We discovered a set of cell surface signatures that enable the isolation of NSC, neurons and glia derived from neural induction cultures of pluripotent stem cells. illustrates a schematic of our cell sorting methods. An important advantage of our approach is that it enables more consistent NSC preparations by allowing for the isolation of NSC without manual dissection of rosettes. In addition, the sorted glial cells provide a population of replicating astrocytes that can be subsequently cultured in vitro. Most importantly, our approach allows for the isolation of near-pure populations of terminally differentiated neurons that are viable in culture. In addition to cell sorting, these cell surface signatures can also be used to comparatively analyze cell populations. In addition, to the hits we further analyzed, there are many additional markers from our data set to explore that could potentially identify unique subpopulations of NSC, neurons and glia.
Diagram of stages and defined markers for isolation of NSC, neurons and glia from neural induction cultures starting with pluripotent stem cells.
Our unbiased screen suggests novel uses of known markers for the isolation of neural cell types. CD184+
has been described for the isolation of neural cells capable of differentiating to neurons post-FACS 
. We observed that CD184 negative selection, not positive selection was required for isolation of post-mitotic neurons. Another study reported the enrichment of hESC-derived neurons by isolating CD56+
neurons by FACS, but purity was not described 
. Based on our screening data, CD56+
would not adequately discriminate against non-neuronal cells. Pruszak et al. (2009) performed a cell surface marker analysis on expanded and differentiated neural rosettes and reported that CD29, CD15 and CD24 defined distinct cell types within this cell population 
. They described multiple populations based on CD24 immunoreactivity, which we did not observe, possibly due to the differences in the maturation of the neural induction cultures used. It will be interesting to determine the relationship of their signatures to those described here. CD133 has proven useful for isolating NSC in multiple studies 
. We avoided CD133 as a selection marker for NSC due to its high expression in hESC and dim expression in EB-rosette(+) cultures in our screens, which raises the risk of insufficient purification away from pluripotent stem cells. In addition to neural ectoderm, CD184 and CD15 have been shown to mark and isolate progenitors from other germ layers 
. Since we could not examine all possible differentiation methods, more work is needed to determine whether the signatures described here are sufficient for isolating neural cell types in additional in vitro differentiating cell culture systems. Additional negative selection markers may be necessary for discriminating other cell types from very heterogeneous cell cultures. For example, Flk-1 may be useful for excluding mesodermal cells 
. Comparison of our data set to those of subsequent immunophenotyping screens on endodermal and mesodermal cell lineages may ultimately lead to appropriate cell-surface signatures for desired cell types in any culture condition.
We believe that immunophenotyping screens could be performed in higher throughput and greater complexity. For example, it may be possible to take advantage of advances in fluorescent cell bar-coding to enable immunophenotyping of multiple cell cultures simultaneously in one experiment 
. We were successful in correlating neuronal β-III tubulin staining with cell surface markers by image analysis. It may be possible to combine cell-surface immunophenotyping with intracellular FACS analysis to identify cell-surface signatures of specific cell types and subpopulations therein. This type of analysis could be instrumental in defining cell surface marker signatures for different developmental stages in neural induction and regional specification, as well as for specific cell subtypes such as dopaminergic neural progenitors and neurons.
Our grafting studies of 3-week differentiation cultures of CD184+
HUES-9 NSC revealed a time course of neuronal and glial maturation and a clear population of grafted DCX-positive neurons or GFAP-positive astrocytes identified 2-4 weeks after grafting, consistent with our previous studies 
. No hNSE or human specific synaptophysin (hSYN) expression was seen at 4 weeks. In contrast, intense hNSE and hSYN immunoreactivity was seen 8–10 weeks after grafting of proliferating NSC, similar to the neuronal maturation profile of proliferating human fetal spinal neural precursors grafted spinally in the same spinal-ischemia model or in naïve immunosuppressed minipig in earlier studies 
. In addition, a subpopulation of grafted proliferating NSC acquired oligodendrocyte phenotype as evidenced by expression of NG2 and Olig2. In contrast to long-term survival and maturation of grafted differentiated or proliferating NSC, no survival of CD184−
purified neurons was seen at 2 or 4 weeks after grafting. These data suggest that transplantation of proliferating CD184+
NPC is sufficient for long-term neuronal engraftment without tumor formation.
In transgenic SOD1 mutant animals grafted with CD184+/CD44+ glia, no DCX-immunoreactive neurons were identified and the majority of grafted cells expressed GFAP, similar to our long term in vitro culture. More extensive studies need to be performed to determine whether CD184+/CD44+ sorted glia have potential to differentiate to oligodendrocytes, but at this time they appear to be highly enriched for cells that can give rise to astrocytes.
While we observed cells that expressed oligodendrocyte markers 10 weeks after grafting, we did not observe convincing staining with these markers our in vitro cultures. This difference could be niche and/or time dependent. Others have reported spurious oligodendrocyte differentiation of hESC-derived NSC in vitro 
. We suspect that significantly longer culture times and/or different differentiation conditions may be required to achieve mature oligodendrocyte differentiation in vitro 
. It is also possible that the CD184+
NSC subpopulation is more biased to neuronal and astrocyte differentiation.
Our approach to defining cell surface signatures in differentiating cultures of pluripotent human stem cells will potentially enable quantitative benchmarks to be devised for comparing specific cell populations generated by different hESC and hiPSC lines. In addition, further antibody screens may yield cell surface signatures of specific neuronal and neural progenitor subtypes as well as signatures of desired cell types from other germ layers. For applications such as transplantation, our approach will be very useful for obtaining pure, well-defined cell populations. Finally, these methods are applicable to many studies of the characteristics of defined neuronal populations, expression analyses, generating defined mixtures of cell types to test disease models for non-cell-autonomous phenotypes, and for drug testing and development.