In the past few years, stem cell-like tumor precursors have been identified in gliomas. They have been consecutively termed glioma stem cells, brain tumor stem cells or brain tumor initiating cells. They are characterized by self-renewal, limitless proliferation, tumor initiation, multi-differentiation and expression of stem cell surface markers such as CD133 and nestin. However, long-term stable maintenance of GSCs, which will offer much more convenient opportunities for attaining full and accurate understanding of the biological features of this special tumor cell type, has been achieved by only a few groups and does not suffice to meet research requirements [18
]. No pure CD133+ glioma stem cell line has so far been available; proliferation and differentiation of these tumor stem cells in vitro
cannot be stopped completely even in a culture medium favoring stem cell growth. The percentage of CD133+ cells in such lines has varied widely. Accordingly, there are no unanimously agreed criteria for establishing a GSC line. Successful cell lines from other tumors suggest that establishment of a GSC line should meet the following criteria. First, GSCs can be cultured long-term in vitro
while maintaining relatively stable stem cell properties. Secondly, even after long-term maintenance, the GSCs should recapitulate their parent or original tumor. In the current study, SU-1 and SU-2, respectively originating from primary and recurrent gliomas with malignancy progression in the same patient, have been maintained in vitro
for more than three years while retaining their tumor stem cell properties. Though the percentage of CD133+ cells was not high (less than 10%), nestin+ cells were the dominant subgroup (> 90%). Thus, the two newly established cell lines SU-1 and SU-2 could be regarded as glioma stem/progenitor lines. Cryopreservation and resuscitation were successful during long-term serial passages in vitro
. We also noticed differences in configuration between the tumor spheres derived from SU-1 and SU-2. When cultured in defined stem cell growth medium (FBS free), the SU-1 spheres were more compact than those of SU-2, and the percentage of CD133+ cells was lower. When cultured in serum-based medium, SU-2 seemed more resistant to FBS-induced differentiation and remained more morphologically primitive than SU-1. In vivo, direct orthotopic transplantation of SU-1 and SU-2 cells developed into xenografts in immune-deficient mouse cerebrum, but the tumors derived from SU-2 cells were more aggressive than those from SU-1. These data imply that malignancy progression could also occur in tumor stem cells. Taken together, these results suggest that SU-1 and SU-2 could provisionally be regarded as permanent glioma stem/progrnitor cell lines and further utilized as reliable resources for basic research and clinical trials concerning GSCs.
The study of GSCs is actually an extension of that of NSCs, since not only the concepts but also the methods employed are derived from those used for NSCs. The finding that 102
CD133+ tumor cells could produce tumor mass in NOD-SCID mice, while up to 105
CD133- tumor cells could not, proved that the former were brain tumor initiating cells and the latter were not [8
]. So it seemed reasonable to suppose that CD133+ tumor stem cells could proliferate and differentiate into CD133- cells, which could further differentiate into common tumor cells approaching terminal differentiation, as NSCs do. However, Beier's studies revealed that four of 15 cell lines derived from primary glioblastomas grew adherently in vitro
and were driven by CD133- tumor cells that fulfilled stem cell criteria. Both CD133+ and CD133- subtypes of GSCs were similarly tumorigenic in nude mice in vivo [20
], indicating that CD133 expression is not sufficient to identify GSCs; more effort is needed to identify a specific GSC marker. At present, though this functional criterion for GSCs is sophisticated and inconvenient to apply, it is reliable and should not be neglected unless and until a specific marker for GSCs is found.
GSCs do not differentiate terminally under conditions that would induce terminal differentiation in NSCs. Not only was differentiation retarded, but retro-differentiation was also observed in vitro
. Our data showed that soon after treatment with differentiation-inducing agents such as FBS and valproate (VPA), nonadherent tumor spheres dissociated and scattered into adherent spindle-shaped monolayer cells. Most of these were still highly positive for nestin (a marker for neural stem/progenitor cells), while a few cells appeared that were doubly positive for nestin and either GFAP (marker for astrocytes) or β-tubulin III (marker for neurons). Markers of both mature and stem/progenitor cells are very rarely co-expressed during NSC differentiation, but it is common in GSCs [15
]. We also observed a "down-up" trend in the percentages of CD133+ cells in SU-1 and SU-2 during a relatively long differentiation-inducing process in vitro
; that is, the percentage of CD133+ cells decreased at first, then remained low for a time and finally increased a little, suggesting that partially differentiated CD133+ cells (loss of CD133 expression) retro-differentiated into CD133+ GSCs under certain circumstances, which made the GSCs involved in tumor remodeling more sophisticated. There was a concomitant "up-down" trend in the levels of the neural differentiation markers GFAP and β-tubulin-III. These phenomena were more obvious in SU-2 [15
]. Thus, it is easy to infer that GSCs were generally similar to NSCs but showed important differences. Under conditions in which differentiation would be induced in NSCs, GSCs showed an intrinsic potential to maintain their undifferentiated state or to resist differentiation and even tended to retro-differentiate under certain circumstances. Once differentiation was initiated in NSCs, they were transformed step by step into various kinds of mature neural cells.
Amplification of the oncogene EGFR
and deletion of the tumor suppressor PTEN
have been identified as the critical genetic changes in the tumorigenesis of human GBMs or other types of glioma. However, few existing glioma cell lines harbor these genetic abnormalities [21
]. The fact that GSCs of both the SU-1 and SU-2 lines faithfully preserved the EGFR
amplification and PTEN
loss greatly enhances their utility in biological and preclinical studies of human gliomas. Recent studies have shown a close correlation between PTEN
loss and low autophagic activiy [33
]. We also found that PTEN
loss and absence of autophagy were concurrent in both SU-1 and SU-2, and this may suggest potential targets for future molecular intervention. More intriguingly, we discovered amplification of MTA1
in SU-2 but not in SU-1. MTA1
is closely associated with various malignancies and its up-regulation always indicates tumor recurrence and metastasis [34
]. However, the significance of MTA1
in the malignancy progression of gliomas has rarely been considered. In the current study, the particular amplification of MTA1
in GSCs derived from the recurrent tumor makes it reasonable to conjecture that MTA1
activation may contribute to both the aggression of GSCs and the malignancy progression of gliomas.