To the editor

In the February 2002 issue of Nature Medicine, Morshead et al.1 report the inability to generate hematopoietic progeny from neural stem cells (NSCs), raising several issues that must be addressed to avoid misinterpretation. Specifically, although they were purported to be identical, a number of substantial deviations exist between this study and our experiments showing the neurohematopoietic potential of NSCs (ref. 2). The functional characteristics of the NSC cultures described by Morshead et al. show that the cells that they used are unlike any known NSC, particularly those used in our transdifferentiation experiments2. These authors' cultures become consistently transformed, a property that we and others have never observed. Neither human nor mouse NSCs undergo transformation with passaging, but rather exhibit growth factor dependency, unaltered growth kinetics and prompt differentiation upon growth factor withdrawal over time3,4,5,6. Furthermore, Morshead et al. report that “less than 1% of the cells composing an individual neurosphere are NSCs,” which is far below the current standard in this system: from 8% (postnatal) to over 20% (embryonic)3,4,5,6. Thus, Morshead et al. injected at least 20 times fewer NSCs than in our study, using populations of NSCs that displayed altered growth and differentiation capacity. Moreover, as noted by these authors, “transformed, aggressively growing cells would progressively eliminate non-transformed cells,” further exacerbating their NSC deficiency. Eventually, the combination of low NSC number and significant transformation found in the Morshead et al. cultures would lead to the transplantation of a negligible number of NSCs. These authors' use of a high sensitivity method to detect hematopoietic engraftment cannot possibly compensate for such severe deficiency, particularly since the kinetic parameters that apply to engraftment in standard repopulation experiments with blood stem cells cannot be extrapolated to the different phenomenon that is the object of these studies, namely the expression of an hematopoietic fate by NSCs.

So far, neuro–hematopoietic conversion has been reported by three independent groups2,7,8. In particular, Shih et al. confirmed the work by Bjornson et al. by using human NSCs (ref. 8). Although Morshead et al. suggest these results were due to hematopoietic contamination, closer examination of the methods employed by Shih et al. rules out this possibility. Furthermore, additional studies have reported that NSCs transdifferentiate into non-hematopoietic mesodermal derivatives9. Thus, the ability of NSCs to give rise to non- neural cells should be less of a question, particularly in light of the variety of freshly-isolated and cultured somatic stem-cell types which appear capable of transdifferentiation10.

We suggest that a more parsimonious conclusion should be drawn from the study of Morshead et al., and submit that transdifferentiation experiments using transformed/transforming cultures with a negligible content of NSCs would necessarily yield an experimental outcome different than one using normal, highly clonogenic NSC cultures. The failure to detect transdifferentiation even at the single cell level is therefore not surprising. While we concur that hematopoietic transdifferentiation may represent a rare property of NSCs, we suggest that owing to disparate culture conditions and an unexplained NSC deficiency in the Morshead et al. experiments, it is inappropriate to compare this study to that of Bjornson et al.

See Reply to “Hematopoietic potential of neural stem cells” by Morshead et al.