br Discussion Generating neural stem cells from hESCs

Discussion Generating neural stem myc pathway from hESCs and iPSCs has been successfully achieved by a number of groups (Elkabetz et al., 2008; Falk et al., 2012; Koch et al., 2009; Nemati et al., 2011). However, the simplicity and retained – or even enhanced – differentiation versatility combined with economical aspects of their growth (e.g. minimal medium components and reduced hands-on maintenance) are distinct advantages of the EZ sphere method over other published protocols. EZ spheres also recover from cryostorage with high efficiency, thus facilitating experimental use. EZ spheres do not require EB formation or manual selection processes, yet they still can be efficiently patterned into a variety of neural subtypes without acquiring regional differentiation bias over time, an advantage over other published protocols (Falk et al., 2012; Koch et al., 2009; Nemati et al., 2011). Moreover, properly organized rosette structures can be generated even after long-term culture in EGF and FGF-2. Because EZ spheres are captured prior to rosette formation, they do not show rosette markers without neural induction (Fig. 4). This suggests that EZ spheres expand the earliest neural stem cell population so far identified, which can be positioned at a pre-rosette stage and before LT-hESNSCs (Koch et al., 2009), therefore bearing greater flexibility in terms of terminal differentiation. Although selecting pre-formed rosettes increases the purity of the differentiated neural cultures, this may also be limiting their differentiation potential as previous work has shown that hESC-derived neural progenitor cells are restricted by the time SOX1 is expressed in well formed rosettes (Li et al., 2005). It may also be possible that retaining non-neurally committed cell types in the spheres exposes the cells to signaling processes more consistent with in vivo development including early instruction from mesoderm (Lumsden and Krumlauf, 1996). EZ spheres express the mesodermal marker Brachyury as well as the endodermal marker alpha fetoprotein (data not shown), which may be contributing to the diverse differentiation potential. However, EZ spheres are not simply fully undifferentiated stem cells grown in suspension culture similar to those described by Steiner et al. (2010). OCT4, Nanog, SSEA3, and Tra-1-81 expression were significantly down-regulated in EZ spheres almost immediately after sphere formation (Fig. 2) indicating that EGF and FGF-2 in combination are not capable of maintaining a large population of undifferentiated stem cells in suspension. Our longitudinal examination of neural gene expression by PCR showed that EGF and FGF-2 maintain a heterogeneous population of cells within the spheres. Although central and peripheral neural progenitor genes nestin, SOX2, and HNK1 were consistently expressed, other region specific markers were differentially expressed over time. Because of this heterogeneity, it is possible that the signal for lowly expressing genes was below our level of detection. However, we interpret this dynamic nature of gene expression within EZ spheres as a contributing factor toward their diverse differentiation potential, likely by allowing cells to effectively respond to external differentiation cues even at later passages. As such, when EZ spheres are removed from EGF and FGF-2 and placed into appropriate differentiation media, the cells retain the capacity to up-regulate appropriate regionalization genes and consistently produce specified neurons and glia. Furthermore, terminal differentiation from EZ spheres shows similar efficiencies compared to standard EB protocols. Efficient neural differentiation has been described by using dual SMAD inhibition during hESC and iPSC patterning (Chambers et al., 2009). Importantly, EZ spheres are easily adaptable to this step as inhibitors to TGFβ, activin, and BMP (e.g. noggin and SB341542) can be added directly to floating EZ spheres at the start of the directed differentiation, as was done for the sensory neuron differentiation (Fig. S3). Although dual SMAD inhibition did not increase nestin expression in early passage H9 hESC-derived EZ spheres (Fig. S5), further optimization and increased yield of terminally differentiated neural subtypes may be achieved by incorporating this in the EZ sphere differentiation protocols.