The Glycolytic Switch in Tumors: How Many Players Are Involved?

Abstract

Reprogramming of cellular metabolism is a hallmark of cancers. Cancer cells more readily use glycolysis, an inefficient metabolic pathway for energy metabolism, even when sufficient oxygen is available. This reliance on aerobic glycolysis is called the Warburg effect, and promotes tumorigenesis and malignancy progression. The mechanisms of the glycolytic shift in tumors are not fully understood. Growing evidence demonstrates that many signal molecules, including oncogenes and tumor suppressors, are involved in the process, but how oncogenic signals attenuate mitochondrial function and promote the switch to glycolysis remains unclear. Here, we summarize the current information on several main mediators and discuss their possible mechanisms for triggering the Warburg effect.

Introduction

Reprogramming of glucose metabolism is a key event in tumorigenesis. Cancer cells undergo a metabolic switch from oxidative phosphorylation (OXPHOS) to glycolysis in which a molecule of glucose is degraded to two molecules of pyruvate (Fig 1). Depending on the supply of oxygen for the cells, pyruvate is either reduced to lactate in the absence of oxygen via an anaerobic glycolysis pathway, or oxidized to yield acetyl-coenzyme A in the presence of oxygen and then oxidized completely to CO₂ and H₂O via citric acid cycle. The majority of cancer cells depend on high rates of glycolysis for growth and survival, even when there is sufficient oxygen ^{1, 2}. This type of aerobic glycolysis is called the Warburg effect, and the mechanisms underlying this reprogramming are not fully understood. The Warburg effect has long been linked to hypoxia, but it is not solely adaptive to hypoxia, as it also occurs under normoxic conditions ^{1, 2}. Although mitochondrial dysfunction in cancer cells can cause a shift in energy metabolism, a majority of tumor cells demonstrate normal mitochondrial function and OXPHOS ^3-5, and the high glycolytic flux in cancer cells does not mean impairment of OXPHOS ⁶. The high rates of glycolysis provide advantages for the survival and growth of cancer cells ⁷. Three possible explanations for tumor cell use of the glycolysis pathway, an inefficient metabolic pathway, have been proposed ^{8, 9}. First, compared to OXPHOS, the rate of ATP production through glycolysis is much more rapid ¹⁰. Secondly, high glycolytic flux provides sufficient glycolytic intermediates to meet the biosynthesis needs of the rapidly proliferating cells ^11-13. Finally, NADPH, derived from the enhanced pentose phosphate pathway (PPP) due to the accumulation of glycolytic intermediates, enables cancer cells to maintain adequate levels of reduced forms of glutathione for resistance to chemotherapeutic agents.

Figure 1.

Main steps in glycolysis and possible key enzymes regulated in the Warburg effect. The three reactions catalyzed by hexokinase (HK), phosphofructokinase-1 (PFK1), and pyruvate kinase (PK) in this process are rate-limiting steps. During glycolysis, four molecules of ATP are produced per molecule of oxidized glucose via substrate-level phosphorylation, and the net yield is two molecules of ATP after deduction of two ATPs consumed in phosphorylation. The fate of pyruvate depends largely on the availability of oxygen for the cells. Pyruvate is reduced to lactate under hypoxia via an anaerobic glycolysis pathway or, under aerobic conditions, oxidized to yield acetyl-coenzyme A, which is then oxidized completely to CO₂ via the citric acid cycle, resulting in the production of large amounts of ATP. G-6-P, glucose-6-phosphate; G6DP, glucose-6-phosphate dehydrogenase; GLUT, glucose transporter; Fru-2,6-P2, fructose-2,6-bisphosphate; LDH, lactate dehydrogenase; MCT, monocarboxylate transporter; TPI, triose phosphate isomerase.

Several mechanisms have been proposed to enable cancer cells to maintain high glycolytic flux ¹¹. First, phosphofructokinase-1 (PFK1) serves as a critical driver of glycolytic flux. The expression of PFK2 is upregulated in cancer cells and promotes the production of fructose-2,6-bisphospate, which acts as a potent allosteric activator of PFK1 to overcome negative allosteric feedback inhibition of PFK1 by high ATP levels. Second, re-generation of NAD⁺ and lactate production mediated by lactate dehydrogenase (LDH) is instrumental in maintaining glycolysis. In addition, the expression of pyruvate kinase M2 (PKM2) is upregulated in cancer cells. Allosteric and covalent inhibition of PKM2 channels glycolytic intermediates upstream of pyruvate into biosynthetic pathways ¹¹. Although metabolic reprogramming has long been observed as a feature of neoplasia and tumor growth, the mechanism triggering and modulating this process remains largely unclear. In this review, we mainly focus on the mechanism underlying the regulation of glycolytic switch in tumors. In addition to signal molecules and transcription factors HIF-1α, c-Myc, Akt, and mTOR, the main regulators which have been well documented, several other regulators including oncogene K-Ras, tumor suppressor p53, energy sensor adenosine monophosphate activated protein kinase (AMPK), non-coding RNAs, and sirtuin family proteins and deacetylation will also be discussed.

Master regulator HIF-1α

Hypoxia-inducible factor-1 (HIF1) consists of two subunits, HIF-1α and HIF-1β, also known as ARNT. Under physiological oxygen levels, the HIF-1α subunit is sensitive to oxygen concentration and is hydroxylated by prolyl-hydroxylases (PHD) and targeted for proteasomal degradation. Increase of reactive oxygen species (ROS) under hypoxia inhibits PHD and stabilizes the HIF-1α subunit. HIF-1α is a master regulator of glycolysis and plays an important role as an activator of aerobic glycolysis and lactate production. It potentiates the transcription of glucose transporters (GLUT) and glycolytic enzymes including GLUT1, hexokinase II (HKII), pyruvate dehydrogenase kinase (PDK), and PKM2 ^14-16. The upregulation of HIF-1-mediated HKII results in a high glycolytic rate in hypoxic solid tumor ¹⁷. Phosphorylation of pyruvate dehydrogenase leads to its inactivation and inhibits conversion of pyruvate into acetyl-CoA for the tricarboxylic acid (TCA) cycle ^{18, 19}. PKM2 is less catalytically active than is PKM1. A higher level of PKM2 in tumor cells, leading to accumulation of carbohydrate intermediates, facilitates the biosynthesis of macromolecules and tumor cell proliferation. HIF-1α drives expression of many glycolytic enzymes, and hypoxic glycolysis is, in turn, necessary for maintaining HIF-1α activity. This constitutes a novel feed-forward mechanism of glycolysis-HIF-1α signaling (Fig 2) ²⁰.

Figure 2.

HIF-1α is a master regulator of the Warburg effect and plays a critical role as an activator of aerobic glycolysis. Hypoxia increases the production of ROS, which stabilizes HIF-1α. HIF-1α induces expression of glucose transporters and glycolytic enzymes, facilitating glycolysis that is, in turn, essential for maintaining HIF-1α activity. Many oncogenes and tumor suppressors are involved in the regulation of HIF-1α. EZH2, enhancer of zeste 2 polycomb repressive complex 2; GRIM-19, gene associated with retinoid-interferon-induced mortality-19; LSD1, lysine specific demethylase 1; ROS, reactive oxygen species; RPS7, ribosomal protein S7; TKTL1, transkelolase-like 1; VEGF, vascular endothelial growth factor; VHL, von Hippel-Lindau; WWOX, WW domain-containing oxidoreductase.

HIF-1α activity is tightly regulated by oncogenes and other factors. For example, the WW domain-containing oxidoreductase (WWOX), lacking in many cancer types, interacts with HIF-1α and modulates its levels and transactivation function. WWOX absence is associated with enhanced glycolysis, and WWOX-deficient cells are more tumorigenic ²¹. Enhancer of zeste 2 polycomb repressive complex 2 (EZH2), a multifaceted oncogenic protein, promotes glioblastoma tumorigenesis and malignant progression through activation of HIF-1α and the Warburg effect. HIF-1α activation is necessary for EZH2-mediated metabolic adaption ²². Ribosomal protein S7 (RPS7) inhibits glycolysis in colorectal cancer by suppressing the expression of HIF-1α as well as of GLUT4 and lactate dehydrogenase B (LDHB) ²³. Vascular endothelial growth factor (VEGF) enhances glycolysis in pancreatic cancer via upregulation of HIF-1α ²⁴. Histone demethylase JMJD1A facilitates glycolysis via coactivation of HIF-1α and promotes cancer progression ²⁵.

HIF-1α activity can be regulated by modifying its stabilization. Transkelolase-like 1 (TKTL1) contributes to carcinogenesis through increased HIF-1α stabilization and the upregulation of downstream glycolytic enzymes and aerobic glycolysis ²⁶. The von Hippel-Lindau (VHL) gene is a tumor suppressor involved in the regulation of HIF-1α stability. VHL protein serves as an E3 ligase that ubiquitinates HIF-1α and results in its degradation by the proteasome. HIF-1α becomes constitutively activated in the absence of VHL ²⁷. The gene associated with retinoid-interferon-induced mortality-19 (GRIM-19), a potential tumor suppressor, promotes VHL-mediated HIF-1α ubiquitination and degradation in glioblastoma cells ²⁸. Lysine specific demethylase 1 (LSD1), a histone demethylase, prevents HIF-1α from subsequent acetylation-dependent degradation and maintains the HIF1α-dependent glycolytic process ²⁹.

Akt and mTOR signaling

Akt is a serine/threonine kinase that promotes cancer growth and has been called 'Warburg kinase,' because it facilitates a glycolytic switch in tumor cells under normoxic conditions ^{30, 31}. Akt activation promotes the expression and activity of glucose transporters and glycolytic enzymes. The transcription of GLUT1, a widely expressed glucose transporter, is enhanced upon the activation of Akt ^{32, 33}. Lack of S-phase kinase-associated protein 2 (Skp2), an E3 ligase, impairs Akt activation, GLUT1 expression, glycolysis, and cancer progression ³⁴. Akt signaling induces the expression of HKII, a rate-controlling enzyme of glycolysis ^{33, 35}. Akt phosphorylates and activates PFK2 to produce fructose-2,6-bisphosphate, an allosteric activator of PFK1 ³⁶. Active Akt accumulates in the mitochondria during hypoxia and phosphorylates pyruvate dehydrogenase kinase 1 (PDK1) to inactivate the pyruvate dehydrogenase complex, switching tumor metabolism toward glycolysis ³⁷. Importantly, Akt-mediated aerobic glycolysis does not affect the rate of OXPHOS. Increased glycolytic flux is required for rapidly proliferating tumor cells to obtain essential metabolic intermediates. Akt-mediated enhanced aerobic glycolysis results in acquired radioresistance of tumor cells ³⁸. Constitutively active Akt leads to cell death in low-glucose conditions ³¹.

The mammalian target of rapamycin (mTOR) is also a serine/threonine kinase downstream of Akt and consists of two complexes, mTORC1 and mTORC2. mTOR acts as a central activator of the Warburg effect by inducing expression of glycolytic enzymes under normoxic conditions. mTOR-mediated upregulation of PKM2, a rate-limiting glycolytic enzyme expressed exclusively in tumor cells, is critical to aerobic glycolysis and tumor growth ³⁹. Tuberous sclerosis protein 1 and 2 complex (TSC1/TSC2) negatively regulates the expression of GLUT3 through the inactivation of mTORC1 signaling ⁴⁰. The transmembrane mucin MUC16 increases glycolysis through activation of mTOR. The mTOR-mediated expression of glycolytic proteins involves activation of HIF-1α, NFκB, and c-Myc ^39-42. Upon stimulation, the receptor tyrosine kinases (RTKs) activate membrane PI3K, which recruits and activates Akt. Thus, RTKs-PI3K-Akt-mTOR signaling plays a critical role in the regulation of aerobic glycolysis and tumor growth (Fig 3) ^43-45.

Figure 3.

The RTKs-PI3K-Akt-mTOR signal pathway plays an important role in the regulation of aerobic glycolysis. Akt and mTOR are central activators of the Warburg effect, promoting the expression of glucose transporters and glycolytic enzymes, which is regulated by many signal molecules. AMPK, adenosine monophosphate activated protein kinase; MUC16, mucin 16; mTOR, mammalian target of rapamycin; RTKs, receptor tyrosine kinases; Skp2, S-phase kinase-associated protein 2; TSC1/TSC2, tuberous sclerosis protein 1 and 2 complex.

Oncogenes and tumor suppressors

Oncogenic K-Ras promotes metabolic reprogramming in tumors ^{46, 47}. Mutated K-Ras has been found to upregulate the expression of the GLUT1 and facilitate cell survival in low-glucose culture conditions via increased glucose uptake and glycolysis ⁴⁸. Thus, K-Ras mutated tumor cells are highly vulnerable to the glycolytic inhibitor ⁴⁸. The small guanosine triphosphatase (GTPase) ADP-ribosylation factor 6 (ARF6) is a target of mutanted K-Ras and promotes the Warburg effect and pancreatic cancer growth ⁴⁹. The K-Ras G12D mutation stimulates glucose uptake and drives glycolytic intermediates into the nonoxidative PPP ⁵⁰. K-Ras (G12V) activation leads to mitochondrial dysfunction, promoting a metabolic switch from OXPHOS to glycolysis and enhancing the tumorigenicity of the transformed cells ⁵¹. The K-Ras G13D mutation is associated with increased expression of glycolytic proteins in colorectal cancer ⁵².

Tumor suppressor p53 negatively regulates cellular glycolysis, contributing to tumor metabolic reprogramming via promotion of mitochondrial OXPHOS and suppression of glycolysis via several routes ^{53, 54}. It downregulates the expression of glucose transporters GLUT1, GLUT3, and GLUT4 ^{55, 56} and promotes the ubiquitination-mediated degradation of phosphoglycerate mutase (PGM) ⁵⁷. p53 also directly inhibits glucose-6-phosphate dehydrogenase (G6PD), the first and rate-limiting enzyme in the PPP ⁵⁸. Inactivation of p53 and the resultant enhanced PPP glucose flux may increase glucose consumption and channel glucose to biosynthesis in tumor cells. In addition, p53 may inhibit glycolysis through its target genes ^59-61. For example, p53 induces Ras-related associated with diabetes (RRAD), which in turn inhibits the translocation of GLUT1 and glycolysis in lung cancer cells ⁵⁹. p53 downregulates glycolysis by transcribing TP53-induced glycolysis and apoptosis regulator (TIGAR) ^{60, 62}. TIGAR degrades fructose-2,6-bisphosphate (Fru-2,6-P2) to fructose-6-phosphate and causes a significant reduction in cellular Fru-2,6-P2 levels. Fru-2,6-P2 serves as an allosteric activator of PFK1 and promotes the production of fructose-1,6-bisphosphate in glycolysis. p53 also negatively regulates the PI3K-Akt-mTOR pathway through its target genes. p53 activates adenosine monophosphate activated protein kinase (AMPK), a major upstream negative regulator of mTOR, and induces Pten and TSC2 to negatively regulate PI3K-Akt signaling and mTOR activity ^{63, 64}. HKII-mediated aerobic glycolysis is required for Pten-/p53-deficiency-driven tumor growth in xenograft mouse models of prostate cancer ⁶⁵. Pten deletion promotes HKII mRNA translation via the activation of the Akt-mTORC1-4EBP1 axis ⁶⁵. Absence of p53 enhances HKII mRNA stability through the inhibition of miR143 biogenesis ⁶⁵.

Given the high mutation rate of p53 in human tumors, the loss of p53 function could be an important factor contributing to the Warburg effect. It has been determined that tumor-associated mutant p53 (mutp53) drives the Warburg effect under normoxia, and inhibition of glycolysis impairs mutp53-promoting tumorigenesis ⁶⁶. Mutant R175H and R273H p53 proteins trigger PKM2 phosphorylation via mTOR signaling ⁶⁷. CD147 promotes reprogramming of glucose metabolism by inhibiting the p53-dependent signaling pathway ⁶⁸.

Upregulation of glucokinase, PK, and PFK2 levels was observed in the liver of c-Myc transgenic mice about two decades ago, suggesting that transcription factor c-Myc is a regulator of glycolytic enzymes ⁶⁹. c-Myc promotes glucose uptake via the upregulation of GLUT1 ^{14, 70} and potentiates transcription of glycolytic enzymes HKII, PFK ¹⁴, and lactate dehydrogenase A (LDHA) ^{14, 71, 72}. c-Myc upregulates the expression of monocarboxylate transporter (MCT) through direct transcriptional activation or by suppressing transcription of miR-29a and miR-29c ⁷³. c-Myc promotes transcription of polypyrimidine tract binding protein (PTB), which binds to PKM pre-mRNA and switches PKM splicing to favor the PKM2 variant, ensuring a high PKM2/PKM1 ratio ^{74, 75}. Inhibition of c-Myc in tumor cells blunts hypoxia-dependent glycolytic reprogramming and is a potential strategy for tumor therapy ^{76, 77}.

Some molecules promote glycolysis via regulation of c-Myc activity. Proto-oncogene human pituitary tumor-transforming gene (PTTG) regulates GLUT1 and several glycolytic enzymes via the c-Myc pathway ⁷⁸. N-Myc downstream-regulated gene (NDRG) family members can manipulate Myc-mediated tumor metabolic pathways and ultimately modify the Warburg effect ⁷⁹. NDRG2, a tumor suppressor, acts as a critical regulator of glycolysis via repression of c-Myc through downregulation of c-Myc transcriptional activator β-catenin, consequently suppressing the expression of GLUT1, HKII, PKM2, and LDHA in colorectal cancer cells ⁸⁰. Inhibitor of differentiation 1 (Id1), a transcription factor, promotes a metabolic shift to aerobic glycolysis in hepatocellular carcinoma cells by regulating the expression levels of c-Myc ⁸¹. lncRNA-MIF, a c-Myc-activated long non-coding RNA, inhibits aerobic glycolysis by promoting c-Myc degradation. lncRNA-MIF acts as a molecular sponge for miR-586, competing with Fbxw7 mRNA for miR‐586. Fbxw7 serves as an E3 ligase for c-Myc that promotes c-Myc degradation ⁸².

Energy sensor

Cellular energy metabolism is strictly regulated. Adenosine monophosphate activated protein kinase (AMPK) is a metabolic sensor that helps maintain cellular energy homeostasis ⁸³. Increases in AMP:ATP and ADP:ATP ratios activate AMPK, potentiating the metabolic process from an anabolic condition to a catabolic state by switching off the synthesis of lipids, carbohydrates, ribosomal RNA, and proteins ^{84, 85}. This leads to downregulation of glycolytic enzymes and glucose transporters. Thus, AMPK negatively regulates aerobic glycolysis in tumor cells and suppresses tumor growth in vivo ⁸⁶. Inactivation of AMPK promotes a metabolic shift to aerobic glycolysis, which requires normoxic stabilization of HIF-1α ⁸⁶.

AMPK is involved in the regulation of glycolysis in many tumors, but its underlying mechanism remains unclear. The role of AMPK in glycolytic shift is also controversial. Several studies report a glycolysis-promoting effect of AMPK. For example, AMPK supports the growth of aggressive experimental tumors in part through positive regulation of glycolysis ⁸⁷. Manganese superoxide dismutase (MnSOD/SOD2) upregulation in cancer cells increases mitochondrial ROS that sustains AMPK activation and the metabolic shift to glycolysis ⁸⁸. Astrocyte elevated gene-1 (AEG-1) mediates glycolysis and tumorigenesis in colorectal carcinoma cells via AMPK signaling ⁸⁹. Prostate cancer cell growth mediated by androgen receptor signaling is involved in an AMPK-mediated metabolic switch to glycolysis ⁹⁰. miR-101-3p targets AMPK in triple negative breast cancer to regulate glycolysis ⁹¹. AMPK is also essential to balance glycolysis and mitochondrial metabolism in acute T cell lymphoblastic leukemia ⁹².

Non-coding RNAs

microRNAs (miRNAs) are involved in the genesis of various cancers and may inhibit aerobic glycolysis via regulation of glucose uptake and glycolytic enzymes (Fig 4) ⁹³. miR-22, miR-144, and miR-1291 directly target glucose transporter GLUT1 in breast ⁹⁴, ovarian ⁹⁵, and renal cancer cells ⁹⁶, respectively, while miR-195-5p targets GLUT3 in bladder cancer cells ⁹⁷. Thus, downregulation of these miRNAs in tumors stimulates aerobic glycolysis. HKII, a key mediator of glycolysis, is another main target of miRNAs. miR-143 directly inhibits the expression of HKII and regulates cancer glycolysis ^98-102. The miR-143 level inversely correlates with HKII protein expression in several cancers, including head and neck squamous cell carcinoma (HNSCC) ¹⁰⁰, breast cancer ¹⁰², glioma ⁹⁹, and lung cancer ⁹⁸. The absence of miR-143-mediated repression of HKII may contribute to the shift toward aerobic glycolysis in tumors ¹⁰¹ and enhance stemness of glioblastoma stem-like cells ⁹⁹. miR-143 can be downregulated by mTOR activation ⁹⁸ or by miR-155 ¹⁰², which also stimulates HKII transcription via activating the signal transducer and activator of transcription 3 (STAT3). In addition to miR-143, miR-98 and miR-199a-5p directly targets HKII ^{103, 104}, and miR-29b downregulates HKII/PKM2 through directly targeting Akt ¹⁰⁵. Expression of these miRNAs is downregulated in several cancers. miR-378* induces glycolytic shift in breast cancer cells via the PGC-1β/ERRγ transcription pathway ¹⁰⁶.

Figure 4.

Non-coding RNAs target glucose transporters and glycolytic enzymes. The downregulation of several miRNAs in some tumors facilitates aerobic glycolysis and promotes the development and progression of the tumors. HIF-1α is a primary target of non-coding RNAs. miRNA absence, or lncRNA-mediated HIF-1α stabilization, enhances HIF-1α activity, contributing to the Warburg effect. PDH, pyruvate dehydrogenase; PDK, pyruvate dehydrogenase kinase.

In addition to HKII, other glycolytic enzymes and signal molecules are miRNA targets. miR-320a regulates PFK1 expression and, consequently, its lactate production ¹⁰⁷. miR-26b and miR-206 downregulate PFK2-driven glycolysis ^{108, 109}. A set of miRNAs targets LDHA and regulates glycolysis in colorectal cancer ¹¹⁰. miR-129-5p blocks glycolysis to retard hepatocarcinogenesis via targeting mitochondrial pyruvate dehydrogenase kinase 4 (PDK4) ¹¹¹. miR-448 promotes glycolytic metabolism in gastric cancer by downregulating KDM2B, a reader for methylated CpGs ¹¹². miR-21 acts as a molecular switch to regulate aerobic glycolysis in bladder cancer cells ¹¹³.

Along with targeting glucose transporters and glycolytic enzymes, many miRNAs exert their functions by targeting HIF-1α, a master regulator of glycolysis (Fig 4). It has been reported that miR-18b ¹¹⁴, miR-186 ¹¹⁵, miR-199a ^{116, 117}, and miR-592 ¹¹⁸ inhibit aerobic glycolysis through directly targeting HIF-1α in several cancer types. Low expression of these miRNAs facilitates aerobic glycolysis and promotes the development and progression of the tumors. The upregulation of HIF-1α under hypoxic conditions, in turn, suppresses miRNA expression and promotes glycolysis ^{103, 117}. miR-150 targets VHL, a specific E3 ligase for HIF-1α, and promotes the Warburg effect in glioma ¹¹⁹.

Long non-coding RNA (lncRNA) is also an important player in the regulation of the Warburg effect ^{120, 121}. lncRNA-p21 is hypoxia-responsive and is essential for hypoxia-enhanced glycolysis. It binds to HIF-1α and VHL, disrupting VHL-HIF-1α interaction and VHL-mediated HIF-1α ubiquitination, resulting in HIF-1α accumulation ¹²⁰. Long intergenic non-coding RNA for kinase activation (LINK-A), a cytoplasmic lncRNA, mediates BRK-dependent HIF-1α phosphorylation, leading to HIF-1α stabilization under normoxic conditions ¹²¹. LINK-A-dependent normoxic HIF-1α signaling promotes breast cancer glycolysis reprogramming and tumorigenesis ¹²¹.

Sirtuin family proteins and deacetylation

Sirtuins are a highly conserved family of nicotinamide adenine dinucleotide (NAD⁺)-dependent protein deacetylases that regulate a large number of cellular processes ¹²². Growing evidence demonstrates that sirtuins are involved in the regulation of cancer metabolism ^{123, 124}. Of the seven mammalian sirtuins (SIRT1-7), SIRT1, SIRT3, and SIRT6 have been implicated in the regulation of glucose utilization ^{125, 126}.

The histone deacetylase SIRT6 has been identified as a tumor suppressor that regulates aerobic glycolysis in cancer cells. Deficiency of SIRT6 in mice results in severe hypoglycemia ¹²⁷. SIRT6 acts as a histone H3K9 deacetylase to function as a co-repressor of HIF-1α and Myc and control the expression of multiple glycolytic genes ^{128, 129}. HIF-1α activity and glycolysis are increased in SIRT6-deficient cells ^{128, 129}. Lack of SIRT6 can lead to tumor formation even without activation of known oncogenes ¹²⁹.

SIRT3 is the major deacetylase within the mitochondrial matrix and works as a tumor suppressor by inhibiting the Warburg effect ^{130, 131}. SIRT3 regulates the stability of HIF-1α via lowering cellular ROS levels ^{130, 131}. Absence of SIRT3 increases cellular ROS, leading to stabilization of HIF-1α and metabolic reprogramming ^{131, 132}. In contrast, SIRT3 overexpression represses glycolysis and proliferation in breast cancer cells ¹³¹. The SIRT3-mediated alterations in ROS are attributed to deacetylation and activation of isocitrate dehydrogenase 2 (IDH2) and superoxide dismutase 2 (SOD2) ¹³³. In addition, SIRT3 deacetylates glutamate oxaloacetate transaminase 2 (GOT2) to inhibit its binding to malate dehydrogenase 2 (MDH2), consequently preventing the malate-aspartate shuttle in the mitochondrial intermembrane space ¹³⁴. The malate shuttle is able to restore cytosolic NAD⁺, which is essential for a high rate of glycolysis. SIRT3 also deacetylates and activates pyruvate dehydrogenase A1 (PDHA1) and PDH phosphatase 1 (PDP1) of the PDH complex (PDC), promoting the conversion of pyruvate to acetyl-CoA for OXPHOS ^{135, 136}

It has been reported that SIRT1 stimulates the expression of glycolysis genes and the tumor cell proliferation in pancreatic neoplastic lesions ¹³⁷. A SIRT1-mTOR/HIF-1α glycolytic pathway is required for differentiation of myeloid-derived suppressor cells into the M1 phenotype ¹³⁸.

Other regulators

Although the roles of the several mentioned master controllers are critical to the Warburg effect, other regulators are also involved in the glycolytic shift in cancer cells. Wnt signaling-mediated PDK1 expression promotes glycolysis and tumor growth ¹³⁹. CUE domain-containing protein 2 (CUEDC2) facilitates aerobic glycolysis and tumorigenesis via upregulating the GLUT3 and LDHA ¹⁴⁰. Pro-inflammatory cytokine interleukin-22 facilitates aerobic glycolysis in colon cancer cells via c-Myc and STAT3-mediated up-regulation of HKII ¹⁴¹. Carboxyl terminus of Hsc70-interacting protein (CHIP), an E3 ligase, inhibits aerobic glycolysis progression of ovarian carcinomas through CHIP-mediated PKM2 degradation ¹⁴². iNOS/NO promotes glycolysis via inducing PKM2 nuclear translocation ¹⁴³. Mitochondrial calcium uptake 1 (MICU1) increases aerobic glycolysis and chemoresistance in ovarian cancer ¹⁴⁴. Epidermal growth factor (EGF) promotes aerobic glycolysis, inducing epithelial-mesenchymal transition (EMT) and cancer stem-like cell properties in human oral carcinoma cells ¹⁴⁵. Toll-like receptor 3 signaling ¹⁴⁶ and serotonin signaling ¹⁴⁷ also trigger metabolic reprogramming of cancer cells. Molecular chaperone TNF receptor-associated protein 1 (TRAP1) ¹⁴⁸, focal adhesion kinase (FAK) ¹⁴⁹, plasma membrane-associated protein Caveolin 1 ^150-152, α/β-hydrolase domain-containing 5 (Abhd5) ¹⁵³, Krüppel-like factor 4 (KLF4) ^{154, 155}, Ecdysoneless ¹⁵⁶, and Jumonji C domain-containing dioxygenase (JMJD5) ¹⁵⁷ are associated with the glycolytic switch in tumors. Some viruses or virus-encoded proteins can induce aerobic glycolysis in tumors ^158-160.

Conclusions

The triggering of the Warburg effect is a complex process with the involvement of multiple regulators (Table 1) ¹⁶¹. HIF-1α is a master activator. In tumorigenesis, overproduced or mutated growth factors activate transcription factors HIF-1α, NFκB, and c-Myc via the RTKs-PI3K-Akt-mTOR pathway, leading to the expression of glucose transporters and glycolytic enzymes. Oncogene activation and tumor suppressor inactivation during carcinogenesis modify the key signal molecules of the PI3K-Akt-mTOR pathway and downstream HIF-1α activity, promoting glycolytic flux and tumor development. Oncoproteins may also activate sirtuins, a protein deacetylase family, directly suppressing the transcription of glycolytic enzymes or inhibiting HIF-1α and c-Myc expression. Hypoxia and the ROS accumulation and energy depletion resulting from rapid tumor growth further stimulate HIF-1α activity or regulate the production of glycolytic enzymes and glucose transporter through energy sensor AMPK. Absence of miRNAs or lncRNA dysfunction during carcinogenesis promotes aerobic glycolysis via targeting glycolytic enzymes or regulating HIF-1α. A crucial question is whether the Warburg effect is the cause or the effect of cancer. There is no doubt that aerobic glycolysis is a hallmark of tumor metabolism, and is essential to tumor survival and growth. An important focus of study is the stage in tumorigenesis at which reprogramming of glucose metabolism is initiated. Research has revealed that the expression of glycolytic enzymes is modified in the precancerous stage of some tumors ^{162, 163}. The imaging data also indicated that elevated glycolysis may occur at early-stages of neoplasia and critically contribute to cancer initiation ^{164, 165}. It has been reported that 14-3-3ζ-mediated upregulation of LDHA in early stage precancerous breast epithelial cells promotes glycolysis, contributing to breast cancer initiation ¹⁶⁶. We have found enhanced expression of several enzymes involved in glycolysis in high grade cervical intraepithelial neoplasia, a typical precancerous lesion of the cervix (Yu et al., unpublished data). This implies that the reprogramming of glucose metabolism occurs at an early stage of carcinogenesis. Additional studies are needed to shed light on this topic.

Table 1.

The major players in the glycolytic switch and their main features

Regulators	Downstream molecules	Effects	References
Akt	GLUT1, HK II, PDK1, PFK2	+	32, 33, 35-37
AMPK	HIF-1α	+, -	86
c-Myc	Glucokinase, GLUT1, HKII, LDHA, MCTs, PFK, PK, PKM2	+	14, 69-75
HIF-1α	GLUT1, HK II, PDK, PKM2	+	14-17
K-Ras	GLUT1	+	48
lncRNA	HIF-1α, VHL	+	120, 121
miRNAs	Akt, GLUT1, GLUT3, HIF-1α, HKII, LDHA, PDK4, PFK, PFKFB3, PKM2	+, -	94-105, 107-111
mTOR	GLUT3, HIF-1α, c-Myc, NFκB, PKM2	+	39-42
p53	AMPK, GLUT1, GLUT3, GLUT4, G6PD, miR143, PGM, Pten, RRAD, TIGAR, TSC2	-	55-60, 62-65
SIRT1, SIRT3, SIRT6	HIF-1α, Myc, PDHA1, PDP1	-, +	128-132, 135-137

Increased glycolysis in tumor cells provides a potential target for tumor therapy. Actually, disrupting glycolysis does interfere with tumor growth ^{167, 168}. Glucose transporters, monocarboxylate transporters, and critical glycolytic enzymes such as HK II, LDHA, PFK, and PKM2 have been proposed as potential targets. Several small molecules including lonidamine, 2-deoxyglucose (2-DG), dichloroacetate, and 3-bromopyruvate (3-BP) have been clinically tested, but many candidates are still under experimental studies ¹⁶¹.