The M2 splice isoform of pyruvate kinase (PKM2), an enzyme that catalyzes the later step of glycolysis, is a key regulator of aerobic glycolysis (known as the Warburg effect) in cancer cells. Expression and low enzymatic activity of PKM2 confer on cancer cells the glycolytic phenotype, which promotes rapid energy production and flow of glycolytic intermediates into collateral pathways to synthesize nucleic acids, amino acids, and lipids without the accumulation of reactive oxygen species. PKM2 enzymatic activity has also been shown to be negatively regulated by the interaction with CD44 adhesion molecule, which is a cell surface marker for cancer stem cells. In addition to the glycolytic functions, nonglycolytic functions of PKM2 in cancer cells are of particular interest. PKM2 is induced translocation into the nucleus, where it activates transcription of various genes by interacting with and phosphorylating specific nuclear proteins, endowing cancer cells with a survival and growth advantage. Therefore, inhibitors and activators of PKM2 are well underway to evaluate their anticancer effects and suitability for use as novel therapeutic strategies. Clin Cancer Res; 18(20); 5554–61. ©2012 AACR.
Obtaining sufficient energy is a critical issue for cells to survive. Regardless of local availability of molecular oxygen, mainly cancer cells primarily use glycolysis to produce ATP. This unique feature is called aerobic glycolysis or the Warburg effect (1, 2). Recent studies show that pyruvate kinase M2 (PKM2) is a key glycolytic enzyme that regulates the Warburg effect and is necessary for tumor growth (3). Incorporated glucose is converted into pyruvate through several steps in cytoplasm (Fig. 1). Pyruvate can be converted to lactate or to acetyl-CoA, and such direction is determined by the enzymatic activity of PKM2 (3–6). Low PKM2 activity promotes conversion to lactate and leads to Warburg effect, whereas high activity of both PKM2 and PKM1 promotes conversion to acetyl-CoA (Fig. 1; refs. 3–6). Recently, not only the glycolytic but also the nonglycolytic functions of PKM2 have attracted a great deal of attention. In this review, we describe the functions of PKM2 in cancer cells and discuss potential therapeutic applications.
Four Isoforms of Pyruvate Kinase
Pyruvate kinase (PK) is a glycolytic enzyme that catalyzes a reaction generating pyruvate and ATP from phosphoenolpyruvate (PEP) and ADP (Fig. 1). Four isoforms of PK (L, R, M1, and M2) are present in mammals. The L and R isotypes are encoded by the PKLR gene. Their expression is tissue specific and is regulated by different promoters. The L isotype is expressed in the liver, kidney, and intestine, and the R isotype is expressed in red blood cells (7–10). PKM1 and PKM2 are encoded by the PKM gene and are the products of 2 mutually exclusive alternatively spliced exons (exon 9 and exon 10, respectively; refs. 7, 11, 12): M1 is expressed in most adult differentiated tissues such as brain and muscle, whereas M2 is expressed in embryonic cells, adult stem cells, and cancer cells (3, 8, 9, 13–15). Splicing of PKM is controlled by the splicing repressors, heterogeneous nuclear ribonucleoprotein (hnRNP) A1 and A2, as well as polypyrimidine tract binding protein (PTB, also known as hnRNPI), and the expression of those repressors is upregulated by MYC oncoprotein (Fig. 2; refs. 7, 11). These proteins bind to exon 9 and repress PKM1 mRNA splicing, resulting in the inclusion of exon 10 and thereby contributing to the high levels of PKM2 expression (7, 11, 16).
PKM2 Expression in Cancer Cells
In addition to embryonic cells and adult stem cells, PKM2 is a major isoform expressed in cancer cells (3, 8, 9, 13–15). The expression of PKM2 is necessary for tumor growth (3, 15, 17–20) and a number of regulators of PKM2 expression have been reported (7, 8, 11, 16, 21–23). A recent study showed that PKM2 expression is induced by activated mTOR, which transactivates hypoxia-inducible factor 1 (HIF-1) and promotes the c-Myc-hnRNPs–mediated alternative splicing, leading to the aerobic glycolysis in tumor cells (24).
Enhanced expression of PKM2 is observed both in various cancer cell lines and in samples such as blood, serum, and stool from cancer patients (3, 9, 15, 25–27). Given that the upregulation of PKM2, concomitant with the downregulation of PKM1, is induced in skin cells within 24 hours of treatment with the tumor promoter, 12-O-tetradecanoylphorbol-13-acetate, PKM2 expression might be an early event in carcinogenesis (28). Thus, PKM2 can be a useful biomarker for the early detection of tumors.
Glycolytic Functions of PKM2
The expression and lower glycolytic enzyme activity of PKM2 are necessary for the Warburg effect, which provides cancer cells with selective advantages, including tumor growth and suppression of reactive oxygen species (ROS; refs. 3, 4, 15, 18) for the following 2 reasons. First is that the glycolytic pathway generates ATP more rapidly than the oxidative phosphorylation (29), allowing faster incorporation of carbon into its biomass (4, 30). Between yield and rate of ATP production, a trade-off has been reported to be present in sugar degradation by glycolysis and mitochondrial respiration. Then, glycolysis generates ATP at a high rate but low yield via massive consumption of glucose (29, 31). The second reason is that lower activity of PKM2 facilitates the production of glycolytic intermediates to enter the glycolysis branch pathways, such as glycerol synthesis and the pentose phosphate pathway, which generates NADPH to suppress ROS production and is also involved in nucleotide synthesis (Fig. 1; refs. 4, 15, 30, 32, 33). In other words, the increase in glycolysis induced by the lower activity of PKM2 can supply cancer cells with varied resources of substrates necessary for their rapid proliferation.
PKM2 exists as either a low-activity dimeric or high-activity tetrameric form, whereas PKM1 constantly exists as a high-activity tetrameric form (Figs. 1 and 2; refs. 8, 9, 34). Cancer cells predominantly express the low-activity dimeric form of PKM2 (3, 5, 6), whereas normal proliferating cells express the high-activity tetrameric form (9). Christofk and colleagues (3) and Vander Heiden and colleagues (35) from Cantley's group reported that PKM1-expressing cells showed much higher PK activity than PKM2-expressing cells; these cells consumed more oxygen, produced less lactate, and were highly sensitive to the mitochondrial ATP synthesis inhibitor, oligomycin (3). In addition, Hitosugi and colleagues reported that tyrosine phosphorylation (Tyr 105) of PKM2 disrupts the active tetrameric form of PKM2, leading to the suppression of its activity. Furthermore, PKM2-mutated cells, in which tyrosine residue 105 is replaced with a phenylalanine, had increased PK activity as observed in PKM1-expressing cells (6). Therefore, the low activity of dimeric PKM2 is a very important driver for glycolysis. In contrast, the high activity of PKM2 and PKM1 tetramers drives the tricarboxylic acid (TCA) cycle (Figs. 1 and 2; refs. 3–6).
Various factors have been reported to control the switch between the dimeric and tetrameric forms of PKM2 (5, 6, 8, 19, 36–44). For example, fructose-1,6-bisphosphate (fructose-1,6-P2), a glycolytic intermediate, binds allosterically to PKM2 and facilitates the formation of the active tetramer. Serine, which is produced from a glycolytic intermediate 3-phosphoglycerate, is also a positive regulator of PKM2 (Figs. 1 and 2A; refs. 8, 45–48). In contrast, tyrosine phosphorylation of PKM2 induces the release of fructose-1,6-P2, which causes PKM2 to convert from tetrameric form to less active dimeric form (5, 6). In addition, oncoproteins such as HPV-16 E7 and activated pp60v-src kinase dissociate the tetrameric form to yield the dimeric form (49). Furthermore, recent studies show that oxidative stress causes dissociation of the tetramer and a subsequent reduction in PKM2 activity (Figs. 1 and 2; ref. 18), and that acetylation of lysine residue within PKM2 suppresses its catalytic activity and induces the degradation by chaperone-mediated autophagy (22). In addition, it has been reported that mucin 1 phosphorylated by EGF receptor (EGFR) interacts with PKM2 and suppresses its activity (50).
Nonglycolytic Functions of PKM2
A variety of molecules interact with PKM2 (5, 6, 8, 14, 18, 19, 21, 23, 24, 28, 36–45, 50–65). Many of them affect the glycolytic functions of PKM2, which directly regulate the Warburg effect. However, an increasing number of reports document the nonglycolytic functions of PKM2. In particular, the role of PKM2 in transcription is attracting attention. It has been reported that PKM2 interacts directly with the HIF-1 subunit and promotes transactivation of HIF-1 target genes (Fig. 3; ref. 21). As HIF-1 also activates the transcription of the genes encoding PKM2, cancer cells may have the positive feedback loop between PKM2 and HIF-1, which contributes to the characteristic metabolism in cancer cells.
There have been several studies about PKM2 nuclear translocation. It is induced not only by such events as interleukin-3 stimulation and EGFR activation, which relate to cell proliferation (19, 52), but also by apoptotic stimuli such as somatostatin analogues, hydrogen peroxide (H2O2), and UV light (58). Nuclear PKM2 has been shown to activate gene transcriptions and cell proliferation (14, 19–21, 52, 61). Translocation of PKM2 into the nucleus induced by EGFR activation was reported to promote β-catenin transactivation, leading to expression of cyclinD1 and c-Myc (Fig. 3; ref. 52). Given that c-Myc upregulates transcription of hnRNPs contributing to the high PKM2/PKM1 ratio (7, 11), and that c-Myc promotes glycolysis by driving the expression of glucose transporter1 (GLUT1) and lactate dehydrogenase A (66, 67), the events induced by the translocation of PKM2 into the nucleus may be connected with a feed-forward loop to drive glycolysis.
Nuclear PKM2 has been reported to exist as a dimeric form, whereas cytoplasmic PKM2 exists as both dimeric and tetrameric form (20). In addition, the nuclear PKM2 was shown to act as a protein kinase that can phosphorylate STAT3 and thereby activate transcription of cancer-relevant genes such as Mek5 (Fig. 3; ref. 20). Taking these findings together, we speculate that PKM2 has 2 faces for conferring great benefits on cancer cells. First is that dimeric PKM2 in cytoplasm acts as a PK in which enzymatic activity is low for maintaining the glycolytic phenotype and promoting glycolytic intermediates to a collateral pathway for biosynthesis. The second is that the dimeric PKM2 translocated in the nucleus acts as an active protein kinase to phosphorylate specific nuclear proteins. Recently, many other roles of PKM2 in various condi-tions such as immunologic responses (43, 44, 63, 64), genomic instability (37), angiogenesis (57), pathogenesis (36, 41, 62), and some disease (54) have been reported. It is important to define whether those events are associated with glycolytic or nonglycolytic functions of PKM2.
PKM2 and the Management of Oxidative Stress
CD44 is a major adhesion molecule and a cell surface marker for cancer stem cells. CD44 has been shown to be implicated in tumorigenesis as well as tumor growth and metastasis (51, 68, 69). Recently, we found that CD44 interacts with PKM2. The CD44/PKM2 interaction suppresses PKM2 enzymatic activity via tyrosine phosphorylation, thereby promoting glycolysis and increasing the flux to the pentose phosphate pathway (PPP) in glycolytic cancer cells that are either p53 dysfunctional or hypoxic (Fig. 4; ref. 51). CD44 ablation enhances mitochondrial respiration through metabolic shift from glycolysis, which may suppress GLUT1 expression, thereby reducing glucose uptake, and subsequent PPP flux. Such metabolic changes induced by CD44 ablation lead to the downregulation of reduced glutathione (GSH) synthesis in cells, along with a consequent increase in ROS accumulation (Fig. 4; ref. 51). Interestingly, PKM2 activity is inhibited by oxidative stress as well as tyrosine phosphorylation (18). Oxidative stress induces the oxidization of Cys358 within PKM2, which promotes glycolysis and PPP flux, leading to the production of GSH and consequent ROS depletion (18, 70, 71). Thus, cancer cells have multiple mechanisms for avoiding ROS accumulation, which gives them a survival advantage in terms of tumor growth and therapeutic resistance (18, 51, 72).
Therapeutic Possibilities for PKM2
Cancer-specific metabolism, which is associated with therapeutic resistance, is an attractive target for cancer therapy (30, 66, 67, 73–78). As outlined above, PKM2 plays many important roles in metabolism and signals in cancer cells and, thereby, can be an ideal therapeutic target. The RNA interference and the peptide aptamers which ablate PKM2 have been reported to elicit anticancer effects, such as the impairment of tumor growth, induction of apoptotic cell death, and increasing sensitivity to chemotherapy (3, 17, 56, 79, 80). Small-molecule inhibitors of PKM2 have already been identified, and these inhibitors suppress glycolysis and cause cell death. However, therapies targeting PKM2 expression are problematic. One problem is that PKM2 is also expressed in some normal tissues (8). The other is that PKM2 knockdown by short hairpin RNA does not completely disrupt cancer cell proliferation; nevertheless, it induces near-complete depletion of PKM2 activity (77). Thus, the compounds that activate the catalytic function of PKM2 may have a role as therapeutic modalities (32, 33, 77, 81). Given that the activity of PKM2 expressed in normal tissues is high, whereas that in cancer cells is low (9), such activators may inhibit glycolysis and proliferation in cancer cells with less toxic effects in normal cells. PKM2 activators show a similar effect of fructose-1,6-P2, which induces tetrameric formation of PKM2 (33). Because nucleic PKM2 exists as a dimer, PKM2 activators have a potential to prevent PKM2 from moving into the nucleus and thereby suppress functions of nucleic PKM2. In other words, PKM2 activators may be capable of inhibiting both glycolytic and nonglycolytic functions of PKM2.
Inhibition of glycolytic function of PKM2 can suppress glycolysis, leading to an increase in ROS production and a decreasing supply of varied resources of substrates necessary for rapid cancer cell proliferation. Inhibition of nonglycolytic function of PKM2 can suppress the activation of cancer-relevant genes such as Mek5, c-Myc, and various HIF-1 target genes (Fig. 3). Further studies are required to ensure their suitability for cancer therapy.
Our group reported that CD44 ablation increases the enzyme activity of PKM2, thereby inducing metabolic changes, including a shift from glycolysis to TCA cycle, leading to suppression of glucose uptake mediated downregulation of GLUT1 expression (51). Such metabolic changes induced by CD44 ablation reduce the flux to PPP and consequently accumulate ROS (Fig. 4). This sensitizes glycolytic cancer cells such as p53-mutated cells and hypoxic cells to anticancer drugs. Thus, our data suggest that indirect activation of PKM2 may have a synergistic effect when used in combination with existing therapies.
Conclusions and Future Directions
Both the glycolytic and nonglycolytic functions of PKM2 provide cancer cells with growth and survival advantages. PKM2 expression was shown to be involved in early tumorigenesis (28) and can be detected in the blood and stool of cancer patients (9, 82); furthermore, increases in PKM2 levels were reported to correlate with tumor size and stage (9). As described in another article in this CCR Focus section by Yang and colleagues, mutations of IDH1 and IDH2 are also reported to be easily detectable using small amounts of tumor samples (83). Accordingly, these enzymes involved in cancer metabolism can be useful biomarkers.
It was recently shown that the microenvironment also confers metabolic heterogeneity on cancer tissues (15, 77). Such metabolic heterogeneity may be caused by the differences in the supply of nutrients and oxygen because each cancer cell is located at a different distance from the blood vessels (77). Acetylation of PKM2, which inhibits its catalytic activity, is triggered by high glucose levels (22, 30, 84). PKM2 acts as a metabolic sensor in a manner that is dependent upon the glucose supply (79). The regulation of HIF-1 and PKM2 also differs according to the degree of hypoxia (21, 85). In addition, Hamanaka and colleagues suggested that PKM2 inhibitors suppress ATP production under severely hypoxic conditions; however, PKM2 activators would increase oxidative damage under moderately hypoxic conditions (30). Interestingly, PKM2-overexpressing stromal cells have been reported to supply recycled chemical building blocks such as lipids or nucleotides via autophagy, increase ketone body secretion, and fuel mitochondrial respiration of adjacent cancer cells (86).
In conclusion, the status of PKM2 is regulated by both intrinsic and extrinsic factors. Through the glycolytic and nonglycolytic functions, PKM2 contributes to the malignant phenotype of cancer cells, suggesting it could be an excellent target for cancer therapy. However, PKM2 has multiple faces and the intracellular events elicited by PKM2 are far more complicated than previously assumed. Furthermore, the effects of targeting PKM2 on normal cells have not been fully assessed. Therefore, further studies are needed before inhibitors and activators of PKM2 can be used as therapeutic interventions.
Disclosure of Potential Conflicts of Interest
H. Saya, commercial research grants, Kyowa Hakko Kirin Co. Ltd and Daiichi Sankyo Co. Ltd. No potential conflicts of interest were disclosed by the other authors.
Conception and design: M. Tamada, H. Saya
Development of methodology: M. Tamada
Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Tamada
Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Tamada
Writing, review, and/or revision of the manuscript: M. Tamada, M. Suematsu, H. Saya
Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Tamada
The authors thank members of the division of Gene Regulation, IAMR, School of Medicine, Keio University, for their important suggestions, and K. Arai for help with preparation of the manuscript.
This work was supported in part by a Grant-in-Aid for Scientific Research from the Ministry of Education, Culture, Sports, Science and Technology (MEXT), Japan (H. Saya), and the Global COE Program, MEXT, Japan (M. Tamada).