- Email: [email protected]
Oncogenic alterations of metabolism
REVIEWS
TIBS 24 – FEBRUARY 1999
Neuhaus, H. E. (1998) J. Biol. Chem. 273, 9630–9636 Miroux, B. and Walker, J. E. (1996) J. Mol. Biol. 260, 289–298 Krause, D. C., Winkler, H. H. and Wood, D. O. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 3015–3019 Dunbar, S. A. and Winkler, H. H. (1997) Microbiology 143, 3661–3669 Möhlmann, T., Scheibe, R. and Neuhaus, H. E. (1994) Planta 194, 492–497 Hatzfeld, W-D., Dancer, J. and Stitt, M. (1990)
24 25 26 27 28
Planta 180, 205–211 29 Huber, S. C. and Edwards, G. E. (1976) Biochim. Biophys. Acta 440, 675–687 30 Batz, O., Scheibe, R. and Neuhaus, H. E. (1995) Planta 196, 50–57 31 Cai, J. and Winkler, H. H. (1996) J. Bacteriol. 178, 5543–5545 32 Cai, J. and Winkler, H. H. (1997) Acta Virol. 41, 285–288 33 Martin, W. and Müller, M. (1998) Nature 392, 37–41 34 Gupta, R. S. (1995) Mol. Microbiol. 15, 1–11
Oncogenic alterations of metabolism Chi V. Dang and Gregg L. Semenza Over seven decades ago, classical biochemical studies showed that tumors have altered metabolic profiles and display high rates of glucose uptake and glycolysis. Although these metabolic changes are not the fundamental defects that cause cancer, they might confer a common advantage on many different types of cancers, which allows the cells to survive and invade. Recent molecular studies have revealed that several of the multiple genetic alterations that cause tumor development directly affect glycolysis, the cellular response to hypoxia and the ability of tumor cells to recruit new blood vessels. A GENETICALLY ALTERED neoplastic cell has special metabolic requirements for its development into a three-dimensional tumor mass. Monolayer cultures do not reflect the three-dimensional cellular growth of an avascular tumor, which can be mimicked in soft-agar anchorage-independent-growth assays. When a tumor has grown to a detectable size, the local environment of the cancer cells often becomes heterogeneous1. Small (,1 mm diameter) tumor nodules, as well as microregions of larger tumors, often have microecological niches that display significant gradients of critical metabolites such as oxygen, glucose and other nutrients or growth factors (Fig. 1a). Tumors, in contrast to normal tissue, exist in acidic environments that result from production C. V. Dang is at the Depts of Medicine, Oncology, Pathology, and Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA; and G. L. Semenza is at the Depts of Pediatrics and Medicine, and the Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA. Email: [email protected]
68
of lactate and other acids. The cytosolic pH of tumor cells, however, is maintained as it is in normal cells. Hypoxia occurs in tumor tissue that is .100–200 mm away from a functional blood supply2. Thus, tumor survival depends, in part, on the ability to recruit new blood microvessels through angiogenic factors (Fig. 1b). Hypoxia tends to be widespread in solid tumors, however, because cancer cells are more prolific than the invading, recruited endothelial cells, which commonly form a new, disorganized blood supply. Human tumors endure profound hypoxia, which indicates that adaptation to hypoxic conditions is a crucial step in tumor progression. The anaerobic use of glucose as an energy source through glycolysis (Fig. 2a) is, therefore, a feature common to most solid tumors. A better understanding of cancers at the molecular level has provided insights into the causes of altered metabolism in oncogenesis. Various metabolic changes have been observed in tumors. Here, however, we emphasize changes in glucose metabolism and cellular responses to hypoxia, and discuss three specific areas: (1) physiological
35 Gray, M. W. (1995) in Molecular Biology of Plant Mitochondria (Sevings, C. S., III and Vasil, I. K., eds), pp. 635–659, Kluwer Academic Publishers BV 36 Olsen, G. J., Woese, C. R. and Overbeek, R. (1994) J. Bacteriol. 176, 1–6 37 Viale, A. and Arakaki, A. K. (1994) FEBS Lett. 341, 146–151 38 Andersson, S. G. E. et al. (1998) Nature 396, 133–143 39 Martin, W. and Schnarrenberger, C. (1997) Curr. Genet. 32, 1–18
responses used by tumor cells to adapt to hypoxia; (2) oncogenic changes that affect glucose metabolism; and (3) tumor metabolism and apoptosis.
Hypoxia in tumors and normal tissues: activation of genes that encode glycolytic enzymes and vascular endothelial growth factor Normal tissue displays an oxygen gradient across a distance of 400 mm from a blood supply. By contrast, in situ measurements of oxygen tension in human tumors and tumor xenografts revealed significant hypoxia: cells adjacent to capillaries displayed a mean oxygen concentration of 2%, and cells located 200 mm from the nearest capillary displayed a mean oxygen concentration of 0.2% (Ref. 2). The profoundly hostile environment selects for cells that are adapted to chronic hypoxia. In normal cells, a critical response to hypoxia is the induction of the hypoxiainducible transcription factor HIF-1, a basic–helix-loop-helix (bHLH) transcription factor that consists of two subunits, HIF-1a and HIF-1b (Ref. 3). HIF-1b is also known as the arylhydrocarbon-receptor nuclear translocator (ARNT)3. HIF-1 binds to the DNA sequence 59-RCGTG-39 and increases the expression of genes that encode glycolytic enzymes, including aldolase A, enolase 1, lactate dehydrogenase A (Fig. 2b), phosphofructokinase L, phosphoglycerate kinase 1 and pyruvate kinase M, as well as the vascular endothelial growth factor (VEGF) gene, which is important for angiogenesis4–8 (Fig. 3). In addition to alterations in oxygen tension, changes in glucose concentration also activate many glycolytic enzyme genes through the carbohydrate-response element (ChoRE; 59CACGTG-39), which matches the consensus binding-site sequences for MYC and HIF-1 (Fig. 2b)9,10. Studies of knockout mice have implicated HIF-1 and the HLH–leucine-zipper transcription factor USF2, which also binds to the 59CACGTG-39 sequence, in the regulation
0968 – 0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
PII: S0968-0004(98)01344-9
REVIEWS
TIBS 24 – FEBRUARY 1999 of glycolytic-enzyme gene expression by changes in the concentrations of cellular oxygen and glucose, respectively6–8,11. Activation of the HIF-1 or USF pathway might therefore mediate adaptive responses to hypoxia and hypoglycemia in cancer cells. Alternatively, the activation of oncogenes or loss of tumor suppressors by somatic mutation might lead directly to non-physiological alterations of cellular metabolism and provide a selective advantage in hostile metabolic environments. These two mechanisms are not mutually exclusive.
Oncogenes, tumor suppressors and aerobic glycolysis Meticulous studies by Warburg12 over seven decades ago demonstrated that the vast majority of human and animal tumors display a high rate of glycolysis under aerobic conditions (a phenomenon known as the Warburg effect). Although Warburg’s hypothesis that defective oxidative metabolism underlies this high rate of glycolysis is not supported by many studies, the original observation has been confirmed repeatedly13. Magnetic-resonance spectroscopy and positron-emission tomography (PET) studies with 2-[18F]fluoro-2-deoxy-Dglucose have consistently demonstrated that different clinical tumors show about an order of magnitude more glucose uptake in vivo than does normal tissue13. Furthermore, glucose uptake correlates with tumor aggressiveness and prognosis, and the expression of the glucose transporter GLUT1 is also increased in cancer cells14. Increases in glucose transport and type II hexokinase activity in cancer cells contribute to the increased flux of glucose through the cancer cells15. Type II hexokinase plays a role in initiation and maintenance of high rates of glucose catabolism in rapidly growing tumors. The enzyme converts glucose to glucose 6-phosphate, the initial phosphorylated intermediate of the glycolytic pathway. The gene that encodes type II hexokinase is amplified fivefold in a hepatoma cell line15. Its promoter contains a potential glucose-response element, as well as putative p53-responsive elements, and type-II-hexokinase expression is markedly decreased in HIF-1a-deficient embryonic stem (ES) cells6. A mutated p53 allele stimulates transcription of the type-II-hexokinase promoter, which suggests that mutant p53 plays a role in tumor metabolism (Fig. 3)16. The ability of HIF-1a to interact with and stabilize p53 protein suggests that p53 plays a
1
(a)
Lactate, H
O2, glucose, growth factors
150 µm
(b) Cell proliferation Activation of oncogenes MYC RAS BCR–ABL
Apoptosis Hypoxia
Promotion of cell survival BCL-2 BCR–ABL p53
HIF-1 Metabolic adaptation
Angiogenesis RAS pVHL p53 Metastasis
Figure 1 (a) Schematic cross section of a multicellular spheroid, showing an outer layer of living cells (shown in green) surrounding a core of dead cells (shown in brown). Gradients of metabolites are indicated. (b) Selected genetic alterations involved in tumor progression. Tumor suppressors are shown in red. Activation of several oncogenes results in deregulated cell proliferation, which is often linked to apoptosis when cells are deprived of growth factors. The outcome is an increased rate of cell death (black circles), which causes no net gain in tumor mass. The promotion of cell survival by genetic alterations that protect against apoptosis, such as activation of BCL-2 or loss of p53, results in a net gain in tumor mass. Diffusion limitations cause hypoxia and nutrient deprivation, and restrict the growth in tumor mass. Activation of hypoxia-inducible transcription factor 1 (HIF-1) by hypoxia or oncogenes induces metabolic adaptation and angiogenic factors, which result in the recruitment of new microvessels. Ultimately, cells migrate out of the tumor mass into the circulating blood and establish distant metastases, the major cause of cancer mortality.
direct role as a transcription factor in response to hypoxia, although p53 protein is induced by near anoxic conditions (,0.02% oxygen)17, whereas HIF-1a expression increases exponentially when oxygen levels fall below 5%. Additional studies will be required if we are to elucidate the details of the functional interactions between p53 and HIF-1a. Glucose usage in mutant Chinese hamster ovary (CHO) cells that lack lactate dehydrogenase A (LDH-A)18, an enzyme that is not considered to be rate limiting for glycolysis, is drastically diminished, compared with that in parental cells.
This indicates that alternative pathways of energy metabolism, such as the use of glutamine, exist in cancer cells. Cells transformed by the oncogenes v-SRC or activated H-RAS exhibit increased rates of aerobic glycolysis. Although cells that express v-SRC display increased expression of HIF-1 and its targets VEGF and enolase 1 (Ref. 19), in SRC-deficient cells hypoxiainducible expression of either VEGF or the gene that encodes GLUT1 does not differ from that in wild-type cells20 (Fig. 3). H-RAS also stimulates transcription of the gene that encodes VEGF in a
69
REVIEWS
TIBS 24 – FEBRUARY 1999
(a)
Glucose
NAD
NAD
ADP
NADH
ATP
Lactate
Pyruvate
O2
TCA cycle and oxidative phosphorylation
Mitochondrion
(b) v-SRC
MYC
RAS
HIF-1 MYC–MAX USF ChoRE
ACACGTGGGTTCCCGCACGTCCGC
LDH-A
HIF-1 Figure 2 (a) Glucose utilization through the glycolytic pathway and the Krebs (TCA) cycle. Normal cells utilize oxygen for efficient production of ATP by oxidative phosphorylation. When they are deprived of oxygen, pyruvate is not metabolized through the TCA cycle but is converted to lactate to replete NAD. Tumor cells rely on the less efficient glycolytic pathway to produce ATP, even in the presence of oxygen. (b) Transcription-factor-binding sites within the proximal promoter of the gene that encodes lactate dehydrogenase A (LDH-A). The E box (59-CACGTG-39) is the consensus core of the carbohydrate-response element (ChoRE) and overlaps with the consensus binding sites for hypoxia-inducible transcription factor 1 (HIF-1), MYC–MAX and USF. MYC and HIF-1 can bind the cis elements directly, whereas v-SRC and activated H-RAS enhance the activity of HIF-1 and other factors that bind to these elements and activate glycolysis.
manner dependent on the presence of an intact HIF-1-binding site21. Whether HIF-1 links activated RAS to expression of VEGF remains to be established. The MYC gene, which is frequently activated in human cancers, encodes a transcription factor that heterodimerizes with a partner protein, MAX, to bind to a 59-CACGTG-39 consensus sequence10. Using representational difference analysis, Lewis et al.22 identified genes that are differentially expressed in MYCtransformed Rat1a fibroblasts. The gene that encodes LDH-A is a MYC target22, and its expression is frequently in-
70
creased in human cancers23. LDH-A has been used as a marker of neoplastic transformation and is induced by hypoxia through the activity of HIF-14,5. During normal mitogenesis, MYC expression transiently increases in G1 phase, which might account for activation of glycolysis and an eightfold increase in LDH-A levels as primary lymphocytes enter into S phase24. Anaerobic glycolysis might protect DNA from damage by oxygen radicals that are produce by oxidative phosphorylation25. Studies of transgenic animals that overexpress MYC in the liver provided
additional evidence for the in vivo induction of glycolysis by MYC (Ref. 26). These animals have increased glycolyticenzyme activity in liver and overproduce lactic acid26. Stably transfected rodent fibroblasts that overexpress LDH-A alone, and those transformed by MYC, overproduce lactate; this suggests that overexpression of LDH-A is sufficient to induce the Warburg effect23. LDH-A overexpression is required for MYC-mediated transformation: decreasing the expression levels of the former reduces the soft-agar clonogenicity of MYC-transformed fibroblasts, MYCtransformed human lymphoblastoid cells, and Burkitt’s lymphoma cells. Likewise, CHO cells that lack LDH-A display a decreased soft-agar clonogenicity induced by activated RAS; they form small colonies in vitro and, in vivo, form tumors that, compared with parentalcell populations, have a large proportion of necrotic or apoptotic cells18. Intriguingly, inactivation of the von Hippel-Lindau tumor-suppressor protein (pVHL) is associated with the development of highly vascularized renal and central nervous system tumors. The pVHL protein has been implicated in post-transcriptional destabilization of hypoxia-inducible mRNAs under normoxic conditions27. In particular, reintroduction of pVHL into renal carcinoma cells that lack pVHL caused inhibition of the production of mRNAs that are regulated by hypoxia, such as VEGF and GLUT1 (Ref. 28; Fig. 3). These studies further emphasize the multitude of pathways that can be altered in cancers and cause a metabolic profile of aerobic glycolysis that is necessary for tumor progression (Table 1).
Tumor microenvironment and apoptosis Tumor angiogenesis is stimulated by hypoxia and hypoglycemia, which induce expression of VEGF among other angiogenic factors29. VEGF recruits new microvessels, which allow delivery of nutrients and expansion of the tumor mass (Fig. 1b). Prior to microvessel recruitment, a small tumor can remain dormant. This dormancy state is a consequence of the fact that the apoptosis rate equals that of mitosis. As new blood vessels are recruited, the rate of tumor apoptosis decreases, the rate of mitosis increases and the tumor rapidly grows in size. Hypoxia remains a strong selective force, however, because new microvessels are limited and disorganized, and oxygen-consumption rates tend to exceed the supply rate.
REVIEWS
TIBS 24 – FEBRUARY 1999 Although tumor cells adapt to the hypoxic and acidic microenvironment, spheroid clusters of avascular tumor cells invariably display a core of necrotic or apoptotic cells that separates from the 150-mm-thick shell of live cells (Fig. 1a)1. Many factors contribute to cell death under these conditions (see below). A major consequence of an elevated rate of glycolysis in tumor cells is that glucose carbons are converted primarily to lactate and are therefore no longer the major carbon source for aerobic respiration. In tumor cells that are able to consume limited amounts of oxygen, glutamine is the major oxidizable substrate that enters an abnormal truncated Krebs cycle. Reactive oxygen intermediates that arise from oxidative metabolism of glutamine in L929 fibrosarcoma cells appear to be required for tumor necrosis factor a (TNFa)-mediated apoptosis30. Depletion of glutamine, but not of glucose, from the culture medium desensitizes the L929 cells to TNFa cytotoxicity. Note that the anti-apoptotic, BCL-2 family of proteins, and hexokinase II, are implicated in the formation or regulation of the mitochondrial permeability transition (PT) pore complex, a multiple-conductance channel and regulator of mitochondrial Ca2+ and pH homeostasis among other activities31. Disruption of the mitochondrial inner membrane potential causes opening of the PT pore and the release of cytochrome C, which triggers the final cell-death pathway, which is mediated by caspases. The BCL-2 protein, which is found in the mitochondrial outer membrane, is thought to inhibit opening of the mitochondrial PT pore and thereby inhibit the release of cytochrome C, preventing apoptosis31. In a model of growth-factorwithdrawal-mediated apoptosis, expression of BCL-2 inhibits apoptosis of interleukin 3 (IL-3)-depleted hematopoietic B0 cells32. Upon withdrawal of IL-3, glycolysis ceases and the ATP concentration is maintained temporarily as cells begin to undergo apoptosis. BCL-2 expression delays apoptosis and confers resistance to ATP depletion, which is a strong apoptotic signal in the parental cells deprived of IL-3. These observations suggest that BCL-2 can cause a state of metabolic dormancy, perhaps by regulating mitochondrial homeostasis, and prevent apoptosis. Hypoxia is a potent stimulus of p53dependent and p53-independent G1 cellcycle checkpoints in non-transformed fibroblasts. In MYC–RAS transformed, confluent primary rat embryonic fibroblasts,
Glucose (external)
pVHL
GLUT1 GLUT3
Glucose (internal)
?
p53
HK1 HK2 Glucose 6-phosphate GPI Fructose 6-phosphate
v-SRC HIF-1 H-RAS ? ?
MYC
PFKL* Fructose 1,6-bisphosphate ALDA* Dihydroxyacetone phosphate TPI Glyceraldehyde 3-phosphate GAPDH 1,3-Bisphosphoglycerate PGK1 3-Phosphoglycerate PGM 2-Phosphoglycerate ENO1 Phosphoenolpyruvate PKM* Pyruvate LDHA Lactate
Figure 3 Transcriptional regulation of glycolysis and its modulation by oncogenes and tumor-suppressor genes. The transport of glucose across the cell membrane by glucose transporters (GLUT1 and GLUT3) and the subsequent catabolism of glucose by the glycolytic pathway are shown. Points of regulation by the tumor suppressors p53 and von Hippel-Lindau protein (pVHL) as well as the MYC oncoprotein and hypoxia-inducible transcription factor 1 (HIF-1) are also shown. Whether MYC regulates other glycolytic enzyme genes through known carbohydrate-response elements (which are present in the asterisked enzymes) remains to be studied. v-SRC induces HIF-1 expression, whereas the mechanism by which H-RAS stimulates glycolysis has not been determined. ALDA, aldolase A; ENO1, enolase 1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GPI, phosphoglucose isomerase; HK, hexokinase; LDHA, lactate dehydrogenase A; PFKL, phosphofructokinase L; PGK1, phosphoglycerate kinase 1; PGM, phosphoglycerate mutase; PKM, pyruvate kinase M; TPI, triose phosphate isomerase.
hypoxia induced an acidic environment and apoptosis in a p53-dependent manner. However, when extracellular pH was buffered to ~7.0, apoptosis decreased and the MYC–RAS transformed primary rat embryonic cells were able to bypass the normal G1–S phase hypoxia-induced checkpoint33. Likewise, hypoxia induces apoptosis in MYC-transformed cells in a p53-dependent manner, unless acidosis is prevented34. When the pH is buffered, a large fraction of the MYCtransformed cells can cycle in hypoxia, which suggests that acidosis is a potent effector of apoptosis under hypoxic conditions23,33. Like the gradient of oxygen tension (which diminishes toward the center of a tumor mass), the concentration of glucose also tapers drastically, which contributes to the triggering of cell death in the tumor core1. Glucose deprivation is
a particularly potent inducer of apoptosis in transformed cells that depend on glucose as a major source of energy. Glucose deprivation or treatment with 2-deoxyglucose, a glycolytic inhibitor, caused non-transformed cells to arrest in G1 phase, whereas MYC-transformed fibroblasts or lymphoblastoid cells underwent extensive apoptosis, which was blocked by elevated BCL-2 expression35. HIF-1 modulates gene expression in tumors and induces both angiogenesis and tumor growth36. Hypoxia induces the expression of certain genes, such as VEGF and phosphoglycerokinase 1, through HIF-1a, whereas the effects of glucose deprivation on gene expression might be HIF-1 independent6–8. Intriguingly, HIF-1a appears to influence the extent of apoptosis in tumors derived from ES cells injected into immunocompromised nude mice7.
71
REVIEWS
TIBS 24 – FEBRUARY 1999
Table 1. Putative effectors of the Warburg effect or aerobic glycolysis in cancer cells Molecule
Function
Activity in cancer
Effect
Refs
HIF-1
Hypoxia-inducible transcription factor
Adaptive or constitutive expression; gain of function
3, 4, 7, 15
MYC
Oncogenic transcription factor
Constitutive expression; gain of function
RAS v-SRC
Oncogenic GTP-binding protein Oncogenic non-receptor tyrosine kinase Tumor suppressor; transcription factor
Activation by mutations; gain of function Oncogenic retrovirally transduced gene; gain of function Mutated; loss of function
Tumor suppressor; mRNA stability modulator
Mutated; loss of function
Increases expression of genes encoding glycolytic enzymes, VEGF, and other proteins involved in hypoxic adaptation Increases expression of LDH-A; increases glycolysis and lactate production in transgenic mouse liver Increases glycolysis and VEGF expression Increases glycolysis; induces enolase and VEGF mRNA expression through HIF-1 Stabilized by HIF-1α; activates hexokinase II; induces apoptosis under hypoxic and acidic conditions Destabilizes hypoxia inducible transcripts: VEGF and GLUT1
p53
pVHL
6, 17, 18, 21
16 15 12, 13, 27, 28
22
Abbreviations used: HIF, hypoxia-inducible transcription factor; LDH-A, lactate dehydrogenase A; VEGF, vascular endothelial growth factor; pVHL, von HippelLindau tumor-suppressor protein.
Hypoxia and glucose deprivation reduced proliferation and increased apoptosis in wild-type ES cells but not in HIF-1a2/2 ES cells. Consequently, tumors derived from HIF-1a2/2 ES cells exhibit reduced VEGF expression, decreased vascularity and a more hypoxic microenvironment. Paradoxically, HIF-1a2/2 tumors are larger than wild-type tumors, presumably because of decreased apoptosis. This result contradicts those of Ryan et al.8, who showed that loss of HIF-1a in ES cells dramatically reduced VEGF expression and retarded tumor growth in RAG1-null immunocompromised mice. The difference in results is probably due to differences in the behavior of the ES cells or the host environment – for example, a variation in the availability of growth factors in nude mice versus RAG1-deficient mice. Both insulin and insulin-like growth factor 1 induce HIF-1 activity independently of hypoxia and elevate the expression of downstream target genes, such as VEGF and GLUT1 (Ref. 37). Although these studies illustrate the complexities involved in the development of a tumor, the role of HIF-1 in tumor growth and metabolism should soon emerge from additional studies of these genetically defined systems.
Conclusions and outlook Recent molecular studies of cancer have revealed that, in addition to the contributions of oncogenes and tumorsuppressor genes to the growth and apoptotic phenotypes of cancer cells, some of these genes directly affect cellular energy metabolism. The products of these genes alter the expression of transcription factors that regulate genes that encode metabolic enzymes and angiogenic factors. Because hypoxia is a key selective pressure on the progression of
72
cancer cells, physiological and nonphysiological (oncogenic) activation of HIF-1 and perhaps other transcription factors might play a central role in promoting the survival of cancer cells in adverse tumor microenvironments. Rekindling of interest in the classical biochemical pathways and their intersections with newly discovered signal transduction pathways will provide novel molecular insights into the alterations in metabolic profile that have long been known to occur in cancers. The frequency and severity of tumor hypoxia, and its association with malignant progression, suggest that therapeutic strategies designed to prevent metabolic adaptation would be particularly efficacious.
Acknowledgements We thank the NIH for support. We apologize to our colleagues for omission of references to many important original articles, which could not be cited owing to space limitations.
References 1 Sutherland, R. M. (1988) Science 240, 177–184 2 Helmlinger, G., Yuan, F., Dellian, M. and Jain, R. K. (1997) Nat. Med. 3, 177–182 3 Wang, G. L., Jiang, B-H., Rue, E. A. and Semenza, G. L. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 5510–5514 4 Firth, J. D., Ebert, B. L. and Ratcliffe, P. J. (1995) J. Biol. Chem. 270, 21021–21027 5 Semenza, G. L. et al. (1996) J. Biol. Chem. 271, 32529–32537 6 Iyer, N. V. et al. (1998) Genes Dev. 12, 149–162 7 Carmeliet, P. et al. (1998) Nature 394, 485–490 8 Ryan, H. E., Lo, J. and Johnson, R. S. (1998) EMBO J. 17, 3005–3015 9 Towle, H. C. (1995) J. Biol. Chem. 270, 23235–23238 10 Grandori, C. and Eisenman, R. N. (1997) Trends Biochem. Sci. 22, 177–181
11 Vallet, V. S. et al. (1998) J. Biol. Chem. 273, 20175–20179 12 Warburg, O. (1930) The Metabolism of Tumours, Constable 13 Gatenby, R. A. (1995) Cancer Res. 55, 4151–4156 14 Younes, M. et al. (1996) Cancer Res. 56, 1164–1167 15 Rempel, A. et al. (1996) Cancer Res. 56, 2468–2471 16 Mathupala, S. P., Heese, C. and Pedersen, P. L. (1997) J. Biol. Chem. 272, 22776–22780 17 An, W. G. et al. (1998) Nature 392, 405–408 18 Yamagata, M., Hasuda, K., Stamato, T. and Tannock, I. F. (1998) Br. J. Cancer 77, 1726–1731 19 Jiang, B. H., Agani, F., Passaniti, A. and Semenza, G. L. (1997) Cancer Res. 57, 5328–5335 20 Gleadle, J. M. and Ratcliffe, P. J. (1997) Blood 89, 503–509 21 Mazure, N. M. et al. (1996) Cancer Res. 56, 3436–3440 22 Lewis, B. C. et al. (1997) Mol. Cell. Biol. 17, 4967–4978 23 Shim, H. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 6658–6663 24 Greiner, E. F., Guppy, M. and Brand, K. (1994) J. Biol. Chem. 269, 31484–31490 25 Brand, K. A. and Hermfisse, U. (1997) FASEB J. 11, 388–395 26 Valera, A. et al. (1995) FASEB J. 9, 1067–1078 27 Lonergan, K. M. et al. (1998) Mol. Cell. Biol. 18, 732–741 28 Gnarra, J. R. et al. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 10589–10594 29 Hanahan, D. and Folkman, J. (1996) Cell 86, 353–364 30 Goossens, V., Grooten, J. and Fiers, W. (1996) J. Biol. Chem. 271, 192–196 31 Green, D. and Kroemer, G. (1998) Trends Cell Biol. 8, 267–271 32 Garland, J. M. and Halestrap, A. (1997) J. Biol. Chem. 272, 4680–4688 33 Schmaltz, C., Hardenbergh, P. H., Wells, A. and Fisher, D. E. (1998) Mol. Cell. Biol. 18, 2845–2854 34 Graeber, T. G. et al. (1996) Nature 379, 88–91 35 Shim, H., Chun, Y. S., Lewis, B. C. and Dang, C. V. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 1511–1516 36 Maxwell, P. H. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 8104–8109 37 Zelzer, E. et al. (1998) EMBO J. 17, 5085–5094
TIBS 24 – FEBRUARY 1999
Neuhaus, H. E. (1998) J. Biol. Chem. 273, 9630–9636 Miroux, B. and Walker, J. E. (1996) J. Mol. Biol. 260, 289–298 Krause, D. C., Winkler, H. H. and Wood, D. O. (1985) Proc. Natl. Acad. Sci. U. S. A. 82, 3015–3019 Dunbar, S. A. and Winkler, H. H. (1997) Microbiology 143, 3661–3669 Möhlmann, T., Scheibe, R. and Neuhaus, H. E. (1994) Planta 194, 492–497 Hatzfeld, W-D., Dancer, J. and Stitt, M. (1990)
24 25 26 27 28
Planta 180, 205–211 29 Huber, S. C. and Edwards, G. E. (1976) Biochim. Biophys. Acta 440, 675–687 30 Batz, O., Scheibe, R. and Neuhaus, H. E. (1995) Planta 196, 50–57 31 Cai, J. and Winkler, H. H. (1996) J. Bacteriol. 178, 5543–5545 32 Cai, J. and Winkler, H. H. (1997) Acta Virol. 41, 285–288 33 Martin, W. and Müller, M. (1998) Nature 392, 37–41 34 Gupta, R. S. (1995) Mol. Microbiol. 15, 1–11
Oncogenic alterations of metabolism Chi V. Dang and Gregg L. Semenza Over seven decades ago, classical biochemical studies showed that tumors have altered metabolic profiles and display high rates of glucose uptake and glycolysis. Although these metabolic changes are not the fundamental defects that cause cancer, they might confer a common advantage on many different types of cancers, which allows the cells to survive and invade. Recent molecular studies have revealed that several of the multiple genetic alterations that cause tumor development directly affect glycolysis, the cellular response to hypoxia and the ability of tumor cells to recruit new blood vessels. A GENETICALLY ALTERED neoplastic cell has special metabolic requirements for its development into a three-dimensional tumor mass. Monolayer cultures do not reflect the three-dimensional cellular growth of an avascular tumor, which can be mimicked in soft-agar anchorage-independent-growth assays. When a tumor has grown to a detectable size, the local environment of the cancer cells often becomes heterogeneous1. Small (,1 mm diameter) tumor nodules, as well as microregions of larger tumors, often have microecological niches that display significant gradients of critical metabolites such as oxygen, glucose and other nutrients or growth factors (Fig. 1a). Tumors, in contrast to normal tissue, exist in acidic environments that result from production C. V. Dang is at the Depts of Medicine, Oncology, Pathology, and Molecular Biology and Genetics, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA; and G. L. Semenza is at the Depts of Pediatrics and Medicine, and the Institute of Genetic Medicine, Johns Hopkins University School of Medicine, Baltimore, MD 21205, USA. Email: [email protected]
68
of lactate and other acids. The cytosolic pH of tumor cells, however, is maintained as it is in normal cells. Hypoxia occurs in tumor tissue that is .100–200 mm away from a functional blood supply2. Thus, tumor survival depends, in part, on the ability to recruit new blood microvessels through angiogenic factors (Fig. 1b). Hypoxia tends to be widespread in solid tumors, however, because cancer cells are more prolific than the invading, recruited endothelial cells, which commonly form a new, disorganized blood supply. Human tumors endure profound hypoxia, which indicates that adaptation to hypoxic conditions is a crucial step in tumor progression. The anaerobic use of glucose as an energy source through glycolysis (Fig. 2a) is, therefore, a feature common to most solid tumors. A better understanding of cancers at the molecular level has provided insights into the causes of altered metabolism in oncogenesis. Various metabolic changes have been observed in tumors. Here, however, we emphasize changes in glucose metabolism and cellular responses to hypoxia, and discuss three specific areas: (1) physiological
35 Gray, M. W. (1995) in Molecular Biology of Plant Mitochondria (Sevings, C. S., III and Vasil, I. K., eds), pp. 635–659, Kluwer Academic Publishers BV 36 Olsen, G. J., Woese, C. R. and Overbeek, R. (1994) J. Bacteriol. 176, 1–6 37 Viale, A. and Arakaki, A. K. (1994) FEBS Lett. 341, 146–151 38 Andersson, S. G. E. et al. (1998) Nature 396, 133–143 39 Martin, W. and Schnarrenberger, C. (1997) Curr. Genet. 32, 1–18
responses used by tumor cells to adapt to hypoxia; (2) oncogenic changes that affect glucose metabolism; and (3) tumor metabolism and apoptosis.
Hypoxia in tumors and normal tissues: activation of genes that encode glycolytic enzymes and vascular endothelial growth factor Normal tissue displays an oxygen gradient across a distance of 400 mm from a blood supply. By contrast, in situ measurements of oxygen tension in human tumors and tumor xenografts revealed significant hypoxia: cells adjacent to capillaries displayed a mean oxygen concentration of 2%, and cells located 200 mm from the nearest capillary displayed a mean oxygen concentration of 0.2% (Ref. 2). The profoundly hostile environment selects for cells that are adapted to chronic hypoxia. In normal cells, a critical response to hypoxia is the induction of the hypoxiainducible transcription factor HIF-1, a basic–helix-loop-helix (bHLH) transcription factor that consists of two subunits, HIF-1a and HIF-1b (Ref. 3). HIF-1b is also known as the arylhydrocarbon-receptor nuclear translocator (ARNT)3. HIF-1 binds to the DNA sequence 59-RCGTG-39 and increases the expression of genes that encode glycolytic enzymes, including aldolase A, enolase 1, lactate dehydrogenase A (Fig. 2b), phosphofructokinase L, phosphoglycerate kinase 1 and pyruvate kinase M, as well as the vascular endothelial growth factor (VEGF) gene, which is important for angiogenesis4–8 (Fig. 3). In addition to alterations in oxygen tension, changes in glucose concentration also activate many glycolytic enzyme genes through the carbohydrate-response element (ChoRE; 59CACGTG-39), which matches the consensus binding-site sequences for MYC and HIF-1 (Fig. 2b)9,10. Studies of knockout mice have implicated HIF-1 and the HLH–leucine-zipper transcription factor USF2, which also binds to the 59CACGTG-39 sequence, in the regulation
0968 – 0004/99/$ – See front matter © 1999, Elsevier Science. All rights reserved.
PII: S0968-0004(98)01344-9
REVIEWS
TIBS 24 – FEBRUARY 1999 of glycolytic-enzyme gene expression by changes in the concentrations of cellular oxygen and glucose, respectively6–8,11. Activation of the HIF-1 or USF pathway might therefore mediate adaptive responses to hypoxia and hypoglycemia in cancer cells. Alternatively, the activation of oncogenes or loss of tumor suppressors by somatic mutation might lead directly to non-physiological alterations of cellular metabolism and provide a selective advantage in hostile metabolic environments. These two mechanisms are not mutually exclusive.
Oncogenes, tumor suppressors and aerobic glycolysis Meticulous studies by Warburg12 over seven decades ago demonstrated that the vast majority of human and animal tumors display a high rate of glycolysis under aerobic conditions (a phenomenon known as the Warburg effect). Although Warburg’s hypothesis that defective oxidative metabolism underlies this high rate of glycolysis is not supported by many studies, the original observation has been confirmed repeatedly13. Magnetic-resonance spectroscopy and positron-emission tomography (PET) studies with 2-[18F]fluoro-2-deoxy-Dglucose have consistently demonstrated that different clinical tumors show about an order of magnitude more glucose uptake in vivo than does normal tissue13. Furthermore, glucose uptake correlates with tumor aggressiveness and prognosis, and the expression of the glucose transporter GLUT1 is also increased in cancer cells14. Increases in glucose transport and type II hexokinase activity in cancer cells contribute to the increased flux of glucose through the cancer cells15. Type II hexokinase plays a role in initiation and maintenance of high rates of glucose catabolism in rapidly growing tumors. The enzyme converts glucose to glucose 6-phosphate, the initial phosphorylated intermediate of the glycolytic pathway. The gene that encodes type II hexokinase is amplified fivefold in a hepatoma cell line15. Its promoter contains a potential glucose-response element, as well as putative p53-responsive elements, and type-II-hexokinase expression is markedly decreased in HIF-1a-deficient embryonic stem (ES) cells6. A mutated p53 allele stimulates transcription of the type-II-hexokinase promoter, which suggests that mutant p53 plays a role in tumor metabolism (Fig. 3)16. The ability of HIF-1a to interact with and stabilize p53 protein suggests that p53 plays a
1
(a)
Lactate, H
O2, glucose, growth factors
150 µm
(b) Cell proliferation Activation of oncogenes MYC RAS BCR–ABL
Apoptosis Hypoxia
Promotion of cell survival BCL-2 BCR–ABL p53
HIF-1 Metabolic adaptation
Angiogenesis RAS pVHL p53 Metastasis
Figure 1 (a) Schematic cross section of a multicellular spheroid, showing an outer layer of living cells (shown in green) surrounding a core of dead cells (shown in brown). Gradients of metabolites are indicated. (b) Selected genetic alterations involved in tumor progression. Tumor suppressors are shown in red. Activation of several oncogenes results in deregulated cell proliferation, which is often linked to apoptosis when cells are deprived of growth factors. The outcome is an increased rate of cell death (black circles), which causes no net gain in tumor mass. The promotion of cell survival by genetic alterations that protect against apoptosis, such as activation of BCL-2 or loss of p53, results in a net gain in tumor mass. Diffusion limitations cause hypoxia and nutrient deprivation, and restrict the growth in tumor mass. Activation of hypoxia-inducible transcription factor 1 (HIF-1) by hypoxia or oncogenes induces metabolic adaptation and angiogenic factors, which result in the recruitment of new microvessels. Ultimately, cells migrate out of the tumor mass into the circulating blood and establish distant metastases, the major cause of cancer mortality.
direct role as a transcription factor in response to hypoxia, although p53 protein is induced by near anoxic conditions (,0.02% oxygen)17, whereas HIF-1a expression increases exponentially when oxygen levels fall below 5%. Additional studies will be required if we are to elucidate the details of the functional interactions between p53 and HIF-1a. Glucose usage in mutant Chinese hamster ovary (CHO) cells that lack lactate dehydrogenase A (LDH-A)18, an enzyme that is not considered to be rate limiting for glycolysis, is drastically diminished, compared with that in parental cells.
This indicates that alternative pathways of energy metabolism, such as the use of glutamine, exist in cancer cells. Cells transformed by the oncogenes v-SRC or activated H-RAS exhibit increased rates of aerobic glycolysis. Although cells that express v-SRC display increased expression of HIF-1 and its targets VEGF and enolase 1 (Ref. 19), in SRC-deficient cells hypoxiainducible expression of either VEGF or the gene that encodes GLUT1 does not differ from that in wild-type cells20 (Fig. 3). H-RAS also stimulates transcription of the gene that encodes VEGF in a
69
REVIEWS
TIBS 24 – FEBRUARY 1999
(a)
Glucose
NAD
NAD
ADP
NADH
ATP
Lactate
Pyruvate
O2
TCA cycle and oxidative phosphorylation
Mitochondrion
(b) v-SRC
MYC
RAS
HIF-1 MYC–MAX USF ChoRE
ACACGTGGGTTCCCGCACGTCCGC
LDH-A
HIF-1 Figure 2 (a) Glucose utilization through the glycolytic pathway and the Krebs (TCA) cycle. Normal cells utilize oxygen for efficient production of ATP by oxidative phosphorylation. When they are deprived of oxygen, pyruvate is not metabolized through the TCA cycle but is converted to lactate to replete NAD. Tumor cells rely on the less efficient glycolytic pathway to produce ATP, even in the presence of oxygen. (b) Transcription-factor-binding sites within the proximal promoter of the gene that encodes lactate dehydrogenase A (LDH-A). The E box (59-CACGTG-39) is the consensus core of the carbohydrate-response element (ChoRE) and overlaps with the consensus binding sites for hypoxia-inducible transcription factor 1 (HIF-1), MYC–MAX and USF. MYC and HIF-1 can bind the cis elements directly, whereas v-SRC and activated H-RAS enhance the activity of HIF-1 and other factors that bind to these elements and activate glycolysis.
manner dependent on the presence of an intact HIF-1-binding site21. Whether HIF-1 links activated RAS to expression of VEGF remains to be established. The MYC gene, which is frequently activated in human cancers, encodes a transcription factor that heterodimerizes with a partner protein, MAX, to bind to a 59-CACGTG-39 consensus sequence10. Using representational difference analysis, Lewis et al.22 identified genes that are differentially expressed in MYCtransformed Rat1a fibroblasts. The gene that encodes LDH-A is a MYC target22, and its expression is frequently in-
70
creased in human cancers23. LDH-A has been used as a marker of neoplastic transformation and is induced by hypoxia through the activity of HIF-14,5. During normal mitogenesis, MYC expression transiently increases in G1 phase, which might account for activation of glycolysis and an eightfold increase in LDH-A levels as primary lymphocytes enter into S phase24. Anaerobic glycolysis might protect DNA from damage by oxygen radicals that are produce by oxidative phosphorylation25. Studies of transgenic animals that overexpress MYC in the liver provided
additional evidence for the in vivo induction of glycolysis by MYC (Ref. 26). These animals have increased glycolyticenzyme activity in liver and overproduce lactic acid26. Stably transfected rodent fibroblasts that overexpress LDH-A alone, and those transformed by MYC, overproduce lactate; this suggests that overexpression of LDH-A is sufficient to induce the Warburg effect23. LDH-A overexpression is required for MYC-mediated transformation: decreasing the expression levels of the former reduces the soft-agar clonogenicity of MYC-transformed fibroblasts, MYCtransformed human lymphoblastoid cells, and Burkitt’s lymphoma cells. Likewise, CHO cells that lack LDH-A display a decreased soft-agar clonogenicity induced by activated RAS; they form small colonies in vitro and, in vivo, form tumors that, compared with parentalcell populations, have a large proportion of necrotic or apoptotic cells18. Intriguingly, inactivation of the von Hippel-Lindau tumor-suppressor protein (pVHL) is associated with the development of highly vascularized renal and central nervous system tumors. The pVHL protein has been implicated in post-transcriptional destabilization of hypoxia-inducible mRNAs under normoxic conditions27. In particular, reintroduction of pVHL into renal carcinoma cells that lack pVHL caused inhibition of the production of mRNAs that are regulated by hypoxia, such as VEGF and GLUT1 (Ref. 28; Fig. 3). These studies further emphasize the multitude of pathways that can be altered in cancers and cause a metabolic profile of aerobic glycolysis that is necessary for tumor progression (Table 1).
Tumor microenvironment and apoptosis Tumor angiogenesis is stimulated by hypoxia and hypoglycemia, which induce expression of VEGF among other angiogenic factors29. VEGF recruits new microvessels, which allow delivery of nutrients and expansion of the tumor mass (Fig. 1b). Prior to microvessel recruitment, a small tumor can remain dormant. This dormancy state is a consequence of the fact that the apoptosis rate equals that of mitosis. As new blood vessels are recruited, the rate of tumor apoptosis decreases, the rate of mitosis increases and the tumor rapidly grows in size. Hypoxia remains a strong selective force, however, because new microvessels are limited and disorganized, and oxygen-consumption rates tend to exceed the supply rate.
REVIEWS
TIBS 24 – FEBRUARY 1999 Although tumor cells adapt to the hypoxic and acidic microenvironment, spheroid clusters of avascular tumor cells invariably display a core of necrotic or apoptotic cells that separates from the 150-mm-thick shell of live cells (Fig. 1a)1. Many factors contribute to cell death under these conditions (see below). A major consequence of an elevated rate of glycolysis in tumor cells is that glucose carbons are converted primarily to lactate and are therefore no longer the major carbon source for aerobic respiration. In tumor cells that are able to consume limited amounts of oxygen, glutamine is the major oxidizable substrate that enters an abnormal truncated Krebs cycle. Reactive oxygen intermediates that arise from oxidative metabolism of glutamine in L929 fibrosarcoma cells appear to be required for tumor necrosis factor a (TNFa)-mediated apoptosis30. Depletion of glutamine, but not of glucose, from the culture medium desensitizes the L929 cells to TNFa cytotoxicity. Note that the anti-apoptotic, BCL-2 family of proteins, and hexokinase II, are implicated in the formation or regulation of the mitochondrial permeability transition (PT) pore complex, a multiple-conductance channel and regulator of mitochondrial Ca2+ and pH homeostasis among other activities31. Disruption of the mitochondrial inner membrane potential causes opening of the PT pore and the release of cytochrome C, which triggers the final cell-death pathway, which is mediated by caspases. The BCL-2 protein, which is found in the mitochondrial outer membrane, is thought to inhibit opening of the mitochondrial PT pore and thereby inhibit the release of cytochrome C, preventing apoptosis31. In a model of growth-factorwithdrawal-mediated apoptosis, expression of BCL-2 inhibits apoptosis of interleukin 3 (IL-3)-depleted hematopoietic B0 cells32. Upon withdrawal of IL-3, glycolysis ceases and the ATP concentration is maintained temporarily as cells begin to undergo apoptosis. BCL-2 expression delays apoptosis and confers resistance to ATP depletion, which is a strong apoptotic signal in the parental cells deprived of IL-3. These observations suggest that BCL-2 can cause a state of metabolic dormancy, perhaps by regulating mitochondrial homeostasis, and prevent apoptosis. Hypoxia is a potent stimulus of p53dependent and p53-independent G1 cellcycle checkpoints in non-transformed fibroblasts. In MYC–RAS transformed, confluent primary rat embryonic fibroblasts,
Glucose (external)
pVHL
GLUT1 GLUT3
Glucose (internal)
?
p53
HK1 HK2 Glucose 6-phosphate GPI Fructose 6-phosphate
v-SRC HIF-1 H-RAS ? ?
MYC
PFKL* Fructose 1,6-bisphosphate ALDA* Dihydroxyacetone phosphate TPI Glyceraldehyde 3-phosphate GAPDH 1,3-Bisphosphoglycerate PGK1 3-Phosphoglycerate PGM 2-Phosphoglycerate ENO1 Phosphoenolpyruvate PKM* Pyruvate LDHA Lactate
Figure 3 Transcriptional regulation of glycolysis and its modulation by oncogenes and tumor-suppressor genes. The transport of glucose across the cell membrane by glucose transporters (GLUT1 and GLUT3) and the subsequent catabolism of glucose by the glycolytic pathway are shown. Points of regulation by the tumor suppressors p53 and von Hippel-Lindau protein (pVHL) as well as the MYC oncoprotein and hypoxia-inducible transcription factor 1 (HIF-1) are also shown. Whether MYC regulates other glycolytic enzyme genes through known carbohydrate-response elements (which are present in the asterisked enzymes) remains to be studied. v-SRC induces HIF-1 expression, whereas the mechanism by which H-RAS stimulates glycolysis has not been determined. ALDA, aldolase A; ENO1, enolase 1; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; GPI, phosphoglucose isomerase; HK, hexokinase; LDHA, lactate dehydrogenase A; PFKL, phosphofructokinase L; PGK1, phosphoglycerate kinase 1; PGM, phosphoglycerate mutase; PKM, pyruvate kinase M; TPI, triose phosphate isomerase.
hypoxia induced an acidic environment and apoptosis in a p53-dependent manner. However, when extracellular pH was buffered to ~7.0, apoptosis decreased and the MYC–RAS transformed primary rat embryonic cells were able to bypass the normal G1–S phase hypoxia-induced checkpoint33. Likewise, hypoxia induces apoptosis in MYC-transformed cells in a p53-dependent manner, unless acidosis is prevented34. When the pH is buffered, a large fraction of the MYCtransformed cells can cycle in hypoxia, which suggests that acidosis is a potent effector of apoptosis under hypoxic conditions23,33. Like the gradient of oxygen tension (which diminishes toward the center of a tumor mass), the concentration of glucose also tapers drastically, which contributes to the triggering of cell death in the tumor core1. Glucose deprivation is
a particularly potent inducer of apoptosis in transformed cells that depend on glucose as a major source of energy. Glucose deprivation or treatment with 2-deoxyglucose, a glycolytic inhibitor, caused non-transformed cells to arrest in G1 phase, whereas MYC-transformed fibroblasts or lymphoblastoid cells underwent extensive apoptosis, which was blocked by elevated BCL-2 expression35. HIF-1 modulates gene expression in tumors and induces both angiogenesis and tumor growth36. Hypoxia induces the expression of certain genes, such as VEGF and phosphoglycerokinase 1, through HIF-1a, whereas the effects of glucose deprivation on gene expression might be HIF-1 independent6–8. Intriguingly, HIF-1a appears to influence the extent of apoptosis in tumors derived from ES cells injected into immunocompromised nude mice7.
71
REVIEWS
TIBS 24 – FEBRUARY 1999
Table 1. Putative effectors of the Warburg effect or aerobic glycolysis in cancer cells Molecule
Function
Activity in cancer
Effect
Refs
HIF-1
Hypoxia-inducible transcription factor
Adaptive or constitutive expression; gain of function
3, 4, 7, 15
MYC
Oncogenic transcription factor
Constitutive expression; gain of function
RAS v-SRC
Oncogenic GTP-binding protein Oncogenic non-receptor tyrosine kinase Tumor suppressor; transcription factor
Activation by mutations; gain of function Oncogenic retrovirally transduced gene; gain of function Mutated; loss of function
Tumor suppressor; mRNA stability modulator
Mutated; loss of function
Increases expression of genes encoding glycolytic enzymes, VEGF, and other proteins involved in hypoxic adaptation Increases expression of LDH-A; increases glycolysis and lactate production in transgenic mouse liver Increases glycolysis and VEGF expression Increases glycolysis; induces enolase and VEGF mRNA expression through HIF-1 Stabilized by HIF-1α; activates hexokinase II; induces apoptosis under hypoxic and acidic conditions Destabilizes hypoxia inducible transcripts: VEGF and GLUT1
p53
pVHL
6, 17, 18, 21
16 15 12, 13, 27, 28
22
Abbreviations used: HIF, hypoxia-inducible transcription factor; LDH-A, lactate dehydrogenase A; VEGF, vascular endothelial growth factor; pVHL, von HippelLindau tumor-suppressor protein.
Hypoxia and glucose deprivation reduced proliferation and increased apoptosis in wild-type ES cells but not in HIF-1a2/2 ES cells. Consequently, tumors derived from HIF-1a2/2 ES cells exhibit reduced VEGF expression, decreased vascularity and a more hypoxic microenvironment. Paradoxically, HIF-1a2/2 tumors are larger than wild-type tumors, presumably because of decreased apoptosis. This result contradicts those of Ryan et al.8, who showed that loss of HIF-1a in ES cells dramatically reduced VEGF expression and retarded tumor growth in RAG1-null immunocompromised mice. The difference in results is probably due to differences in the behavior of the ES cells or the host environment – for example, a variation in the availability of growth factors in nude mice versus RAG1-deficient mice. Both insulin and insulin-like growth factor 1 induce HIF-1 activity independently of hypoxia and elevate the expression of downstream target genes, such as VEGF and GLUT1 (Ref. 37). Although these studies illustrate the complexities involved in the development of a tumor, the role of HIF-1 in tumor growth and metabolism should soon emerge from additional studies of these genetically defined systems.
Conclusions and outlook Recent molecular studies of cancer have revealed that, in addition to the contributions of oncogenes and tumorsuppressor genes to the growth and apoptotic phenotypes of cancer cells, some of these genes directly affect cellular energy metabolism. The products of these genes alter the expression of transcription factors that regulate genes that encode metabolic enzymes and angiogenic factors. Because hypoxia is a key selective pressure on the progression of
72
cancer cells, physiological and nonphysiological (oncogenic) activation of HIF-1 and perhaps other transcription factors might play a central role in promoting the survival of cancer cells in adverse tumor microenvironments. Rekindling of interest in the classical biochemical pathways and their intersections with newly discovered signal transduction pathways will provide novel molecular insights into the alterations in metabolic profile that have long been known to occur in cancers. The frequency and severity of tumor hypoxia, and its association with malignant progression, suggest that therapeutic strategies designed to prevent metabolic adaptation would be particularly efficacious.
Acknowledgements We thank the NIH for support. We apologize to our colleagues for omission of references to many important original articles, which could not be cited owing to space limitations.
References 1 Sutherland, R. M. (1988) Science 240, 177–184 2 Helmlinger, G., Yuan, F., Dellian, M. and Jain, R. K. (1997) Nat. Med. 3, 177–182 3 Wang, G. L., Jiang, B-H., Rue, E. A. and Semenza, G. L. (1995) Proc. Natl. Acad. Sci. U. S. A. 92, 5510–5514 4 Firth, J. D., Ebert, B. L. and Ratcliffe, P. J. (1995) J. Biol. Chem. 270, 21021–21027 5 Semenza, G. L. et al. (1996) J. Biol. Chem. 271, 32529–32537 6 Iyer, N. V. et al. (1998) Genes Dev. 12, 149–162 7 Carmeliet, P. et al. (1998) Nature 394, 485–490 8 Ryan, H. E., Lo, J. and Johnson, R. S. (1998) EMBO J. 17, 3005–3015 9 Towle, H. C. (1995) J. Biol. Chem. 270, 23235–23238 10 Grandori, C. and Eisenman, R. N. (1997) Trends Biochem. Sci. 22, 177–181
11 Vallet, V. S. et al. (1998) J. Biol. Chem. 273, 20175–20179 12 Warburg, O. (1930) The Metabolism of Tumours, Constable 13 Gatenby, R. A. (1995) Cancer Res. 55, 4151–4156 14 Younes, M. et al. (1996) Cancer Res. 56, 1164–1167 15 Rempel, A. et al. (1996) Cancer Res. 56, 2468–2471 16 Mathupala, S. P., Heese, C. and Pedersen, P. L. (1997) J. Biol. Chem. 272, 22776–22780 17 An, W. G. et al. (1998) Nature 392, 405–408 18 Yamagata, M., Hasuda, K., Stamato, T. and Tannock, I. F. (1998) Br. J. Cancer 77, 1726–1731 19 Jiang, B. H., Agani, F., Passaniti, A. and Semenza, G. L. (1997) Cancer Res. 57, 5328–5335 20 Gleadle, J. M. and Ratcliffe, P. J. (1997) Blood 89, 503–509 21 Mazure, N. M. et al. (1996) Cancer Res. 56, 3436–3440 22 Lewis, B. C. et al. (1997) Mol. Cell. Biol. 17, 4967–4978 23 Shim, H. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 6658–6663 24 Greiner, E. F., Guppy, M. and Brand, K. (1994) J. Biol. Chem. 269, 31484–31490 25 Brand, K. A. and Hermfisse, U. (1997) FASEB J. 11, 388–395 26 Valera, A. et al. (1995) FASEB J. 9, 1067–1078 27 Lonergan, K. M. et al. (1998) Mol. Cell. Biol. 18, 732–741 28 Gnarra, J. R. et al. (1996) Proc. Natl. Acad. Sci. U. S. A. 93, 10589–10594 29 Hanahan, D. and Folkman, J. (1996) Cell 86, 353–364 30 Goossens, V., Grooten, J. and Fiers, W. (1996) J. Biol. Chem. 271, 192–196 31 Green, D. and Kroemer, G. (1998) Trends Cell Biol. 8, 267–271 32 Garland, J. M. and Halestrap, A. (1997) J. Biol. Chem. 272, 4680–4688 33 Schmaltz, C., Hardenbergh, P. H., Wells, A. and Fisher, D. E. (1998) Mol. Cell. Biol. 18, 2845–2854 34 Graeber, T. G. et al. (1996) Nature 379, 88–91 35 Shim, H., Chun, Y. S., Lewis, B. C. and Dang, C. V. (1998) Proc. Natl. Acad. Sci. U. S. A. 95, 1511–1516 36 Maxwell, P. H. et al. (1997) Proc. Natl. Acad. Sci. U. S. A. 94, 8104–8109 37 Zelzer, E. et al. (1998) EMBO J. 17, 5085–5094