Human embryonic stem cells can give rise to all cell types in the body and are a valuable tool to make differentiated cells for drug screening, predictive toxicology and potentially therapeutic applications. Therapeutic use of stem cell derived cell populations would be augmented if they did not induce an immune response in the host that would result in rejection of cells. Although preliminary data suggest that cell transplant populations differentiated from hESCs in vitro do not express an immunogenic HLA profile 
, the long-term immune effects are unknown. It is most likely that an autologous or immune-matched transplant would be best tolerated by the recipient. phESCs are pluripotent stem cells that can be made HLA homozygous for simplified immune matching, but their differentiation potential and biases are understudied. A greater understanding of the differentiation potential and differentiation bias of multiple cell types is needed to allow researchers to select the cell type that best suits the research or clinical need at hand. In this study we identify points of equivalence and non-equivalence between hESC and phESC.
Our data indicate that the in vitro directed NPC differentiation protocol resulted in poor yield from phESC compared to hESC, most likely due to a difference in cell-cell interaction resulting from differences in ECM expression. We corrected for the possibility that decreased proliferation in the undifferentiated state led to an apparent difference in final yield by maintaining undifferentiated cultures as long as was necessary to reach comparable confluence. The NPC protocol included a culture step consisting of neurosphere formation, which relied heavily on cell-cell interaction. Our PCR data revealed that phESCs and hESCs had a different expression pattern of extracellular matrix proteins, suggesting that while the neurosphere step may be ideal for the hESC cultures, it may be detrimental to the phESC cultures. Indeed, we observed a lack of adherence between phESC derivates, resulting in poor sphere forming capability, and poor adherence to surfaces. At day 6 through day 14 of the differentiation protocol, hESCs typically formed round, relatively stable spheres that held together during feeding of cells in suspension. phESCs had a tendency to dissociate and form smaller spheres or dissociate to the single-cell level. Since this feeding selects for the larger, uniform hESC-type spheres, it is not surprising that a problem at this stage would lead to dramatically decreased yield.
The possibility that phESCs cannot efficiently produce cells of the neural lineage is ruled out by the successful generation of RPE, which are of the neural lineage. The RPE yield and purity was equivalent in the phESC and hESC cultures. Notably, the RPE differentiation protocols did not rely on a sphere-formation step during culture, suggesting that the apparent inequality in differentiation potential demonstrated in the NPC differentiation study may have resulted from differences in ECM proteins and was thus an artifact of the protocol.
Our PCR data support the idea that the altered ECM profile of phESC derivates may contribute to their poor spherogenic nature. Multiple adhesion molecules and their regulators are expressed at different levels between hESC and phESC derivates. Relevant to successful generation of RPE from phESC, Col7A1, a collagen present in several layers of the retina 
, was elevated in phESC derivates. hESC derivates express elevated levels of four MMPs, indicating expression of these MMPs does not preclude sphere formation at this stage, possibly because co-expression of their regulators or targets ameliorate their anti-adhesive effects. For example, MMP13 (elevated in hESC derivates) preferentially cleaves type II collagen 
, which is also elevated in hESC as Col11A1. In addition, hESC derivates express elevated TIMP2, an inhibitor of MMPs 
. hESC derivates express 26.42 fold higher levels of NCAM1 than their phESC counterparts. Decreased NCAM1 expression may contribute to loss of cells, both by direct dissociation leading to cell loss during aspiration of consumed media, and by loss of trophic support. NCAM1 activates the FGF receptor when clustered NCAM1 is involved in trans-homophilic binding 
. FGF provides not only differentiation cues, but also trophic support 
, without which the phESC in our NPC protocol may have been at an additional disadvantage.
For therapeutic use, there are a number of practical considerations when choosing the appropriate cell type. Should their differentiation methods be optimized, parthenotes offer tremendous advantages for immune tolerance. Similar advantages could be realized with hESCs and iPS, however; there are drawbacks to using these two cell types that can be avoided with phESCs. hESCs express no human leukocyte antigen (HLA)-DR, DP, DQ and only low levels of HLA-A, B, C even when exposed to pro-inflammatory cytokines 
, but it is not yet clear how differentiated cells would be tolerated long-term. iPS technology suffers from increased tumorigenesis, low efficiency, and the time it takes to generate an immune-matched line would render it inadequate for acute needs. An additional commercial strength of phESCs is that their increased immune tolerance may broaden the range of patients who stand to benefit from each line. With rising health care costs, efficient production may be essential to opening the possibility of therapy to the greatest number of patients. A system in which a new line must be created for each patient may not be economically feasible, and will be technically inferior. phESCs can be immune matched for simplified long-term tolerance and can be banked and stored for efficient production and timely availability. When choosing a line for commercial scale up, cellular yield is an important measure to consider. The cost of expansion and differentiation can be immense, so addition of financial and temporal burdens caused by decreased yield is not desirable. These findings and needs underscore the importance of determining comparative differentiation biases and differentiation potential.
In conclusion, our data indicate that stem cell differentiation protocols are specific to stem cell types, and that phESCs have a unique ECM and methylation profile which renders differentiation protocols containing a spherogenic step unsuitable for this stem cell population. Thus, inappropriate differentiation protocols mask differentiation potential.