Cell surface markers CD133 and CD15 have been recognized as neural stem cell markers and have been applied to enrich for neural precursor cells from various sources 
. However, to our knowledge, purified CD133 or CD15 expressing neural precursor cells have not been maintained as homogeneous cell populations in vitro
. In neurosphere cultures this is obscured because neural precursor cells spontaneously differentiate into glial lineages 
. By analyzing purified human NS cell lines 
, we find CD133 and CD15 are only expressed by a subpopulation of neural stem cells. We demonstrated that single CD133 and/or CD15 negative cells could generate clonal and tripotent neural stem cell lines. The two cell surface markers were invariably expressed heterogeneously in proliferating cultures. Therefore cultured neural stem cells are not constitutively CD133 or CD15 positive. This could be related to the observation that CD133 expression appears linked to cell cycle phase in NS cells. CD133 negative cells are disproportionately represented in G0/G1. Recently, cell cycle dependent variation in CD133 expression has been described for colon cancer and melanoma cell lines 
. These authors also find that CD133 levels are down-regulated in G0-G1, suggesting that this may be a generic feature.
In our adherent culture conditions CD133 negative human NS cells exhibited higher colony formation efficiency than CD133 positive cells. Previous analyses using primary human foetal CNS tissue have indicated that few or even no neurospheres could be derived from CD133 negative cell populations 
. However, primary tissues contain overwhelming numbers of differentiated cells that are CD133 negative. Therefore CD133 sorting will give a major enrichment even if only capturing a fraction of the stem cells. It is also possible that CD133 negative stem cells may not initiate colony formation in suspension culture, even though they do so efficiently when adherent. The higher cloning efficiency we observed for CD133 negative versus CD133 positive cells might be because G0/G1 cells are intrinsically more clonogenic, or they may be less fragile and more resistant to flow sorting compared cells in S, G2, or M phase. Alternatively they may have some other feature that resists stress-induced differentiation or promotes attachment.
Interestingly, although human and mouse NS cells are cultured under identical conditions, human NS cell cultures harbour fewer CD133+ cells and exhibit a longer doubling time than mouse counterparts. Ki67 staining and BrdU incorporation experiments indicated that a subpopulation of human NS cells may withdraw from the cell cycle. The fraction of non-cycling human NS cells, estimated from BrdU incorporation () is around 5% whereas this is less than 0.2% in mouse NS cell cultures. These non-cycling cells do not show features of differentiation or senescence, and retain precursor markers suggesting they could be dormant stem cells. To investigate further we applied Ara-c, an antimitotic drug, to eliminate dividing cells. Approximately 5% of human NS cells remained viable. These cells retained Nestin and Sox2 expression but did not express CD133 or Ki67. Parallel BrdU labelling confirmed they were not dividing. Crucially however, a proportion of these cells resumed proliferation after the antimitotic drug was removed. This strongly suggests that human NS cells in culture can suspend proliferation but retain the capacity to re-enter the cell cycle.
A major interest in neural stem cell biology is the relationship with brain tumor cells [42,43 
]. CD133 has been successfully applied to isolate brain tumor initiating cells, also called cancer stem cells 
. It was found that malignant brain tumors have a higher CD133 index than low-grade tumors 
. Based on the present observations, however, this does not necessarily indicate an increased frequency of stem cells but may reflect a higher proliferative index in the stem cell compartment. Indeed we have found that glioblastoma stem cells also exhibit heterogeneous CD133 and CD15 expression in vitro 
, similar to the profile observed in human NS cells. If CD133 expression in brain cancer stem cells is regulated in a cell cycle dependent fashion, extra caution must be taken when this marker is used to define tumour stem cells. This would be particularly significant if CD133 is not expressed by quiescent brain cancer stem cells.
In conclusion, our data demonstrate that mammalian neural stem cells are not constitutively CD133 or CD15 positive, and that down-regulation of CD133 protein and mRNA correlates with an enrichment of cells in G0-G1 phase of the cell cycle. These observations point to the potential absence of CD133 expression in slow-cycling or dormant neural stem cells. It will be informative to investigate whether CD133 expression is similarly down-regulated in G0/G1 in adult neural stem cells in vivo. The mechanisms and regulation of NS cell dormancy are also of interest for future study.