Introduction: Tumor cells, just as other living cells, possess the potential for proliferation, differentiation, cell cycle arrest, and apoptosis. There is a specific metabolic phenotype associated with each of these conditions, characterized by the production of both energy and special substrates necessary for the cells to function in that particular state. Unlike that of normal living cells, the metabolic phenotype of tumor cells supports the proliferative state. Aim: To present the metabolic hypothesis that (1) cell transformation and tumor growth are associated with the activation of metabolic enzymes that increase glucose carbon utilization for nucleic acid synthesis, while enzymes of the lipid and amino acid synthesis pathways are activated in tumor growth inhibition, and (2) phosphorylation and allosteric and transcriptional regulation of intermediary metabolic enzymes and their substrate availability together mediate and sustain cell transformation from one condition to another. Conclusion: Evidence is presented that demonstrates opposite changes in metabolic phenotypes induced by TGF-
, a celltransforming agent, and tumor growth-inhibiting phytochemicals such as genistein and Avemar, or novel synthetic antileukemic drugs such as STI571 (Gleevec). Intermediary metabolic enzymes that mediate the growth signaling pathways and promote malignant cell transformation may serve as highefficacy nongenetic novel targets for cancer therapies.

Decades of intensive research have identified many possible mechanisms for the development of common human malignancies including tumor-inducing genes, environmental factors, and signal transduction pathways. The genetic alterations reported to date in pancreatic cancer include frequent mutations of the K-ras, p53, p16, and Smad4 genes, which were reported to be associated with accelerated disease progression and poor prognosis (1,2). Genetic abnormalities that influence cellular responses to hormonal growth regulators and their signaling pathways have been reported in connection with other major tumor types as well (3–8). The general understanding of malignant cell behavior currently emphasizes strong genetic regulation of human cell functions. It has been well established through molecular genetic studies that there is a great variance in the expression of human genes in response to environmental changes, nutrition, lifestyle, and age. The continued genetic characterization of human cancer reveals an ever more complicated architecture of potential cancer promoter and suppressor genes that show great variability and are extremely difficult to modulate. The diverse mechanisms of tumor induction have in common their resultant influence on metabolism-altering normal potential for differentiation, cell cycle arrest, and apoptosis (Fig. 1). 

In spite of their great genetic potential to express different phenotypes corresponding to the different degree of differentiation, tumor cells are uniformly characterized by a specific high glucose-utilizing metabolic phenotype and poor differentiation (9–11). Tumor cells use glucose primarily for intracellular anabolic processes, mainly the synthesis of nucleotides, whereas other anabolic reactions such as lipid and protein synthesis pathways are depleted of glucose carbons (12). It is our overall hypothesis that the effects of growth signals and genetic factors that transform human cells from the resting state to the proliferating one are mediated by metabolic phenotype changes aimed toward supplying the rapidly dividing cells with an optimal source of carbons at the expense of cell differentiation. In this article, we review the evidence for the hypothesis, delineating the conversion of growth signals into changes in metabolic pathway activity during the growth and death of pancreatic, lung, and epithelial malignancies. CHARACTERIZATION OF METABOLIC PHENOTYPES IN TUMOR CELLS A well-known characteristic of malignant cells is their ability to compete for glucose about 10 to 50 times more intensely than the surrounding normal tissue (9–11).

 This metabolic characteristic of tumors is used when the radioactive analog glucose tracer 18fluoro-deoxy-glucose is detected by positron emission tomography (FDG-PET) for the diagnosis and classification of human malignancies. It has been demonstrated using PET that the increased rate of glucose accumulation in cancer cells strongly correlates with increased malignancy and invasiveness (13–16). Proliferating cells are highly dependent on de novo synthesis of purines and pyrimidines as well as ribose for nucleic acid synthesis. Glucose is a particularly important substrate from which tumor cells produce ribose, the backbone of RNA and DNA molecules. Before the advent of stable isotope and mass spectrometry, the use of glucose in tumor cells had been studied using enzymology methods. Simple studies of glucose uptake and oxidation can also be done using radioactive tracers. The advantage of the stable isotope and mass spectrometry technique is in its ability to determine the quantity and position of label incorporation into biomolecules that are synthesized during cell cycle progression and proliferation by normal and tumor cells. Such capability was first demonstrated in the study of the glycogen synthesis pathway using uniformly labeled U13C-glucose in experimental animals (17). In consequent studies, the analysis of label incorporation into glutamate for the purpose of sampling -keto-glutarate pools provided the basis for the study of pyruvate dehydrogenase (PDH) and pyruvate carboxylase activity as well as the anaplerotic flux of the TCA cycle (18). 

Anaplerosis refers to the reactions that allow the entry of carbon into the TCA cycle intermediate pools other than via citrate synthase. Any carbon that enters the cycle as acetyl-CoA is oxidized to carbon dioxide and water; any carbon that enters the citric acid cycle via an anaplerotic pathway is not oxidized, but must be disposed of by some other route. Glutamate dehydrogenase is one possible route providing equilibrium between alpha-ketoglutarate and glutamate; some other reactions include pyruvate carboxylation, transamination reactions, and propionate carboxylation. Taking advantage of the knowledge of mass isotopomer distribution of a biomolecule, fatty acid synthesis can directly be studied using 1, 2–13C-acetate or deuterated water (19). We recently have developed a method of using [1, 2-13C2]glucose for metabolic phenotyping studies that provides unique information on the carbon flow through the oxidative and nonoxidative branches of the pentose cycle by its labeling pattern in ribose. [1, 2-13C2]glucose symmetrically labels oxaloacetate when it is carboxylated via pyruvate carboxylase and carbon 2 and 3 of -keto-glutarate of the TCA cycle. When it is converted to acetyl-CoA by pyruvate dehydrogenase, it labels carbon position 4 and 5 of -ketoglutarate for the study of glucose oxidation and generates a mass isotopomer distribution pattern in fatty acid for the determination of the new fraction of lipid synthesis as well as the 13C enrichment of acetate in pancreatic adenocarcinoma cells (20). Future studies using labeled fatty acids, glycerol, or ribose can be developed to further characterize the role of these substrates in different metabolic phenotypes. 

Such studies can reveal metabolic activity and the identification of structural and regulatory macromolecules that are synthesized using primarily glucose carbons in human cells (21,22). Extensive reviews and methodologic reports published in specialty journals have demonstrated the usefulness of the stable isotope method in combination with biologic mass spectrometry in mammalian cell metabolic studies to reveal glucose-derived specific synthesis pathways of nucleotides, lipids, and amino acids associated with tumor proliferation, differentiation, or apoptosis (23–25). The stable isotope approach in combination with biologic mass spectrometry complements molecular genetic approaches, which measure the expression, transcription, and activation of metabolic enzymes. Although molecular genetic studies provide information on the regulatory action of genes, hormones, and signal transduction pathways, nutrient and substrate availability ultimately influence substrate distribution in tumor cells and determine cell functions (Table 1). The most measurable significance of directly characterizing substrate distribution in tumor cells is the ability to determine the metabolic phenotype characteristic of tumor cell proliferation and to characterize the overall metabolic effect of tumor growth-regulating agents.METABOLIC PHENOTYPE CHANGES IN RESPONSE TO TUMOR GROWTH-PROMOTING AND INHIBITING SIGNALS Many potential growth-modulating factors have been identified and characterized through their signal transduction pathways. 

They all fall into two major signal types: one, such as steroid hormones, that acts on intracellular receptors and influences gene expression and the other, such as transforming growth factors, that acts on cell surface receptors and influences multiple enzyme activities by protein phosphorylation. We have previously characterized the overall metabolic effect of transforming growth factor (TGF-
) on lung epithelial carcinoma cells, which uses the tyrosine kinase signaling mechanism through the cell surface TGF-
receptor family and promotes the invasive transformation of various human cells (26). The accumulation of glucose carbons in nucleic acid ribose demonstrated a significant, dosedependent increase in response to this growth factor (Fig. 2). Concomitant metabolic changes in response to TGF-
treatment included decreased glucose oxidation in the pentose and TCA cycles, indicating that invasive cell transformation is accompanied by nonoxidative metabolic changes and increased glucose utilization toward anabolic metabolic reactions of nucleotides (27). This increase in the nonoxidative metabolism of glucose in the pentose cycle provided an explanation at the molecular level for the principal metabolic disturbance observed in human tumors: increased glucose uptake with increased glucose utilization for nucleic acid synthesis and decreased glucose oxidation. These metabolic changes also explain how tumor cells are capable of dividing rapidly in the hypoxic environment. Recent evidence has indicated that genistein, the natural tumor growth-regulating agent found in soy bean, decreases glucose uptake and glucose carbon incorporation into nucleic acid ribose in MIA pancreatic adenocarcinoma cells (Fig. 3) (20).

 Genistein has marked tyrosine kinase- and protein kinase-inhibiting properties (28,29), resulting in cell cycle arrest (30) and limiting angiogenesis (31) in several tumor models. The opposite changes in glucose carbon deposition into nucleic acid ribose, lactate, glutamate, and fatty acids after treating tumor cells with growth-promoting (27) or growthinhibiting agents (20) indicate that glucose carbon redistribution between major metabolic pathways plays a critical role in the cell proliferation process. METABOLIC HYPOTHESIS OF TUMOR GROWTH Tumor cells, like other living cells, possess the potential for proliferation, differentiation, cell cycle arrest, and apoptosis. Associated with each of these conditions is a specific distribution of substrates between macromolecule synthesis pathways according to the metabolic needs of the given cell. Macromolecule synthesis pathways directly determine proliferating, differentiating, cell cycle arrest, and apoptotic phenotypes, which are characterized by different needs for the production of energy and substrates necessary for cells to function under different pathophysiologic conditions. It is clear from previous work on pancreatic, lung, and epithelial malignancies that invasive transformation is associated with characteristic metabolic changes. The typical metabolic phenotypes related to tumor cell formation and death are summarized in Table 2. Cell transforming agents, such as TGF-
and organophosphate pesticides, induce a severe imbalance in glucose carbon redistribution between cell proliferation-related and cell differentiation-related structural and regulatory macromolecules. 

This phenotype is characterized by increased glucose utilization specifically for nucleic acid synthesis through the nonoxidative branch of the pentose cycle with a concomitant decrease in glucose oxidation and the synthesis of glutamate, palmitate, and stearate directly from glucose. Conversely, agents that inhibit tumor growth such as genistein and the fermented wheat germ extract FWGE alter glucose utilization from a pattern typical of the proliferative phenotype to one of differentiated or apoptotic cells. Such evidence suggests the hypothesis that (1) cell transformation and tumor growth are associated with the activation of metabolic enzymes that increase glucose carbon utilization toward nucleic acid synthesis, whereas lipid and amino acid synthesis pathway enzymes are activated during tumor growth inhibition. (2) phosphorylation, allosteric regulation, and transcriptional regulation of intermediary metabolic enzymes and their substrates together mediate and sustain cell transformation from one condition to another. The effects of pathophysiologic and genetic factors that transform tumor cells from one functional state to another are mediated by metabolic phenotype changes aimed toward supplying the dividing cell with an optimal source of carbons at the expense of cell differentiation.Figure 1 demonstrates the relation between extrinsic tumor growth-influencing factors and metabolic enzymes that are targets of growth-regulating signals. 

Tumor growth-modulating factors in the form of cell surface hormones bind cell-transforming factors that activate specific signal transducer pathways, resulting in activity changes of metabolic enzymes through either protein phosphorylation or genetic regulation. Superimposed on this background, glucose, intracellular glucose metabolites, and various nutrients directly affect cell function in providing the substrates for intracellular synthetic reactions and energy production and as key allosteric regulators of metabolic enzymes. The metabolic characteristics of four distinctive phenotypes, namely proliferation, differentiation, cell cycle arrest, and apoptosis, that have been described in tumor cells in response to tumor growth-modulating treatments are summarized in Table 2. Tumor cells adapt to a high rate of glucose utilization and macromolecule synthesis. These processes, in turn, become highly dependent on the availability of glucose carbons and increased activity of glycolytic, pentose, and TCA cycle enzymes. As a feedback mechanism, the production of intermediary metabolites also regulates gene expression through intracellular nuclear receptors. Our proposed model is based on the direct determination of substrate flow through various metabolic pathways that control the oncogenic process under the influence of various signaling and growth regulatory events. In support of our model, there are strong interactions between newly discovered signal transduction pathways and fundamental metabolic pathways such as glycolysis and the pentose cycle. Glucose deprivation is capable of inducing apoptosis of tumor cells on its own, even when other nutrients are plentiful (32,33). 

The c-myc oncogene directly regulates glucose transporter 1 and glycolytic gene expression in several tumor cells (34). Our hypothesis emphasizes that there is an apparent functional hierarchy within growth signaling, the translation of genes into proteins and enzymes that regulate various key metabolic pathways for macromolecule synthesis and energy production. The flow of information from the exterior of cells using specific signal transducer pathways and the flow of substrates to sustain these signaling events are key elements in the regulation of metabolism. These events also determine the optimum level of metabolic enzymes, which ultimately represent key regulatory check posts of carbon redistribution between major metabolic pathways. Metabolic enzymes control substrate flow according to the physiologic needs of a given cell as determined by its metabolic phenotype. The direct and indirect interactions between signal transducer pathways, substrates, and their intracellular target enzymes allow a fine regulatory mechanism in a functional hierarchy of genes and proteins, which together control cell events such as cell cycle progression, transformation, proliferation, hormone or enzyme secretion, differentiation, apoptosis, and growth. Accordingly, multiple genetic alterations and signaling pathways that cause tumor development directly affect glycolysis (35), the cellular response to hypoxia (27), and the ability of tumor cells to recruit new blood vessels (36).