Therapeutic potential of targeting glucose metabolism in glioma stem cells

Abstract

Glioblastoma is a highly lethal cancer. Glioma stem cells (GSCs) are potentially an attractive therapeutic target and eradication of GSCs may impact tumor growth and sensitize tumors to conventional therapies. The brain is one of the most metabolically active organs with glucose representing the most important, but not the only, source of energy and carbon. Like all other cancers, glioblastoma requires a continuous source of energy and molecular resources for new cell production with a preferential use of aerobic glycolysis, recognized as the Warburg effect. As selected metabolic nodes are amenable to therapeutic targeting, we observed that the Warburg effect may causally contribute to glioma heterogeneity. This Editorial summarizes recent studies that examine the relationship between GSCs and metabolism and briefly provides our views for the future directions. The ultimate goal is to establish a new concept by incorporating both the cellular hierarchical theory and the cellular evolution theory to explain tumor heterogeneity. Such concept may better elucidate the mechanisms of how tumors gain cellular and molecular complexity and guide us develop novel and effective targeted therapies.

Keywords::

• brain tumor
• cancer stem cells
• glioblastoma
• glycolysis
• tumor metabolism

1. Clinical relevance of separating glioblastoma into subtypes

Despite aggressive multimodality therapy, glioblastomas (GBMs) remain universally fatal with current treatment offering only palliation [1]. There is an urgent need to better understand GBM pathophysiology to develop effective therapies. Although it has been long recognized that cancers display aberrant metabolic reprogramming, recent studies of cancer genetics and metabolomics have just started to reveal the intrinsic crosstalk between oncogenic and oncometabolic pathways that facilitate tumor progression and therapeutic resistance [2,3]. For example, whole exome sequencing of secondary GBM by the Vogelstein and Kinzler group discovered the genetic mutations of isocitrate dehydrogenases (IDH)1 [4], which cause neomorphic enzymatic production of an oncometabolite 2-hydroxyglutarate (2-HG), resulting in global changes in DNA methylation and progression from low-grade tumor to high-grade one. However, the IDH1 mutation is uncommon in the vast majority of GBMs, suggesting that most GBMs dysregulate metabolism through alternative, most likely nongenetic, mechanisms [4]. Based on the recent genome-wide transcriptome and DNA methylome analyses, presence of the four subtypes of adult GBM has been proposed: proneural (PN), classical, neural and mesenchymal (MES) [5-7]. Given that distinct signaling pathways are activated in these individual GBM subtypes [5-7], it is reasonable to believe that therapeutic strategies may be more effective if tailored to tumor subtypes instead of applying them to all GBM patients. Nonetheless, metabolic pathways in individual GBM subtypes with distinct genetic mutations remain poorly characterized. Also, clinical relevance of separation of GBM tumors into the four subtypes is still debatable, as patient prognoses of these subtypes are marginally different or contradictory between studies (unlike other cancer subtypes such as medulloblastoma and breast cancer). Despite this controversy, recent data points out that glioma may gain MES identity when they advance malignancy [8]. Gene signature and DNA methylation pattern of classical GBM appear to overlap, at least partially, with those of PN or MES GBM [7]. At least in the experimental GBM models, striking differential responses to individual targeted therapies are observed between PN and MES GBM cell cultures and/or xenografted tumors [2,9]. To date, it remains unanswered whether separation of GBM into four (or more) subtypes has any clinically relevant significance, or gliomas including primary GBM arise from PN tumors, subsequently gaining MES identity during disease progression. We would speculate that the latter (PN to MES progression) takes a major, if not the only, role in GBM establishment (more details described later).

2. Clinical relevance of studying stem cell phenotype in GBM

Additional complexity in GBM heterogeneity is derived from recognition of cellular hierarchies in tumors with subpopulations of self-renewing, highly tumorigenic cells that activate stem cell pathways and give rise to the bulk of the tumor [10-12]. Potential relevance of these stem-like cancer cells (glioma initiating/propagating cells or glioma stem cells [GSCs]) is derived from a number of experimental studies including ours demonstrating roles in therapeutic resistance, invasion into normal tissues and tumor angiogenesis [10-12]. Targeting GSCs (or their dependent signals) may elucidate therapeutic windows that improve tumor-selective targeting with acceptable toxicity to somatic tissues. Collectively, intratumoral and intertumoral heterogeneity may be integrated into models of tumors as dynamic ecologies. Indeed, although convincing evidence exists to support presence of GBM cells with stem-like properties, it is still unclear whether these tumor cells represent a defined and irreversible population or they represent a ‘state’, where tumor cells can flux in and out this state. The evolution of tumors before, during and after treatment is likely created by a combination of cellular hierarchy and cellular evolution (Figure 1). It remains undetermined, however, how cellular hierarchy and cellular evolution coexist to explain the complexity of tumor cells in single tumors.

Figure 1. Schema showing the combined concept of cellular hierarchy model and cellular evolution model. Association with glucose metabolism is also described.

GSCs are not distributed uniformly within tumors but rather mimic the organization of the bone marrow with two functional niches: perivascular and hypoxic. Microenvironment of these two niches for GSCs is quite distinct: perivascular areas in tumors are enriched with oxygen and glucose provided by tumor vessels, while hypoxic areas have very limited nutrition including glucose. In addition, GSCs in perivascular and hypoxic niches are regulated by distinct molecular mechanisms including VEGF and hypoxia-inducible factors (HIFs), respectively [13]. HIFs are critical to GSC maintenance and chronic hypoxic and/or hypo-glucose condition triggers stem cell features in cancer cells. Limited oxygen and nutrition induces stress to challenge tumor cells to survive. It remains elusive, however, whether GSCs in these two niches have different gene signatures and phenotypes. The predominant source of VEGF secretion for tumor cells is vascular endothelial cells. Given that the brain is one of the most vascular-rich organs, premalignant cells that arise and propagate GBM are presumably surrounded with enriched vascular endothelial cells, at least in the initial stage (favorable condition for PN tumor cells). On the other hand, when a tumor expands rapidly, vascularization may not catch up with requirement of nutrition for individual tumor cells, thereby creating a large central necrosis (favorable microenvironment for MES tumor cells). It is possible that GBM arises from PN subtype that eventually shifts its phenotype to MES tumor cells. Future studies will likely tell us the answer.

3. Glucose metabolism in GSCs as a therapeutic target

Under limited energy sources within a tumor mass, glucose uptake and its intracellular metabolism may determine whether individual tumor cells can preferentially survive and grow. We recently reported that restricted glucose conditions lead to increased GSC frequency through a combination of preferential GSC survival and acquisition of stem-like features in non-GSCs [3]. GSCs competed to absorb glucose by co-opting high-affinity glucose transporters normally expressed by neurons and embryonic stem cells glucose transporter 3 (Glut3) [3]. Targeting Glut3 expression in GSCs disrupted stem cell self-renewal and tumor growth, and high Glut3 expression was associated with increased tumor grade and poorer patient survival [3]. Collectively, in the harsh microenvironment, GSCs may instruct tumor heterogeneity because of their enhanced survival potential over nonstem tumor cells.

In a variety of cancers, reprogramming of cellular metabolism is well recognized to be a shared feature of malignancy [14-17]. Almost 90 years ago, Otto Warburg demonstrated that cancer cells preferentially obtain energy from aerobic glycolysis, which produces high amounts of lactate from pyruvate – termed the Warburg effect. The conversion of glucose to lactate in the presence of oxygen is a critical aerobic pathway that allows cancer cells to proliferate rapidly and the amount of lactate produced is correlated well with tumor malignancy [14-17]. In the MES, but not PN, subtype of GSCs, glycolysis pathway is highly upregulated creating heterogeneous cellular populations [2]. Elevated radioresistance of GSCs is likely associated with MES signature and activated glycolysis pathway and disruption of glycolysis-derived MES identity increased aggressiveness and radiosensitivity of tumor cells [2]. Collectively, these data suggest that the Warburg effect is not a consequence of a malignant phenotype but rather a causative metabolic paradigm, leading to tumor development and propagation to create tumor heterogeneity at both the cellular and the global levels. Given these observations, elimination of the glycolysis-induced metabolic reprogramming is likely a promising approach for the treatment of various cancers including GBM [18-21]. Recently, a small-scale first-in-human study was conducted with five GBM patients using the oral glycolysis inhibitor, dichloroacetate (DCA), for up to 15 months together with the oral methylator, temozolomide, with some promising results [22]. Out of five patients, three patients showed radiographical regression of tumors and four patients were clinically stable for up to 15 months. Matched immunohistochemistry analysis with three tumors before and after treatment exhibited that DCA normalizes mitochondrial functions, inhibits the target gatekeeper enzyme of DCA, pyruvate dehydrogenase kinase II and promotes cellular apoptosis, resulting in antitumor effects. In preclinical studies, an IDH1 inhibitor delayed growth and induced differentiation of mutant glioma cells [23]. These early studies suggest that targeting metabolism may offer benefit in GBM treatment.

4. Known unknown

In cancer cells, glucose provides acetyl-CoA that condenses with oxaloacetate to produce citrate and 2-HG production [24,25]. As a critical metabolite, α-ketoglutarate (α-KG) is actively carboxylated by the NADPH-linked mitochondrial and cytosolic IDH1/2 to form citrate and 2-HG, respectively [25,26]. Subsequently, citrate is required for facilitating cancer cell proliferation and 2-HG is an oncometabolite synthesized by carboxylation of α-KG. Increased α-KG production and carboxylation appear to be essential for accelerating cancer cell growth. Nonetheless, despite the accelerated cancer cell growth by α-KG, IDH-mutant tumors are more indolent and less aggressive compared with wild-type tumors. It is possible that citrate promotes lipid synthesis as a critical intermediate, which may serve as the common mechanism for aggressiveness of GBM that do not have IDH mutations. Collectively, these data indicate the complexity of distinct dependence of GBM tumors in various other, yet undetermined, oncometabolic pathways. While α-KG phenocopies malignant transformation of human fibroblasts driven by Rho family GEF oncogenic Dbl, DCA treatment of cultured GBM cells paradoxically increases α-KG expression [27]. The physiological roles of α-KG as glycolysis mediator are still controversial and require more investigation.

Significance of elucidating the molecular mechanisms of tumor metabolism in GBM has no question, and it is critical to translate this knowledge into the development of effective GBM-targeted therapeutic approaches. Successful achievement will likely elucidate GSC-specific signaling pathways and effective targeted therapies for GBM. In addition, to study the molecular mechanisms by which glucose uptake and glycolysis regulate GSCs can be applicable to other stem cells and/or cancers.

Declaration of interest

The author has no relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Acknowledgement

The American Cancer Society MRSG-08-108-01, NIH/NCI P01 CA163205, R21 CA175875, NIH/NINDS R01 NS083767 and R01 NS087913.

Notes