Recent breakthroughs in stem cell biology, especially the development and application of induced pluripotent stem cell techniques, have generated tremendous enthusiasm and efforts to explore the therapeutic potential of stem cells in regenerative medicine. The use of stem cell-derived RPE for the treatment of AMD is under clinical investigation because there are several advantages of targeting the eye as an organ for stem cell-based therapies, for example ease of administration route, size, potential immune privilege, separation from systemic circulation and so forth [21]. However, in applications based on hESC or hiPSC, the safety of the therapeutic product is of prime importance given that residual stem cells may have the capacity of unlimited proliferation and self-renewal resulting in teratomas or teratocarcinomas that can potentially be highly malignant [2].

Current protocols for RPE generation from hESC or hiPSC rely on the differentiation process together with culture conditions; for example, use of culture medium and extracellular matrices that would not support stem cell growth to prevent the presence of residual stem cells in a differentiated RPE population. Furthermore, most cell purifications performed for clinical cell therapy utilize antibody-coupled magnetic bead-based sorting, referred to as magnetic-activated cell sorting (MACS) [22, 23], a population-based method which has a faster purification time compared with FACS. However, the method is limited by a lower efficiency of sorting and does not allow for analysis of individual cells. This is a disadvantage particularly for hESC or hiPSC applications because the benefit of single cell analysis allows a better chance for identifying and isolating stem cells that may be present in a minority.

In this report, we aimed to explore a novel strategy for ensuring purity of a RPE population by identifying and sorting on the basis of a specific cell surface marker that would be expressed on RPE but not on stem cells. We used an unbiased screening approach to identify CD59 expression on RPE but at negligible levels on stem cells and further demonstrated that sorting for CD59 expression can effectively purify RPE and deplete pluripotent stem cells from a mixed population. CD59 is involved in suppression of the complement pathway and contributes towards the potential RPE-dependent immune privilege associated with the eye [18]. CD59 identification is therefore consistent with the functional attributes of RPE. We have not formally ruled out that CD59 is expressed exclusively on RPE and not on other cell types generated during differentiation. However, because sorting for CD59 will lead to removal of stem cells, it is still a beneficial step to be included in the differentiation protocol to ensure that residual stem cells are removed effectively. Further work is also needed to demonstrate that CD59-positive cells do not have any features of pluripotency, for example with the use of teratoma and colony formation assays, embryoid body formation and alkaline phosphatase staining. It will also be interesting to explore strategies based on negative sorting for RPE purification. For instance, pluripotent stem cell specific markers such as SSEA3 or TRA-1-60 could be used to label residual stem cells and the negative fraction would potentially be free of pluripotent stem cells. However, there are technical challenges around the detection of such potentially rare events which require a high signal-to-noise ratio and a large sample number to be accurate.

It is noteworthy that the CD59 transcript can be detected in pluripotent stem cells, albeit at a level that is about 6-fold level lower than that in RPE cells. However, the expression of CD59 protein is negligible in stem cells as demonstrated by our flow cytometry data. This highlights the importance of corroborating transcript levels with protein expression because there may not necessarily be a proportional relationship. In this context, our screen using live, non-permeabilized cells allowed identification of cell surface expressed proteins that were amenable to cell sorting and separation-based applications. Further work is also needed to understand the dynamics of CD59 protein expression during the differentiation time course. This will help to clarify whether progenitors at intermediate stages of differentiation can be separated from mature RPE, in addition to undifferentiated stem cells.