Targeting Glioma Stem Cells

by Stephen Western

Ever since cells with stem cell-like properties were first identified in human brain tumour tissue in 2003 (1), glioma stem cells, or glioma tumour-initiating cells have been a hot topic in glioma research, though their exact nature and role are not yet clear. Cancer stem cells are defined as having potential for limitless replication, self-renewal, multilineage differentiation, and tumorigenicity. It is hypothesized that cancer stem cells are more resistant to therapies and are primarily responsible for tumour recurrence.

According to some models, the cancer stem cell is a distinct population which gives rise to both stem and non-stem cells. Another model, the "stemness phenotype" model, argues that "stemness" is a phenotype, varying according to microenvironmental factors, and that all cells in a tumour have stemness potential (2). According to this model, there should be intermediate phenotypes, somewhere between a pure stem-like cell and a non-stem cell.

CD133 is a cell-membrane protein and is perhaps the most commonly used marker to identify the stem cell population in gliomas. The stemness phenotype model is supported in a recent study of CD133 expression, where CD133 expression in the population increased or declined according to whether the cells were growing under low oxygen or normoxic conditions. The authors of this study perceptively observe:

"Induction of two opposing phenotypes in cells of identical genetic background demonstrates the plasticity, multipotency and opportunistic properties of these tumors. It suggests that malignant glioma cells are capable of adapting to and remodeling their microenvironment, switching phenotypic behaviour to improve their chances for survival." (2)

It is likely that the stem cell-like properties in a fraction of cells found in any tumour is an adaptation to microenvironmental factors such as hypoxia.

Radioresistant and radiosensitive subpopulations in glioma stem cell cultures

August 2, 2014 update.

The consensus opinion formed during the initial years of glioma stem cell research states that glioma stem cells (tumour-initiating cells) display increased radio- and chemo-resistance compared to non-stem glioma cells, and are thus largely responsible for tumour recurrence. More recent research calls this view into question, and reveals that a more complicated scenario exists in reality.

A group centred in Heidelberg, Germany (Lemke, Platten, Wick et al.) published a paper in June, 2014 which significantly advances our understanding of glioma stem cell markers and their relation to radioresistance and proliferation (9). In this study, freshly dissected tumour tissue from multiple glioblastoma patients were cultured in neurosphere medium, forming glioma-initiating cell (GIC) cultures.

Next, established glioma stem cell markers, such as CD133, CD15, CD44, Nestin and SOX2, were examined and compared with the sensitivity of each cell line to radiotherapy (4 Gy). Surprisingly, the stem cell markers Nestin and SOX2 inversely correlated with radioresistance in the clonogenic assay. In other words GIC cultures with higher expression of these markers tended to have less radioresistance/ increased radiosensitivity. Another stem cell marker, CD44, was positively correlated with radioresistance. Further analysis of gene transcripts (mRNA) detected in the various cell cultures showed that common glioma stem cell markers CD133 and Nestin (as well as Musashi1, PLAGL2 and L1CAM) inversely correlated with radioresistance in the clonogenic assay, contrary to expectations based on currently prevailing theories.

A similar analysis was carried out, this time using a proliferation assay. Again, in the glioma-initiating cell cultures, levels of Nestin protein, and mRNA levels of CD133, PLAGL2 and Musashi1 inversely correlated with radioresistance following the proliferation assay.

Following these in vitro experiments, two GIC cultures, representing one radiosensitive and one radioresistant culture, were implanted into mouse brains. Interestingly, the mice implanted with radiosensitive tumours became symptomatic after 71 days, while mice implanted with radioresistant tumours became symptomatic after around 192 days, implying that the radioresistant cell culture was less proliferative/ tumorigenic. As predicted, when the mice were given 6 Gy of cranial irradiation, the survival of the mice with radiosensitive xenografts was prolonged, while survival of mice with radioresistant xenografts was not significantly prolonged.

The Cancer Genome Atlas (TCGA) database was then investigated in the attempt to correlate expression of glioma stem cell markers with disease-free patient survival following radiochemotherapy. No correlation was found.

SOX2 positivity: a proliferative, radiosensitive subpopulation

In the GIC cultures described previously, intracellular markers were next investigated in order to define subpopulations within each GIC culture which might have different sensitivity to radiotherapy. All the GIC cultures were nearly 100% positive for nestin, ruling out a separate subpopulation of nestin-positive cells. On the other hand, GIC cultures were found to have separate subpopulations of SOX2-negative and SOX2 positive cells. Interestingly, the radiosensitive cell line T269 was nearly 100% positive for SOX2, while the radioresistant T325 displayed the lowest fraction of SOX2-positive cells. Further investigations revealed that SOX2 positive cells were more proliferative than SOX2-negative cells. 95-98% of the proliferating cells in the two GIC cell lines were SOX2 positive. Actively cycling cells positive for Ki-67 were almost exclusively SOX2-positive. The SOX2-positive fraction of 5 GIC cell lines were found to be sensitive to irradiation with 4 and 8 Gy.

In summary, this paper calls into question the importance of conventional cell-surface markers such as CD133 in terms of glioma stem cell resistance to radiotherapy. On the contrary, the only stem-cell marker which showed a positive correlation with radioresistance was CD44, pointing to the importance of this marker as a therapeutic target. Another intracellular marker of glioma stem cells, SOX2, was found to make up the proliferative fraction of glioma-initiating cell cultures and this fraction was also sensitive to irradiation. Perhaps a dual targeting of CD44 and SOX2 would therefore be highly effective, dealing with both the radioresistant fraction and the radiosensitive but highly proliferative fraction of glioma-initiating cell cultures derived from freshly dissected glioblastomas.

Lack of correletion between glioma stem cell phenotype and resistance to therapies

Another recent study (10) by researchers at University of California, San Francisco (Fouse, Costello et al.) also called into question the hypothesis of inherent therapeutic resistance of glioma stem cells to radio- and chemo-therapy. Matched cancer stem cell (CSC)- enriched and CSC-depleted cultures from 10 primary glioblastoma patients were developed. The CSC-enriched cultures were highly positive for recognized stemness markers such as nestin, SOX2 and CD133, while the CSC-depleted cultures from the patient samples were mostly negative for these markers. When injected into mouse brains, the CSC-enriched cultures were highly tumorigenic and formed aggressive, diffusely invasive tumours, while the CSC-depleted cultures rarely formed tumours (1 out of 4), and the resulting tumour was circumscribed and less invasive. Thus CSC-enriched and CSC-depleted cultures from the same patient were phenotypically (outwardly) distinct, in terms of behaviour and expression of recognized stemness markers.

25 primary glioblastoma CSC cultures and 10 matched CSC-depleted cultures were then tested for resistance to temozolomide (TMZ). The 25 CSC-enriched cultures showed a wide range of sensitivity to TMZ. Contrary to expectations, the CSC-depleted cultures displayed a similar degree of sensitivity to TMZ when compared to their matched (from the same patient) CSC-enriched counterparts, introducing the concept that resistance/sensitivity to therapy may be more dependent on the individual tumour characteristics than on the differences between glioma stem- and non-stem cells. In support of this concept, each CSC and non-CSC culture pair displayed similar MGMT methylation status and MGMT expression along with similar TMZ sensitivity. When TMZ alone, RT (radiotherapy) alone, and combined TMZ plus RT were tested in the matched samples, there was again no significant difference in response between the CSC-enriched and CSC-depleted cultures, with one exception, in which the CSC-enriched culture was more sensitive to irradiation than the matched CSC-depleted culture. When 10 matched samples were tested in the context of 3-day exposure to irradiation or RT plus TMZ, there was again no significant difference in response between the matched CSC-depleted and CSC-enriched pairs. Contrary to expectations, the CSC-enriched cultures from 3 patients were more sensitive to 3-day TMZ or RT treatment than the matched CSC-depleted cultures.

This study also found that 10 individual patient-derived cell cultures were sensitive to either RT or to TMZ, while the combination of RT plus TMZ did not significantly improve therapeutic efficacy. This finding could have important implications in terms of individualized glioblastoma therapy. This study also failed to find any relation between expression of CD133 and response to TMZ or RT, further calling into question the prevailing view that glioma stem cells are inherently more resistant to therapies than non-stem glioma cells. In two cases, treatment with either TMZ or RT led to a slight increase in the percentage of CD133-positive cells post-therapy, though this was more likely related to increased CD133 expression in formerly non-expressing cells, rather than a selective outgrowth of therapy-resistant CD133-positive cells. In other words, CD133 expression in these cases may be a response to therapy, rather than an indication of therapeutic resistance in pre-existing CD133-positive cells.

In summary, this study challenges the prevailing view that the glioma stem cell fraction within a tumour is inherently more resistant to therapies than the non-stem fraction. In some cases, it was found that the CSC-enriched fraction was in fact more sensitive to therapy. Importantly, CD133 was not found to correlate to therapy resistance. Instead, resistance/sensitivity to TMZ and RT were found to be patient-specific. The authors intend to extend these studies to the in vivo setting.

Targeting glioma stem-like cells

Invasive cells are shown to have higher activity of STAT3 and NF-KB transcription factors. Recent studies have shown that both of these transcription factors are also constitutively activated in glioblastoma stem cells and that they influence the Notch pathway, which is also deregulated in cancer stem cells. STAT3 was also shown to be necessary for proliferation and maintenance of multipotency in glioblastoma stem cells (5). Please see the discussion of these two transcription factors in the Targeting Invasion section.

One potential complication of targeting the glioma stem-like cells, is the inadvertent toxicity to normal neural stem cells. One study showed that the standard of care drug Temodar was more harmful to neural stem cells than it was to glioma stem cells, which had increased drug resistance (6). On the other hand, this study showed that glioma stem cells were far more vulnerable to bortezomib (trade name Velcade, a proteasome inhibitor approved for use in multiple myeloma and mantle cell lymphoma) than neural stem cells. While bortezomib has limited access to the central nervous system, the anti-alcoholic drug disulfiram (Antabuse) is also a proteasome and NF-KB inhibitor (7), and has justifiably been recommended as a promising drug in glioma therapy.

Disulfiram (Antabuse)

The drug disulfiram (Antabuse) has been in clinical use since the 1940s, mainly to discourage alcohol abuse, as consuming alcohol while on this drug leads to some very unpleasant side-effects. It is currently in two clinical trials for glioma therapy due to a multitude of anti-cancer and anti-glioma actions, including aldehyde dehydrogenase inhibition, proteasome (and consequently NF-KB) inhibition, MGMT and P-glycoprotein inhibition, polo-like kinase inhibition, and inhibition of the two matrix metalloproteinases, MMP-2 and MMP-9, which are critical for cell invasion. It must be noted however, that most of the favorable evidence in terms of glioma therapy thus far comes from in vitro laboratory studies.

In vitro, disulfiram is one of the most effective agents for targeting glioma stem cells. In a high-throughput screen of 2000 compounds being tested against glioblastoma stem cells, disulfiram emerged as the most promising candidate for further testing (7). In this study, its effects were attributed to its inhibition of the ubiquitin-proteasome pathway, as mentioned above.

In the same study mebendazole, a drug used against parasitic worms (trade name Vermox in Canada) was also shown to be one of the more effective drugs against glioma stem cells. This drug was shown to be effective in high quality animal models of glioma, and is currently being tested in a phase I clinical trial for newly diagnosed high grade gliomas (estimated primary completion date July 2014).

In a separate study, low concentrations of disulfiram were highly toxic to glioblastoma cells, while not affecting normal human astrocytes. Furthermore, disulfiram was highly effective in cell lines which were unaffected by temozolomide, and in one test the two drugs were synergistic, inhibiting proliferation and self-renewal of glioma cells by 50% at concentrations which had no effect as single agents. In this study, the efficacy of disulfiram was attributed to its action as a polo-like kinase 1 inhibitor (8).

The use of disulfiram to discourage alcoholism is due to its inhibition of aldehyde dehydrogenase, which is likewise important in glioblastoma stem cell biology. However, disulfiram suppresses glioblastoma cell growth at lower doses than are required to inhibit aldehyde dehydrogenase (8).

In short, disulfiram appears to be one of the most effective and clinically safe agents to use in targeting glioma stem cells in vitro. It has so many various potential modes of action that it is still unclear which is the most important. A phase II trial for newly diagnosed glioblastoma (primary completion date March 2016) will hopefully prove its efficacy in brain tumours.


The migratory fraction and the stem cell fraction in a glioma tumour may be to a large degree, overlapping. These two fractions, which are slowly proliferating and resistant to drug-induced apoptosis, will be the most challenging, as well as the most important targets in any successful, and potentially curative anti-glioma strategy.


  1. Identification of a cancer stem cell in human brain tumors. Singh et al. 2003.

  2. Are all glioma cells cancer stem cells? Cruz et al. 2010.

  3. Analysis of gene expression and chemoresistance of CD133+ cancer stem cells in glioblastoma. Liu et al. 2006.

  4. Maternal Embryonic Leucine Zipper Kinase (MELK) reduces replication stress in glioblastoma cells. Kig et al. 2013.

  5. STAT3 is required for proliferation and maintenance of multipotency in glioblastoma stem cells. Sherry et al. 2009.

  6. Neural stem/progenitors and glioma stem-like cells have differential sensitivity to chemotherapy. Gong et al. 2011.

  7. High-throughput chemical screens identify disulfiram as an inhibitor of human glioblastoma stem cells. Hothi et al. 2012.

  8. Disulfiram, a drug widely used to control alcoholism, suppresses the self-renewal of glioblastoma and over-rides resistance to temozolomide. Triscott et al. 2012.

  9. Primary Glioblastoma Cultures: Can Profiling of Stem Cell Markers Predict Radiotherapy Sensitivity? Lemke et al. 2014.
    READ ABSTRACT Email me for a PDF copy

  10. Response of primary glioblastoma cells to therapy is patient specific and independent of cancer stem cell phenotype. Fouse et al. 2013.
    READ ABSTRACT Email me for a PDF copy

This page was created on 12/28/2013 and last updated on 08/21/2014

Our privacy / cookie policy has changed.
Click HERE to read it!