Stem cells and the Heisenberg uncertainty principle

Comment on Wagner et al, page 675

Considerable efforts have been directed at understanding the distinctive genetic programs that underlie the biology of the elusive hematopoietic stem cell.

In this issue of Blood, Wagner and colleagues publish another gene expression profile of human hematopoietic stem cells (HSCs), using the novel strategy of isolating the slow-dividing fraction (SDF) of CD34⁺CD38^– cells. This SDF is determined by its failure to divide after a week in culture, as measured by the retention of the membrane dye PKH26. The rationale for this study is that functional HSCs are largely quiescent, a view supported by a sizeable body of prior evidence.¹  Although this study is technically rigorous, an obvious concern is that the isolation procedures and the time in culture alter the transcriptome of these cells, even though they have not divided. As Theise²  has recently pointed out, all the various manipulations used by different investigators to isolate HSCs (fluorescence-activated cell sorting [FACS], magnetic beads, dye exclusion, and dye retention) may perturb the biology of stem cells that have been ripped from their native environment, thereby, to paraphrase Heisenberg, disturbing the very phenomenon under study.

Another concern with this and all other published HSC gene expression profiles is comparability between studies. Considering that the isolation strategies vary considerably from study to study, as do the original sources of cells, the analytic platforms used, and even the statistical methods applied, it is probably not surprising that the degree of overlap between surveys is limited (Wagner and colleagues [Table 6]). These results suggest that different investigators may in fact be looking at distinct cells or mixtures of cells and introduce uncertainty as to what the transcriptome of the “real” HSC is. Indeed, the comparability of the profiles of CD34⁺CD38^– cells and SDF cells within this study is low, with only one gene, osteopontin, showing high levels of differential expression in both sets of cells (Wagner and colleagues [Tables 1 and 3]).FIG1 

Putting these significant technical concerns aside, these authors are to be commended for using a functional attribute of stem cells, proliferative quiescence, as part of their enrichment strategy. A more distinctive functional attribute of stem cells is their capacity for asymmetric division, leaving one daughter stem cell and sending the other sibling down the differentiation pathway. These authors' own previous work demonstrated that approximately 30% of CD34⁺CD38^– cells appear to divide asymmetrically, giving rise to one quiescent and one proliferating daughter cell. It seems likely that this property of asymmetric division has an anatomic basis, as has been elegantly demonstrated in Drosophila germ cells.³  In this model, hematopoietic stem cells are applied closely to supporting marrow stromal cells (perhaps the osteoblast)⁴  in a tight geometric arrangement that forces one daughter cell away from the stromal layer after cell division. It is the distinct genetic program of that HSC, poised at the moment of an asymmetric division, that we need to understand. Since that genetic program must be regulated epigenetically by intimate interactions between stem cells and their microenvironment, it would be desirable to find ways to perform that analysis on stem cells in their niche.