Over 80 years ago, Warburg discovered that cancer cells generate ATP through the glycolytic pathway, even in the presence of oxygen. The finding of this phenomenon, termed the “Warburg effect,” stimulated much research on tumorigenesis, but few explanations were forthcoming to explain the observation. Recently, advanced developments in molecular biology and high-throughput molecular analyses have revealed that many of the signaling pathways altered by gene mutations regulate cell metabolism in cancer. Furthermore, mutations in isocitrate dehydrogenase 1 and 2 were shown to elevate 2-hydroxyglutarate, which led to changes in α-ketoglutarate-dependent dioxygenase enzyme activity, resulting in an increased risk of malignant tumors. Although these findings led to a renewed interest in cancer metabolism, our knowledge on the specifics of tumor metabolism is still fragmented. This paper reviews recent findings related to key transcription factors and enzymes that play an important role in the regulation of cancer metabolism.
Interest in tumor metabolism has waxed and waned over the past century. In the 1920s, Otto Warburg discovered that cancer cells predominantly produce ATP/energy through the glycolytic pathway rather than through the tricarboxylic acid (TCA) cycle, even in the presence of adequate oxygen.[1, 2] These strategies employed by cancer cells are baffling because most mammalian cells generate ATP by oxidative phosphorylation through the TCA cycle, which utilizes oxygen. A number of cancer researchers have attempted to understand this aberrant tumor metabolism. However, initial research offered inadequate explanations of tumorigenesis, and the oncogene revolution pushed tumor metabolism to the margins of cancer research.
In recent years, high-throughput sequence data suggest that the mutations leading to tumorigenesis are more numerous and heterogeneous than previously thought.[4, 5] Indeed, there are thousands of point mutations, translocations, amplifications and deletions that may contribute to cancer development, and mutational change can differ even among histopathologically identical tumors. Moreover, detailed bioinformatics analyses have suggested that cancer-related driver mutations affect a dozen or more core signaling pathways and processes responsible for tumorigenesis. Despite this heterogeneity in tumors, it is becoming clear that certain metabolic alterations are essential for malignant cancers.
These observations have renewed interest in cancer metabolism, and recent studies have shown that cancer cells may shift metabolic pathways to facilitate the uptake and incorporation of nutrients into cell building blocks, such as nucleotides, amino acids and lipids, that are needed for highly proliferating cells. This review covers the accumulated findings regarding key transcription factors and metabolic enzymes that induce alterations in cancer metabolism, which can be attractive targets for cancer therapy.
The Warburg Effect
One of the prominent characteristics of rapidly growing tumor cells is their capacity to sustain high rates of glycolysis for ATP generation, irrespective of oxygen availability. This phenomenon is known as the Warburg effect.[2, 8, 9] This characteristic seems to be a general property of highly malignant tumors, independent of their carcinogenic origin. In fact, significant activation of glycolysis was observed in human cancer tissues through the use of a leading-edge metabolomic profiling method based on capillary electrophoresis mass spectrometry.[11-13]
In the presence of oxygen, most cells primarily metabolize glucose to pyruvate in glycolysis and then completely oxidize most of this pyruvate to carbon dioxide through the mitochondrial TCA cycle, where oxygen is the final acceptor in the electron transport chain that generates an electrochemical gradient to facilitate ATP production. During these processes, oxidative phosphorylation produces 34 molecules of net ATP per complete oxidation of one glucose molecule and glycolysis generates only two molecules of ATP per glucose molecule, which raises the question of why cancer cells use glycolysis in spite of its inefficient energy production. Although the key factors and pathways of cancer metabolism remain to be elucidated, current literature shows that several transcription factors and metabolic enzymes are crucial in mediating the aberrant metabolic behavior of tumor cells (Fig. 1).
Hypoxia-Inducible Factor 1
Hypoxia-inducible factor 1 and 2 (HIF-1 and HIF-2) complexes are heterodimeric transcription factors that are responsible for gene expression changes in hypoxia. They are composed of the constitutively expressed HIF-1β subunit and either the HIF-1α or HIF-2α subunit, both of which are rapidly stabilized upon exposure to hypoxia. The HIF-1α subunit is ubiquitously expressed whereas the HIF-2α subunit's expression is restricted to endothelial, lung, renal, and hepatic cells. Under normoxic conditions, the HIF-1α subunits undergo oxygen-dependent hydroxylation by prolyl hydroxylase 2 (PHD2) enzyme, resulting in their recognition by von Hippel–Lindau tumor suppressor (VHL), an E3 ubiquitin ligase, and subsequent degradation. However, in hypoxia, prolyl hydroxylation is suppressed, which allows the HIF-1α subunit to escape VHL-mediated destruction and accumulate to high levels. Then, HIF-1α dimerizes with the constitutively present HIF-1β subunit and accumulates in the nucleus. Subsequently, the HIF-1 dimer (i.e. the active complex of HIF-1α and HIF-1β) binds to the hypoxia response element of target genes (Table 1),[18-22] resulting in their transcriptional activation (Fig. 2). Most cancer cells are exposed to chronic hypoxia from the early stage of carcinogenesis. Indeed, measurement of oxygen tension in tumors confirms severe hypoxia in many types of cancer, thus resulting in activation of HIF-1 and expression of glucose transporters and the glycolytic pathway.
Table 1. Target genes of HIF, c-Myc and p53 associated with energy metabolism
In certain cases, the HIF-1α subunit is stabilized, even under normoxic conditions. An inactivating mutation in the VHL tumor suppressor, which is predominantly associated with the development of human glioblastoma and renal clear cell carcinoma, results in stabilization of HIF-1α. Further, inactivation or inhibition of the PHD2 enzyme suppresses hydroxylation of the HIF-1α subunit, which accumulates as a result. Oxygen, 2-oxoglutarate and divalent iron ion are used as substrates of PHD2. Therefore, a decrease in any of these substrates suppresses hydroxylation of HIF-1α, resulting in increased HIF-1 expression (Fig. 2). For this reason, oxygen deprivation (hypoxia) induces HIF-1α expression, followed by HIF-1 transcriptional activity.
In addition, the TCA cycle metabolites fumarate and succinate competitively inhibit HIF-1α hydroxylation, leading to the upregulation of HIF-1 (Fig. 2). Indeed, accumulation of these mitochondrial metabolites is observed in tumors that are associated with mutations in the TCA cycle enzymes succi-nate dehydrogenase and fumarate hydratase.[27, 28] Mutations in subunits B, C or D of succinate dehydrogenase lead to the development of paraganglioma or phaeochromocytoma, and mutations in fumarate hydratase cause leiomyoma, leiomyosarcoma or renal clear cell carcinoma.[29, 30] A shift toward glycolysis is further induced by HIF-1. It actively suppresses mitochondrial oxidative metabolism by increasing the expression of pyruvate dehydrogenase kinase 1, which phosphorylates and inactivates pyruvate dehydrogenase (PDH), the enzyme that converts pyruvate to acetyl-CoA for entry into the TCA cycle.[21, 31] Also, HIF-1 activates the expression of lactate dehydrogenase A, which converts pyruvate to lactate, the end product of glycolysis.
To summarize, the activation of HIF-1 notably contributes to a shift in metabolic flux toward glycolysis by inducing expression of glucose transporters and most of the glycolytic enzymes (Fig. 1).
c-Myc is a basic helix-loop-helix leucine zipper transcription factor that heterodimerizes with c-Myc-associated protein X (Max) at the helix-loop-helix leucine zipper domain to bind a DNA consensus core sequence, CACGTG or E-box.[33-35] In normal cells, c-Myc is induced upon growth factor stimulation, whereas it is expressed at constitutively high levels in transformed cells. Some degree of c-Myc overexpression is estimated to occur in 70% of human tumors.
In transformed cells, high levels of c-Myc promote energy production and biomolecule synthesis, which are required for rapid proliferation, independent of growth factor stimulation.[15, 22] Similar to HIF, c-Myc enhances the glycolytic pathway by increasing target gene expression from glucose transporters through pyruvate kinase as well as lactate dehydrogenase A, thereby allowing efflux of glucose-derived carbon as lactate (Fig. 1).[36, 37] Moreover, c-Myc drives anabolic pathways, with targets that include carbomyl phosphate synthetase aspartate transcarbomylase and dihydroorotase, serine hydroxymethyltransferase, fatty acid synthase and ornithine decarboxylase (Table 1).[38, 39]
Interestingly, c-Myc can collaborate with HIF to confer metabolic advantages to tumor cells. Although HIF-2α accentuates c-Myc-Max heterodimer-mediated transcriptional activation by stabilizing the c-Myc-Max complex, HIF-1α binds Max and renders c-Myc inactive when Myc is regulated at normal levels (Fig. 3). In contrast to the physiological regulation of c-Myc by growth factor stimulation of normal cells, many cancers overexpress c-Myc. When c-Myc is highly increased, its activity is not affected by HIF-1α because the high levels of c-Myc protein maintain c-Myc-Max heterodimers through mass action (Fig. 3). In this way, these transcription factors act in concert to reprogram metabolism, protein synthesis and cell cycle progression.
The transcription factor and tumor suppressor p53 is best known for its function in DNA damage response and apoptosis. p53 also plays an important role in the regulation of glycolysis and oxidative phosphorylation (Table 1). In general, p53 decreases the glycolytic rate, however, mutation or suppression of p53 frequently occurs in cancer, which results in losing control of its functions, thus promoting glycolysis.
p53 inhibits the expression of glucose transporters (GLUT) 1 and GLUT4, and decreases the levels of phosphoglycerate mutase, which converts 3-phosphoglycerate to 2-phophoglycerate.[42, 43] p53 also indirectly regulates glycolysis by modulating the nuclear factor-κ light-chain-enhancer of activated B cells pathway that upregulates glycolysis-promoting genes such as GLUT3. Expression of p53 can downregulate nuclear factor kappa B (NF-κB ) through IκB kinase (IKK), resulting in suppression of GLUT3 gene expression (Fig. 4).
In contrast, p53 increases the expression of Tp53-induced glycolysis and apoptosis regulator, an enzyme that reduces the levels of fructose-2,6-bisphosphate, an activator of phosphofructokinase (Fig. 4).[3, 45] p53 also increases oxidative phosphorylation through upregulation of the synthesis of cytochrome c oxidase 2 gene. Furthermore, wild-type p53 supports the expression of phosphate and tensin homolog deleted on chromosome 10 (PTEN), which inhibits the phosphatidylinositol 3-kinase (PI3K) pathway and leads to decreased activation of v-akt murine thymoma viral oncogene homologue 1 (AKT1) and HIF, both of which are important drivers of glycolysis (Fig. 1).
Glutaminase (GLS) 2 is another p53 target gene that encodes a mitochondrial GLS, which catalyzes the hydrolysis of glutamine to glutamate. Glutamate can be further deaminated to form α-KG, which can enter the TCA cycle for energy metabolism (Fig. 1). Glutamate also preserves total reduced glutathione after oxidative stress.
Several functions of p53 reduce flux through the glycolytic pathway and induce oxidative phosphorylation (Fig. 4). However, mutation or suppression of p53, a frequent occurrence in cancer, results in the loss of control of its metabolic regulation. Cancer cells with mutant p53 promote glycolysis in several ways, including expression of GLUT (GLUT 1, GLUT4 and GLUT3), enhancement of glycolysis enzymes (phosphofructokinase and phosphoglycerate mutase), suppression of mitochondrial respiration by inhibition of synthesis of cytochrome c oxidase 2 and GLS2, and activation of AKT and HIF, which are effectors downstream of PI3K.
Pyruvate Kinase M2
Recently, several papers reported that pyruvate kinase (PK) plays a crucial role in reprogramming of glycolytic metabolism. It is a rate-limiting enzyme that catalyzes the final reaction of glycolysis, converting phosphoenolpyruvate (PEP) to pyruvate and producing ATP (Fig. 1). Four mammalian PK isoenzymes (M1, M2, liver isoform (L) and red blood cell isoform (R)) exist and are present in different cell types. A constitutively active tetrameric form of PK, muscle isoform (PKM1) is found in normal adult cells, whereas PKM2 forms both tetramers and less active dimers. Replacement of embryonic and tumor isoform (PKM2) by PKM1 in tumor cell lines renders them less glycolytically active and diminishes tumor xenograft growth, suggesting that PKM2 is responsible for the Warburg effect.
To form the active tetramer, PKM2 requires fructose-1,6-bisphosphate. This tetrameric form of PKM2, found in differentiated tissues and normal proliferating cells, has a high affinity to its substrate PEP, which promotes the conversion of PEP to pyruvate. When PKM2 is in its less active dimeric form, found predominantly in cancer cells, all glycolytic intermediates above PK accumulate and may be further directed towards anabolic processes for nucleic acid, phospholipid and amino acid synthesis (Fig. 1). Indeed, nucleic acids, phospholipids and amino acids are all important cell building-blocks that are greatly needed by highly proliferating cells, such as tumor cells.
The phosphorylated tyrosine 105 site in PKM2 releases fructose-1,6-bisphosphate from PKM2 to the less active dimeric form, which suggests that phosphorylation of PKM2 decreases its activity. In addition, increases in intracellular concentrations of reactive oxygen species causes oxidation of the cysteine 358 residue in PKM2, which inhibits PKM2 multimer formation and results in its low activity. The suppression of PKM2 allows for accumulation of its substrate, PEP, which inhibits the glycolytic enzyme triose phosphate isomerase that converts the three carbon sugars glyceraldehydes 3-phosphate and dihydroxyacetone phosphate, leading to activation of a pathway alternative to glycolysis, the pentose phosphate pathway. Increased activity of this pathway provides ribose-5-phosphate, used for nucleotide and nucleic acid synthesis, and NADPH, a reducing factor required for antioxidant enzyme activity and for recycling the antioxidant glutathione. Therefore, PKM2 activated cancer cells can produce antioxidant glutathione, which protects against reactive oxygen species (Fig. 1).
Mutations in isocitrate dehydrogenase 1 (IDH1) and IDH2 are found in up to 70% of low-grade glioma and secondary glioblastoma multiforme, as well as in 10% of acute myeloid leukemia in humans.[57-59] Both IDH1 and IDH2 catalyze the oxidative decarboxylation of isocitrate to α-KG in the cytosol/peroxisomes and in the mitochondria as part of the TCA cycle (Fig. 1). Heterozygous somatic mutations at arginine R132 (IDH1) and at R140 or R172 (IDH2) in the active sites of these enzymes confer a gain in function to the enzyme, which can result in production of the oncometabolite 2-hydroxyglutarate from α-KG (Fig. 1).[60, 61]
Elevated levels of 2-hydroxyglutarate competitively inhibit several α-KG-dependent dioxygenase enzymes that use α-KG as a substrate to catalyze a wide range of reactions, including alterations in histone and DNA methylation,[62-64] biosynthesis of collagen or L-carnitine, and response to hypoxia. As described in the section on HIF, PHD2 is an α-KG-dependent dioxygenase enzyme that can destabilize HIF-1 by hydroxylating a proline residue, leading to VHL-mediated proteasome degradation. Mutations in IDH1 R132 or IDH2 R140 or R172 consume α-KG, a substrate of PHD2, and highly increase 2-hydroxyglutarate, a PHD2 inhibitor. This suppresses PHD activity and maintains HIF-1α, resulting in the activation of glycolysis (Fig. 1).
Cancer cells contain an increased number of lipid droplets compared with normal tissues. Cytosolic acetyl-CoA is the central biosynthetic precursor for fatty acid and cholesterol synthesis.[66, 67] Normally, acetyl-CoA is generated primarily from glucose-derived pyruvate through the action of PDH in mitochondria. Because mitochondrial acetyl-CoA cannot be directly transported to the cytosol, acetyl-CoA is first converted to citrate, then exits to the cytosol and is finally converted to acetyl-CoA by ATP-citrate lyase (Fig. 1).[68-70]
Cancer cells exposed to hypoxia, however, can stabilize HIF-1, which inhibits PDH by promoting the expression of pyruvate dehydrogenase kinase, and induces expression of lactate dehydrogenase. Thus, glucose-derived pyruvate is shunted away from the TCA cycle toward glycolysis (Fig. 1). This metabolic shift raises the question of how cancer cells produce cytosolic acetyl-CoA for fatty acid synthesis.
Some papers have recently reported that glutamine can contribute carbon to lipogenic acetyl-CoA through two distinct pathways: glutamine and glutamine-derived α-KG (Fig. 1). Glutamine is transported at a high rate into proliferating cells.[71, 72] Several cell lines can reductively metabolize glutamine-derived α-KG and generate acetyl-CoA for lipid synthesis through a cytosolic IDH1-dependent pathway. Alternatively, glutamine-derived α-KG is reductively carboxylated by mitochondrial IDH2 to form citrate, which can then be transported to the cytosol for generation of acetyl-CoA. These reductive carboxylations of glutamine are also part of the metabolic reprogramming associated with HIF-1. Glutamine metabolism in many cancer cells is regulated by GLS1, which has associated microRNA miR-23a and miR23b that are transcriptionally repressed by c-Myc, resulting in greater expression of their target protein, GLS1 (Fig. 1).[49, 73]
Using global metabolomic analysis and molecular biological technology, the author and his colleagues have recently obtained strong evidence that the energy generation pathway of cancer, especially under hypoxic and nutrient-deprived conditions, relies on fumarate respiration.[74, 75] Fumarate respiration confers upon cells the ability to generate ATP by converting fumarate to succinate, instead of oxygen to water, as the final electron transport step in the reverse reaction of succinate dehydrogenase. Succinate, the end-product of fumarate respiration, was found to be markedly higher, as were fumarate and malate levels, in colon, lung, and prostate tumor tissues than in their corresponding normal tissues (Kenjiro Kami, personal communication, 2012). Therefore, high concentrations of these metabolites in tumors might be attributed to upregulation of fumarate respiration, which would facilitate tumor growth and proliferation, even under hypovascular microenvironment. However, further investigation is necessary to confirm this metabolic reprogramming.
The metabolic shift to aerobic glycolysis is a common hallmark of cancer. Cancer cells appear to have adapted in order to facilitate the incorporation of nutrients into cell building blocks (nucleotides, lipids, amino acids) and anti-oxidant glutathione to produce a new cell. Although there is still much to learn about the regulation of cancer cell metabolism, it is becoming clear that metabolic reprogramming is induced by HIF-1, c-Myc, p53, PKM2, IDH, GLS and other molecules that act both independently and in concert with each other. These cancer-specific metabolic pathways have recently been used for cancer diagnosis and therapy. Aerobic glycolysis is often accompanied by increased glucose uptake, a phenomenon that has been visualized in tumors of patients usingF-fluorodeoxyglucose PET. In contrast, blocking these metabolic pathways or restoring altered pathways can lead to new approaches in cancer treatment, and consequently, some of these metabolic enzymes are currently being considered as therapeutic targets for cancer.
I would like to express my deep gratitude to Dr Mitsuhiro Kitagawa, Dr Kenjiro Kami for his comments, and Kumi Suzuki for her help in preparaing the figures. This work was partly supported by a Grant-in-Aid for Scientific Research on Innovative Areas (No. 22134007), Japan.
The author has no conflict of interest to declare.