factors play important roles in determining the ability of grafted MSCs to adopt a neural-like phenotype. Nakano et al. showed that murine bone marrow cells adopt a neural fate when they are directly injected into the striatum of previously irradiated mice . Similarly, MSCs also adopt neural fate after systemic injection in lethally irradiated mice . Interestingly, the systemic injection of MSCs in non-irradiated but brain-lesioned mice had positive effects on injury repair, but very few MSCs-derived cells expressed neural marker in such conditions.