The main findings of this work were that the short-term removal of growth factors is capable of (i) modifying the distribution of β-tubulin III and GFAP positive cells in whole neurospheres; (ii) increasing the number of β-tubulin III positive cells, while decreasing the number of Nestin positive cells (iii) inducing changes in gene expression profiles; and (iv) promoting neurite extension and changes in orientation of neural stem cells.
Here we show the ability of hNPC to survive and differentiate after culturing in suspension in the absence of growth factors. It has recently been shown that mNPC are able to survive and differentiate after 14 days growth factors starvation 
. However, MFM neurospheres did not grow as much as the CTR ones, as observed in growth rates (). As neurosphere size is established by the balance between proliferation and cell death, the stable size of MFM neurospheres is due to both proliferation and apoptosis decreasing (). As expected, BrdU incorporation decreases after mitogens removal (27.9% for CTR; 10.5% for MFM). On the other hand, we have to take in account that, even after growth factors removal, the level of apoptotic cells was extremely low (about 5%), again, showing that neurospheres should have a mechanism that prevents cell death in this condition. The increased apoptosis after 14 days in culture in the CTR group can be attributed to the increase in neurosphere size. As neurospheres grow, the cells from the core are exposed to lower concentration of growth factors, leading to an increased cell death (). This may be due to the fact that neurospheres deprived of growth factors in suspension are able to produce cytokines and growth factors that, at the same time, can induce differentiation and prevent cell death.
It could be argued that, owing to the fact that EGF and FGF-2 are known to induce self-renewal of NPC, the effects found after removal of these factors on cell differentiation is predictable and expected based on published literature. Yet, Caldwell and colleagues 
showed that the combination of cell–cell interactions during differentiation and growth factor administration, can increase the number of generated neurons . This is only one example of the importance of cell-cell interaction and of the aspects associated to its three-dimensional structure for neurosphere plasticity.
Staining for GFAP, β-tubulin III and Nestin in hNPC from the CTR and MFM groups cultured in suspension revealed two major findings. Removal of growth factors modifies the distribution and proportions of β-tubulin III, GFAP and Nestin positive cells in whole neurospheres. In the CTR group, GFAP (34%) was found across the whole neurosphere, β-tubulin III (4%) was preferably expressed in the neurosphere core () and Nestin (60%) was found in the border of the sphere (). The cell distributions for the CTR group are in agreement with the findings of Campos and colleagues 
for a three-dimensional model of mouse neurospheres. In the MFM group, β-tubulin III positive cells (32%) concentrated in the border of the neurospheres, while GFAP (41%) and Nestin (42%) positive cells kept the same localization. (). We highlight the fact that Nestin localization in the CTR group was the same as that of BrdU positive cells. Despite Nestin localization was not altered, it is important to notice that the percentage of positive cells significantly decreases, explaining the increased differentiation in the borders of the neurosphere after mitogens removal. β-tubulin III
expression increased in hNPC () and displayed a clear tendency of increasing in mNPC (Supplemental figure S1
), after growth factors removal, as showed by real-time PCR .Together, these results reinforce the hypothesis that mitogens removal, even without adhesion and migration, is responsible for an increased cell differentiation.
We hypothesize the shift in the distribution of β-tubulin III positive cells was caused by a decreased gradient of growth factors from the outer layer to the center of neurospheres. Given that the concentration of EGF and FGF-2 inside the neurosphere might be lower than in the outside (in spheres cultured in the presence of these mitogens), cells in the neurosphere core are able to stop proliferation and start differentiation even in suspension. It is reasonable to assume that cells in the inner portion of the neurosphere would more readily be deprived from growth factors and this, in turn, would trigger the endogenous production of growth factors, initially or mostly from the neurosphere core. Growth factors production by the core would be responsible for maintaining an undifferentiated state in the core while setting cells with neuronal markers in the borders of the neurosphere (that lack growth factors). In the core of the MFM groups, the smaller concentrations of such autocrine production of growth factors, as compared to exogenous administration in the medium, would be sufficient to maintain an undifferentiated state while insufficient to trigger mitotic activity. This represents a logical and possible hypothesis for explaining the altered distribution of β-tubulin III positive cells (core versus
shell) as a consequence of growth factors withdrawal. Taken together, these results show that, during growth factors removal, there is a significant shift in the expression of growth factors and neural specific markers (gfap
, β-tubulin III
) by NPC. It was already shown that mouse neurosphere cells deprived of growth factors are able to produce PDGF and FGF-2 among other factors 
. However, for human neurospheres that had not been described yet. Our Real Time-PCR results showed that hNPC from the MFM group had decreased expression of fgf-2
and increased expression of igf-1
(). This, in part, could explain the differences in the level of differentiation and in β-tubulin III positive cells distribution in the neurospheres before and after growth factors removal.
Immunocytochemistry assays showed that, after 7 days plating, hNPC differentiated into β-tubulin III and GFAP positive cells, but no oligodendrocytes for both the CTR and the MFM groups. Lack of oligodendrocytes after human neurosphere differentiation (plated or in suspension) must be attributed to the absence of proliferation of true stem cells in vitro
, as reported by Wright and colleagues 
. We demonstrate that human cell lines in the MFM group showed neurons with longer processes, while the CTR group originated neurons with shorter processes. The longer processes found after mitogens starvation can be explained by the fact that some growth factors that affect neurite extension are up regulated. For instance, PDGFb and IGF-1 promote neurite outgrowth 
. Given our finding of a significant increase in those growth factors expression after mitogens removal, we suggest that PDGFb and IGF-1 could be partially responsible for neurite extension in the MFM group. In addition, before starvation, processes lacked a clear orientation in relation to the neurosphere, while after starvation a substantial number of processes could be seen radially oriented from the neurospheres.
These features could be relevant for priming the cells just before implantation in a different environment in the course of regenerative strategies. Indeed, altering the conditions to which NPC are subjected just prior to being implanted in an injured structure might be critical for defining its potential for survival and integration. Here we did not evaluate whether or not this potential translates to the same effect in vivo. Yet, our data prompt us to hypothesize that growth factors starvation prior to NPC transplantation might enhance the ability of these cells to graft and provide functional recovery.
We also suggest that growth factors starvation prior to NPC transplantation might enhance the ability of these cells to graft and provide functional recovery. The observations concerning the organized sprouting of astrocytes and neurons after growth factors withdrawal could result in a better targeting of the transplanted cells to the lesion site. Rather than providing optimal surviving conditions for cultures, we hypothesize that adjusting culture parameters (including growth factors withdrawal) might be essential for achieving success when grafting these cells in damaged systems. Similar to ischemic cell conditioning where pre-exposure of neural cells to brief ischemic episodes render these cells rather resistant to subsequent ischemia 
, it is possible that transitory withdrawal of EGF and FGF-2 might trigger the expression of specific gene programs for differentiation and/or neuroprotection. The differences found in neuronal and glial differentiation between the CTR and MFM groups, give rise to a new question: Can growth factors removal influence the capability of migration, integration and differentiation of NPC after cell transplantation? This is an important issue to be evaluated and could be relevant for the functional recovery of neurological disorders.