This study details the use of a fully adherent and feeder-free differentiation protocol employing small molecules to obtain neural precursors for transplantation after stroke. Although there is still some batch-to-batch variation and some animal products are still used, this protocol reduces the heterogeneity present in suspension culture and reduces possible contamination from animal products by removing the use of feeder cells in cell culture. Here, we demonstrate that neural precursors derived by using this protocol can develop into electrophysiologically active neurons, suggesting that they have the potential to act as functional neurons in damaged tissue. We further demonstrate that neural precursors survive, differentiate into neurons, improve neural regeneration, and enhance sensory function after transplantation into the penumbra region of stroke.
As we demonstrated during terminal differentiation, it is possible for cells to express neuronal markers like NeuN and neurofilament without exhibiting mature electrophysiological function. It is important to ensure that cells intended to replace lost tissue in the brain can further differentiate into neurons and that those neurons can respond appropriately to electrical signals. However, many studies rely on protein expression, without testing for electrophysiological function. Johnson et al.  studied functional development in hES cell-derived (H9) neurons over the course of 7 weeks of terminal differentiation (10 weeks from the onset of differentiation in hES cells). PAX6+/SOX1+ progenitors were obtained within 2 weeks by using suspension culture and neural rosette isolation, similar to our time course. These were again cultured in suspension for 1 week before plating for terminal differentiation in a medium containing brain-derived neurotrophic factor and glial cell line-derived neurotrophic factor, among other factors. Electrophysiological properties were examined at 1, 3, 4, and 7 weeks after plating for terminal differentiation. High-amplitude, single-spike action potentials were first reported at 4 weeks of terminal differentiation, and repetitive trains were observed in some cells at 7 weeks. In contrast, we obtained high-amplitude, single-spike action potentials at only 2 weeks after plating for terminal differentiation, and bFGF was the only growth factor added to our base medium. We began to obtain repetitive trains at 3 weeks of terminal differentiation, and the proportion of cells firing them increased with another week of culture. We have therefore greatly reduced the time and cost associated with obtaining electrophysiologically active neurons in vitro. After transplantation, cells differentiated into neuronal cells. Although the present study could not verify the electrophysiological properties of these cells, behavioral tests support the hypothesis that the transplanted cells participated in functional repair of damaged brain structures.
In this study, we examined neuronal differentiation in vitro to confirm the ability of our hES cell-derived neural precursors to further differentiate into neurons. While these neurons were expressing receptor subunits and electrical activity consistent with an excitatory phenotype, we did not determine the exact subtype of neurons we derived in vitro. This determination, along with how environments approximating the stroke and penumbra region affect the differentiation, will be important as therapies move toward the clinic. To this same end, it will be important to further identify the non-neuronal cells in culture with an emphasis on demonstrating that the cell types derived become post-mitotic upon differentiation and do not form any inappropriate cell types. In this study, we used bFGF as our only recombinant growth factor, but it is possible that further patterning factors will increase the proportion of neurons in culture and permit the derivation of specific subtypes. Patterning factors are usually recombinant growth factors that can greatly increase the cost of culture, but small molecules may lead to decreased cost in this arena as well. For example, purmorphamine is a sonic hedgehog (shh) agonist that has been used in the derivation of dopaminergic neurons .
In vivo, we observed a very high degree of cell survival after transplantation. This may have been due, in part, to the presence of Matrigel throughout the differentiation process. It was recently reported that hES cell-derived neural precursors cultured with Matrigel before transplantation or injected with a Matrigel scaffold reduced infarct size, improved behavioral outcomes, and differentiated primarily into neuronal cells . However, cells that were not exposed to Matrigel exhibited high levels of cell death and lower proportions of neuronal markers and did not improve infarct size or behavioral outcomes. We have also reported positive effects of Matrigel on hES cell-derived neural precursors in vitro, where we found that cells terminally differentiated on poly-D-lysine/laminin-coated dishes never developed mature action potential responses but that those grown on Matrigel-coated dishes did . Thus, the use of Matrigel throughout our differentiation process may have contributed to the positive results we report here. However, Matrigel will need to be removed from the process if it is ever used in human trials, as the removal of xenogenic products is largely seen as necessary for widespread clinical use .
One major concern with the use of pluripotent stem cells in cell therapy is the fear of tumor formation. We did not observe any teratoma formation and this was likely due the lack of residual pluripotent cells in our cultures. However, pluripotent cells need not be present for tumor formation. For example, small rosette-like tumors can form if hES cell-derived neural precursors are transplanted at a stage of differentiation in which cells are highly proliferative but not yet similar to fetal brain in the expression of neural markers . These cells expressed high levels of PAX6 but had only low-level expression of SOX1, whereas the neural precursor stage, which did not result in any tumor formation, expressed high levels of both markers. The cells we obtain with small-molecule SMAD inhibition also highly expressed both of these markers, and we observed no adverse effects from cell proliferation in the brain tissue. In fact, the total numbers of BrdU-positive cells found in the control and transplant groups were not significantly different, suggesting low levels of proliferation in the graft.
Another concern in our model may have been the use of the Hoechst tag for tracking, as it has the potential to cause problems in DNA replication or leak into neighboring cells. However, this tag has been successfully used in prior studies , and we did not observe any tumor formation in vivo. As argued earlier, it is likely that the transplanted cells were not proliferative, mitigating any problems with DNA replication. In future studies, a better tracking method may be something akin to that used by Daadi et al. . Transfection of cells with an easily identifiable marker that does not leak and can be easily co-stained would be ideal. The additional inclusion of a bioluminescent marker or superparamagnetic iron oxide (SPIO) tag would also allow for in vivo monitoring in live animals.
A different tracking method would also help to better differentiate between endogenous and graft-derived neurogenesis. In previous studies with this stroke model, we have shown that newborn cells that express doublecortin (DCX) and incorporate BrdU migrate from the SVZ to the stroke region and form new neurons. Other interventions, such as whisker stimulation, can increase the number of neuroblasts migrating toward the stroke region at early time points and new neurons in the penumbra region 4 weeks after stroke . This response could be enhanced by the transplantation of neural precursors, but this hypothesis remains to be verified by using more specific markers and technologies.
BrdU incorporation is the most commonly used and clearest measure for tracking the fate of newborn cells in the nervous system [56, 57], as it remains in the cell even after it differentiates and leaves the cell cycle, but there are concerns with its use. Because BrdU incorporates during DNA replication, it is possible that labeled cells may have been undergoing cell repair (successfully or prior to apoptosis) rather than mitosis . In unpublished studies, we have found little to no co-staining of TUNEL and BrdU in this model, and none of the counted cells showed morphological signs of apoptosis or necrosis, so it is unlikely that the BrdU-positive cells we observed were dying. It is also unlikely that our dose of 50 mg/kg would be sufficient to visualize cells undergoing repair , although an increase in successful cell repair would also be a desirable outcome of transplantation. Future studies will need to differentiate between cell repair and true neurogenesis by examining earlier time points and quantifying DCX-positive neuroblasts migrating to the stroke region after treatment.
Our previous unpublished studies have demonstrated no change in graft survival in this model when immune suppression is administered and we chose not to use it in this study. However, it is possible that we would have been able to achieve greater or more consistent levels of neuronal differentiation in vivo if immune suppression had been used. In a transplant model using mouse embryonic stem-cell derived neurospheres, graft survival was unchanged by the administration of cyclosporine A, but inflammatory factors biased cells toward glial differentiation rather than neuronal differentiation when it was not given . This may have occurred in our model as well, although we likely mitigated this effect by transplanting 7 days after stroke, when inflammation in the stroke region has largely subsided. It is also important to note that immune suppression may be detrimental to healing after stroke. Inflammatory signals can attract stem cells to the site of injury , and the immune response may be neuroprotective and necessary for endogenous neurogenesis [62–65]. Additionally, the immune system is already naturally suppressed after stroke , and further suppression may increase the risks of infection and tumor formation [67, 68]. It is thus clear that systemic immune suppression in patients with stroke should be avoided wherever possible.