Exploiting Metabolic Vulnerabilities of Cancer with Precision and Accuracy

Abstract

The links between oncogenic drivers and cancer cell metabolism have emerged over the past several decades, indicating that constitutive oncogenic growth signaling can render cancers susceptible to metabolic interventions. While significant progress has been achieved in identifying metabolic vulnerabilities of cancer cells, the complexity of the tumor microenvironment (TME) and the dynamic nature of organismal circadian metabolism challenge the precision of targeting cancer metabolism. Herein, current progress in the areas of cancer metabolism and TME metabolism is reviewed, highlighting how cancer metabolism can be accurately and precisely targeted.

Introduction

The “Precision Medicine” revolution in cancer has focused on the identification of compounds or biologic agents with high levels of molecular specificity and the tailored use of these agents in subsets of patients whose tumors have unique molecular characteristics. The use of such “precise” therapies can protect from off-target side effects, but not from on-target side effects. Such on-target effects can impact remote tissues, but also non-cancerous cells within the tumor microenvironment. The complex interplay between stromal elements, immune cells, and blood vessels can drastically impact the response to therapy and the development of resistance. Hence, through on-target effects on non-cancerous tissues both within and outside of the tumor microenvironment, “precise” therapies can fail to “accurately” treat a patient's cancer.

This issue is particularly notable with metabolic therapies that target pathways that are ubiquitous in both malignant and normal host cells. In this review, we will summarize alteration in tumor and TME metabolism, current therapies targeting these alterations, and strategies for improving the accuracy of metabolic therapies.

Basic concepts in tumor metabolism

Simplicity for the sake of understanding cancer metabolism has provided key conceptual frameworks over the past decades. The first of these frameworks is that cancer cell metabolism is altered to support biomass accumulation rather than efficient energy production. This was built on the observations of Louis Pasteur and Otto Warburg. In the 1880s, Pasteur showed that oxygen could suppress glycolysis or fermentation in yeast. Warburg then demonstrated in the 1920s that animal and human cancer tissues continued to display high glycolytic rates, even in the presence of oxygen[1, 2]. The former phenomenon is called the Pasteur effect, while the latter is coined the Warburg effect. Warburg's initial interpretation, that tumor mitochondria were dysfunctional, turned out to be incorrect. Rather, oxidative metabolism is held at sub-maximal levels so that alternative pathways may use glucose carbon to support the metabolic demands of constitutive cell growth and division.

The second major framework in tumor metabolism, based on work in the 1990s, is that the metabolic changes are driven by changes in oncogenes and tumor suppressors, as well as the hypoxia inducible factors (HIFs), which rewire metabolic pathways [3-6]. Human oncogenes, such as MYC, PI3K, and RAS, drive neoplastic cell proliferation by stimulating cancer cell metabolism through transcriptional, translational, and post-translational alterations of metabolic enzymes, hence promoting biosynthesis [6].

The constitutive oncogenic drive for cancer cells to grow and proliferate renders cancer cells dependent on a continuous nutrient supply to meet the constant demand of uncontrolled proliferative metabolism, which is distinct from normal maintenance metabolism that occurs in the vast majority of non-proliferative adult mammalian cells and provides a therapeutic window [7]. This window is limited by three major factors, 1) the complexity and heterogeneity of tumor metabolism itself, the cellular complexity of the TME and the role of metabolic pathways in the non-cancerous cells in the TME and in interactions between the tumor and the TME, and the use of tumor-like metabolism by a small subset of normal tissues, largely stem cells.

Complexity of tumor metabolism

The central metabolic pathways are conserved and evolved over billions of years. Specifically, glycolysis is the most conserved pathway that converts glucose to pyruvate (Figure 1), which is further catabolized to lactate, alanine, alcohol, or transported into the mitochondrion depending on the metabolic wiring of the cell [5, 6]. The conversion of glucose to lactate via glycolysis and lactate dehydrogenase A (LDHA) is known as the Warburg effect. The mitochondrial pyruvate carrier transports pyruvate into the mitochondrion, where it is converted by pyruvate dehydrogenase (PDH) to acetyl-CoA for further catabolism through the tricarboxylic acid (TCA) cycle. Hypoxia, through HIF-1, reverses the Pasteur effect and activates glycolytic genes as well as activating pyruvate dehydrogenase kinase to inhibit the conversion of pyruvate to acetyl-CoA while favoring its conversion to lactate via LDHA.

Figure 1.

Overview of Cellular Metabolic Pathways, Central metabolic pathways and their connections are outlined, including glycolysis, the TCA cycle, nucleic acid synthesis, lipid synthesis, and the urea cycle. Cancer cells utilize these pathways to varying extents depending on the genetics and tissue of origin of the tumor, as discussed in the main text. Glucose and glutamine can both converted into substrates that can be oxidized via the TCA cycle, but intermediates of these pathways can be diverted to provide substrates for nucleic acid and amino acid synthesis or to replenish TCA cycle intermediates. Glucose can also be converted to lactate and exported (the Warburg Effect). Alternatively, lactate can also be taken up and oxidized as fuel.

There are exit and entry points into the mitochondrion and the TCA cycle that are involved in catabolism or biosynthesis [5]. Citrate produced from glucose in the mitochondrion (or glutamine through reductive carboxylation) is transported into the cytosol and converted by acetyl-CoA lyase (ACLY) to acetyl-CoA, providing a precursor for fatty acid synthesis or acetylation reactions. Succinyl-CoA, downstream of citrate, provides a precursor for heme biosynthesis. Oxaloacetate is converted by glutamate oxaloacetate transaminase (GOT2) to aspartate, which is exported into the cytosol for nucleotide biosynthesis. Glutamine is converted by glutaminase to ammonia and glutamate, which is further transformed to α-ketoglutarate for catabolism in the TCA cycle [8]. Under specific conditions, such as hypoxia or a truncated TCA cycle (eg., fumarate hydratase (FH) mutations in certain cancers), α-ketoglutarate can undergo reductive carboxylation to produce citrate in support of fatty acid synthesis [9]. Branched chain amino acids (BCAA; Leu, Ile, Val) are converted to ketoacids and ultimately to TCA cycle intermediates for catabolism [10]. Lipids transported into mitochondria can be catabolized through fatty acid oxidation (FAO) [11]. Ketone bodies (acetoacetate, β-hydroxybutyrate, and acetone) can also be converted to acetyl-CoA and catabolized by the mitochondrion. Further, pyruvate can be converted by pyruvate carboxylase (PC) to oxaloacetate to enter the TCA cycle, in addition to its conversion to acetyl-CoA by PDH [12].

These pathways are rewired in cancer cells under specific conditions. For example, studies with human lung cancers reveal the role of PC in the use of glucose [13], whereas BCAA are utilized in a genetically engineered mouse model (GEMM) of pancreatic cancer [10], FH mutant tumors undergo glutamine reductive carboxylation [14], and human liver cancers can catabolize ketone bodies [15]. It should also be noted that cancer cells can also turn on themselves and digest their own constituents via autophagy to generate metabolic intermediates for survival, they can eat albumin from the environment through macropinocytosis, and they can even eat other nearby cells through entosis [16, 17].

While engaging the mitochondria to catabolize nutrients is essential for the survival of cancer cells, the generation of byproducts can also be toxic and expose additional vulnerabilities such as the induction of autophagy by ammonia, the inhibition of NOS by nitric oxide, and oxidative stress generated by reactive oxygen species [18]. Cancer cells can therefore require increased activation of anti-oxidant defense mechanisms. One example of this is the NRF2 transcriptional axis, which is activated through the ROS-induced inactivation of KEAP to release active NRF2, which enters the nucleus and transactivates genes involved in redox homeostasis to reduce oxidants [19, 20]. In a fraction of human lung cancers, KEAP1 is mutated, allowing for the constitutive activity of NRF2, which increases the anti-oxidant programs for cell survival. Another example is the increased synthesis and use of long chain polyunsaturated fatty acids, particularly in high mesenchymal cell state tumors, which leads to increased dependence on glutathione and the lipid peroxidase GPX4 to avoid ferroptosis [21]. Hence, such rewiring exposes new vulnerabilities to oxidant stress in these tumors.

While different metabolic pathways have been shown to be important in different tumor subtypes, there can also be heterogeneity in metabolic characteristics within an individual tumor. Certain regions of a tumor are more hypoxic than others that are more proximal to blood vessels [22, 23]. In fact, it is documented that in some settings hypoxic cancer cells generate lactate that is converted to pyruvate for mitochondrial catabolism in more oxygenated cancer cells [24, 25]. This was further supported by a recent study demonstrating that poorly perfused regions of human non-small cell lung tumors predominantly fuel the TCA cycle with glucose, while well-perfused regions preferentially use non-glucose alternatives[26].

The work summarized above has allowed for a deep understanding of cancer cell intrinsic metabolic changes. However, with a richer appreciation for the cellular complexity of the tumor microenvironment (TME), this understanding is insufficient for a strategy to target tumor metabolism, noting that targets can be precisely hit without accuracy in the context of a complex TME that changes spatially and temporally.

Metabolism in the TME

The tumor tissue is highly complex and variable depending on the tissue of origin [27] (Figure 2). For example, while leukemias and liver cancers tend to be composed primarily of malignant cells, cancers involving the pancreas display highly variable composition with remarkable heterogeneity of the stromal versus cancer cellular content. In addition to the genomically altered cancer cell, the tumor microenvironment can contain stromal fibroblasts, glial cells, macrophages, myeloid derived tumor suppressor cells, and tumor tolerant T and B cells, which are all fed by neo-vasculature and drained by highly variable lymphatics. These tumor-associated cells operate within the same metabolic milieu as the cancer and can be involved in tumor formation, maintenance, and therapy response and resistance. As such, the understanding of the complexity of the community of cells that comprises the tumor tissue is essential for accurate and precise targeting of tumor metabolism.

Figure 2.

Metabolism in the Tumor Microenvironment, In addition to tumor cells, the tumor microenvironment is composed of fibroblasts, macrophages, myeloid-derived suppressor cells, regulatory and cytotoxic T cells, and endothelial cells, among others. These cells depend on specific metabolic substrates for appropriate function, which can be disrupted in the TME or with therapy. Cancer cells can reprogram these cells to produce fuels such as lactate, alanine, and fatty acids that are then consumed by the cancer cells themselves. They can also produce immunosuppressive metabolites, including lactate, adenosine, and kynurenine.

Stromal cells can undergo metabolic changes in response to signals from the tumor cells as well as in response to the hypoxic and acidic microenvironment [28]. These metabolic changes can then feed-back to support tumor growth. For examples, pancreatic stellate cells can produce alanine to support the metabolism of pancreatic cancer cells [29]. Prostate cancer-associated fibroblasts can produce exosomes that supply amino acids to tumor cells and reprogram tumor metabolism away from oxidative energy production [30]. Tumor derived hydrogen peroxide can induce cancer-associated fibroblasts to consume glucose into lactate, which can then be used as an oxidative metabolite by the cancer cells [31]. Ovarian cancer cells have been shown to promote lipolysis in adipocytes and transfer of free fatty acids to tumor cells for use in beta-oxidation [32]. More broadly, the community of cancer and stromal cells co-exists in an abnormal metabolic environment that selects for synergistic commensal interactions, which manifest as clinically detectable cancers.

The tumor neo-vasculature is required for oxygen and nutrient delivery, for removal of waste products, and for delivery of chemotherapeutic agents. Endothelial cells are thus essential to the metabolism of the tumor, but they also have their own unique metabolic features [33]. Endothelial cells in general are highly glycolytic, even in the presence of oxygen, reminiscent of the Warburg effect [34]. There are also roles for glutamine and lipids in endothelial metabolism [33]. Tumor endothelial cells are abnormal and have even higher rates of glycolysis than non-tumor endothelial cells, and blocking glycolysis can lead to normalization of the tumor vasculature and improved delivery of chemotherapeutic agents [35]. In addition to improving drug delivery, normalization of tumor blood vessels can alleviate hypoxia, which reverse tumor metabolic changes and can alleviate selective pressure for tumor evolution and metastasis [36]. In addition to hypoxia, lactate produced from tumors can signal to endothelial cells to promote angiogenesis [37], thus meeting the metabolic demands of the tumor.

The dynamism of the immune system further complicates the TME through the contributions of macrophages, myeloid derived cells, and lymphocytes, which have metabolic profiles and needs that change in different states of activation [38, 39]. T cells provide an instructive example. Quiescent T cells oxidize glucose and fatty acids to produce ATP, but then switch to aerobic glycolysis upon activation. Memory T cells return to a resting state-like metabolic program, but have increased mitochondrial mass and spare respiratory capacity, which provides a proliferative advantage upon re-stimulation with antigen [40]. In a tumor, the metabolic dynamics of T cell activation and persistence are then overlaid with the complexity of tumor cell metabolism and the TME. While these immune metabolism/tumor metabolism interactions pose challenges, they also provide an opportunity to identify cancer-cell dependent pathways that produce immunosuppressive metabolites such as lactate, adenosine, and kynurenine [41].

The challenge in metabolic anti-tumor therapy is therefore not only to identify and collect metabolic inhibitors with appropriate drug properties, but to apply them with an understanding of their effect on the tumor, the TME, and the interaction between them in order to achieve both therapeutic precision and accuracy.

Cancer metabolic inhibitors and drugs

It is notable that the first rationally designed anti-cancer drug, developed by Sidney Farber and colleagues in the 1940's, targeted 1C metabolism by blocking the folate pathway [42]. This success in targeting cancer metabolism triggered studies using 2-deoxyglucose in human studies in the 1950's, which were met with side effects [43, 44]. Interest in targeting cancer metabolism waned and was replaced by a focus on kinases. However, the resurgence of interest in cancer metabolism has revived metabolic inhibitors and led to their use in preclinical studies, and in some cases, clinical trials. For example, metformin, which inhibits mitochondrial Complex I and is used clinically to treat diabetes, and hydroxychloroquine, which inhibits autophagy and is used clinically to treat malaria as well as rheumatic diseases, have been exhumed and placed into cancer clinical trials (clinicaltrials.gov). Increased interest in cancer metabolism has also driven the identification of novel chemical entities that target metabolic pathways for cancer therapy.

Because many, but not all, human cancers display avid uptake of radio-labeled 2-deoxyglucose as determined by positron emission tomography, it stands to reason that targeting glucose metabolism could be of therapeutic importance. Compounds targeting glucose metabolism include inhibitors of glucose import, of glucose phosphorylation by hexokinase 2 (HK2), of conversion of pyruvate into lactate, and of export of lactate out of the cell [45-48]. Pathways involving NAD+ synthesis, glutamine metabolism, fatty acid synthesis and mitochondrial function have been targeted by small molecules [24, 48-52]. Table 1 summarizes metabolic enzymes and inhibitors that have been studied in in vitro or in pre-clinical or clinic settings.

Table 1. Cancer metabolic inhibitors tested in selected studies.

Protein	Compound	Study	Finding	Ref
GLUT1	STF-31	pre-clinical	Xenograft activity	[46]
Xct (SLC7A11)	Sulfasalazine	pre-clinical		[89]
MCT1 (SLC16A1)	AZD3965	clinical	ongoing	[53] ClinicalTrials.gov Identifier: NCT01791595
HK2	Substituted glucosamine	In vitro		[90]
GAPDH	3-bromo pyruvate	pre-clinical in vivo		[64]
PKM2	Activator: TEPP-46	pre-clinical in vivo	Xenograft activity	[57]
LDHA	GNE-140	In vitro	nM, poor PK	[91]
LDHA	2-((3-cyanopyridin-2-yl)thio)acetamides	In vitro	nM, poor PK	[92]
LDHA	Quinoline-3-sulfonamides	In vitro		[93]
LDHA	Pyrazole-based	In vitro	nM, cell active	[94]
LDHA	Fragment-based		nM in vitro	[95]
LDHA	FX-11	pre-clinical in vivo	off-target effects	[47]
NAMPT	APO-866	clinical	Phase I: thrombo-cytopenia	[96]
GLS	BPTES	Pre-clinical	Survival GEMM model liver cancer, kidney	[58, 97]
GLS	CB-839	clinical	Phase I/II Solid tumor Plus nivolumab	[98] ClinicalTrials.gov Identifier: NCT02771626
FASN	TVB-2640	clinical	PD study resectable colon cancer	ClinicalTrials.gov Identifier: NCT02980029
ACC	ND-646	Pre-clinical	GEMM model lung cancer	[99]
Mitochondrial Complex I	phenformin	clinical	Phase I combination in melanoma	ClinicalTrials.gov Identifier: NCT03026517
Mitochondrial Complex 1	metformin	clinical	Interventional with standard therapy; metastatic breast cancer	ClinicalTrials.gov Identifier: NCT01310231
PDH, OGDH	CPI-613	clinical	Phase I Metastatic pancreas cancer; CRs	[100]
IDH1	AG-120	clinical	Interventional plus azacytidine; AML	ClinicalTrials.gov Identifier: NCT02677922
IDH2	Enasidenib (IDHIFA)	clinical	FDA approved 2017; AML	[101]

Notes: GLUT1, glucose transporter; Xct amino acid antiporter; MCT1, monocarboxylate transporter 1; HK2, hexokinase 2; GAPDH, glyceraldehyde-3-phosphate dehydrogenase; PKM2, pyruvate kinase M2; LDH, lactate dehydrogenase; NAMPT, nicotinamide phosphoribosyltransferase; GLS, glutaminase; FASN, fatty acid synthase; ACC, acetyl-CoA carboxylase; PDH, pyruvate dehydrogenase; OGDH, oxoglutarate dehydrogenase; IDH, isocitrate dehydrogenase; nM, nanomolar IC50; PK, pharmacokinetics; GEMM, genetically engineered mouse model; CRs, complete remissions; AML, acute myelogenous leukemia; FDA, Food and Drug Administration.

A number of lessons have been learned from studies summarized in Table 1 as well as other related studies. First is that there is a heterogeneity of responses to specific inhibitors depending on the genetics and tissue origin of the model tested, as well as combination therapies used. For example, an inhibitor of the lactate transporter MCT1 appears to have variable responses in different models, with MCT1 high/MCT4 low tumor responding better [53, 54]. The concomitant use of complex I inhibitors (metformin or phenformin), rewire cells toward a further dependency on glycolysis and sensitize to the effects of MCT1 inhibition [55, 56]. Activators of PKM2 also seems variable depending on whether specific tumors rely heavily on PKM2 for survival, and the role of PKM2 in tumorigenesis appears much more complicated than originally surmised [57]. Glutaminase inhibition in preclinical models appears to depend on specific models and on potential synergies with other inhibitors. For example, while glutaminase (Gls) genetic loss or pharmacological inhibition in a MYC-inducible mouse liver cancer model seems effective in delaying tumorigenesis or prolonging survival, inhibition of glutaminase does not seem to affect lung or pancreas cancer in GEMM models [58-60]. However, under oxidative stress, GLS inhibition appears more effective in pancreas or a subtype of lung cancer that depends on the anti-oxidant activity of activated Nrf2 [61]. The study by Yuneva et al. was prescient in pointing out that both the genetic basis and tissue origin of a cancer could determine its metabolic re-wiring in vivo [62]. Specifically, the Yuneva study documented that in mouse liver cancer, MET oncogene-driven tumors were glycolytic while MYC oncogene-driven tumors were glutaminolytic [62]. This observation suggested that MET oncogene-driven liver cancers would be more sensitive to glycolytic but not glutaminolytic inhibitors. Moreover, they went on to show that a MYC oncogene driven lung cancer model, contrary to the MYC-driven liver cancer, was able to use both glucose and glutamine, but these tumors could also generate glutamine through glutamine synthetase, which is induced in lung tissue.

The second lesson learned from Table I and related studies is that in vitro activity does not portend in vivo activity due to off-target effects, suboptimal pharmacokinetic properties of the compounds in question, or profound differences between in vitro versus in vivo tumor metabolism. For example, many chemical entities have been generated against LDHA and some are quite effective in in vitro assays but are ineffective in vivo due to poor pharmacodynamic properties. Other tool compounds against LDHA, such as FX11, are effective in vivo, but have off-target effects at high concentrations and hence their anti-tumorigenic activity may not be due entirely to LDHA inhibition [47, 63]. In addition to inhibiting GAPDH, 3-bromopyruvate is a highly active alkylating agent that is likely to have pleiotropic effects [64]. Exploitation of increased transporter (MCT1) expression in tumor cells, however, could increase 3-bromopyruvate uptake and enhance target cellular accuracy [65]. Cancer dependency on specific metabolic pathways, and hence sensitivity to pathway inhibition, can be very context-dependent. Since even alteration in composition of culture media can have significant effects on cellular metabolite levels [66], it is perhaps not surprising that there can be profound differences between in vitro and in vivo metabolic dependencies. In non-small cell lung cancer, for example, cultured cells have high glutaminolytic flux and are sensitive to glutaminase inhibitors. However, in vivo tumors utilize less glutamine and are insensitive to glutaminase inhibition [59]. The other consideration, given lessons learned from LDHA, is the importance of cellular concentrations of target enzymes. In this regard, it is intriguing to note from global cellular proteomics studies of NIH3T3 fibroblasts and human Hela cells that the concentrations of glycolytic enzymes vary dramatically by orders of magnitude from HK2 to LDHA [67, 68]. Copies of glycolytic enzymes involved in the catabolism of hexoses tend to be low (∼50,000-200,000 copies/cell) relative to the much higher (50 to over 100 million copies/cell) protein copy numbers per cell of glycolytic enzymes involved in the metabolism of trioses, such as LDHA, estimated to have a concentration of up to 16 μM. Hence, it may be more fruitful to focus on enzymes with low copy number such as HK2, particularly since HK2 is induced in proliferating cells, while HK1 tends to be involved in maintenance metabolism of non-proliferative cells.

The third lesson learned is that metabolic synthetic essentiality can be exploited to enhance tumor selectivity. As alluded to above, oxidative stress can synergize with glutaminase inhibition in a lung cancer model that depends on Nrf2 [61] and inhibitors of oxidative phosphorylation can increase dependence on glycolysis and sensitivity to MCT1 inhibition [45]. mTOR inhibition can also synergize with glutaminase inhibition [69, 70]. The re-wiring of metabolism by targeted kinase inhibition can lead to new vulnerabilities, such as sensitivity to phenformin after BRAF inhibition in melanoma [71]. Loss of isoforms of metabolic genes adjacent to tumor suppressor genes can drive sensitivity to inhibition or loss of the remaining isoforms, a concept dubbed “collateral lethality” [72, 73]. The combination of glycolytic, glutaminolytic and oxidative metabolic inhibitors could hypothetically uncover new vulnerabilities that are likely tumor type specific. The effects of these combinations on non-cancer cells in the TME should be consider for full understanding of their effects in vivo.

The fourth lesson, one still at early stages, is the potential for metabolic inhibitors to affect elements of the TME and therefore antagonize or synergize therapeutic response. Metabolic inhibitors can alter the tumor vasculature, either by changing tumor endothelial cell metabolism directly or by altering metabolic signals in the TME sensed by endothelial cells, thus altering angiogenesis, metabolite and oxygen supply, and therapeutic access [33, 35]. The success of checkpoint inhibition and chimeric antigen receptor T cell therapy in specific clinical settings has put particular focus on metabolic interactions with the anti-tumor function of cytolytic T cells, tumor suppressive macrophages, or myeloid and B suppressor cells [38, 41, 74, 75]. As described above, T cells have specific metabolic requirements at different stages of activation and persistence. The use of metabolic therapies may alter this process when used in combination with immunotherapies. Intriguingly and further demonstrating the complexity of these interactions, one study suggests that phenformin can inhibit myeloid-derived suppressor cells and increase the anti-tumor effect of PD-1 blockade in melanoma [74]. How glycolytic and glutaminase inhibitors alter T cell function or fates will be important areas to explore, particularly given the observation that mTOR inhibition could skew T cell differentiation away from the tumor permissive Treg lymphocytes toward more cytolytic T cells. [76, 77].

Concluding Remarks

While significant advances have been achieved in understanding cancer metabolism and developing metabolic therapies, several issues remain to be addressed (see Outstanding Questions Box). We anticipate that a richer understanding of tumor, organismal, and TME metabolism will allow for application of such therapies with both precision and accuracy.

Outstanding Questions.

1. How can metabolic enzymes be targeted precisely with potent inhibitors without causing an immune-suppressive microenvironment?

2. Could some metabolic drugs promote a tumor-permissive microenvironment and therefore lose efficacy, despite diminishing tumor cell intrinsic growth?

3. What are the synthetic essentialities with specific metabolic drugs that could be exploited to improve safety and efficacy?

4. Could new technologies be developed to assess or visualize the effects of metabolic drugs on components of the tumor microenvironment in vivo?

With regard to tumor intrinsic metabolism, our knowledge of synthetic essentialities between metabolic pathways as well as with other pathways is still rudimentary. It is notable that passenger genomic deletions, such as loss of ENO1 or ME2 found in specific cancers could uncover new druggable targets with the remaining isoenzymes, ENO2 and ME3, respectively [73, 78]. Additional studies to uncover synthetic lethal vulnerabilities will help tremendously in streamlining experimental combination metabolic therapies. It is also expected that blockage of key metabolic pathways would trigger other survival mechanisms such as autophagy and macropinocytosis. As such, better definition of the interplay of these cellular pathways will be essential. Lastly, better definition is needed of links between the genomic alterations of human cancers and their metabolic phenotypic manifestations (by metabolic nuclear medicine imaging approaches).

Improvements in the use of genomic data to define the TME composition [79] are needed to take advantage of this expanding resource. In addition, the use of advanced technologies to directly define the tumor immune microenvironment by flow cytometry will greatly enhance to strategic use of metabolic drugs. Beyond T cells, better understanding of how metabolic inhibitors influence macrophages, myeloid suppressor, and B suppressor cells will aid in determining responses versus resistance due to cell extrinsic mechanisms in immune-competent hosts.

The proliferation of committed normal tissue stem cells in less than 1% of adult mammalian cells relies on biosynthetic pathways similar to those used by cancer cells [80]. However, the normal stem cell cycle oscillator tends to couple with the circadian clock to orchestrate tissue repair and regeneration in synchrony with the solar cycle of fasting and feeding coupled with sleep and wake periods that are linked to cellular metabolism [81, 82]. These normal pathways are under tight control of feed-back loops that return cells to a resting state when nutrients and growth signals are insufficient. Emerging evidence has shown oncogenic disruption of the clock in cancer cells [83-85]. This presents the possibility that metabolic toxicity to normal tissues could be spared and efficacy against cancer cells increased by taking the circadian clock into consideration. For example, a large clinical trial showed that the effectiveness of 5-FU in colorectal cancer depends both on the time of treatment and gender of the patient [86, 87]. Further studies are needed to determine the role of the circadian clock in controlling the metabolism of cells in the TME. Such refinements based on time of administration could allow for the more accurate use of targeted therapies, improving therapeutic outcomes [88].

Trends Box.

• Oncogenes rewire cellular metabolism to meet the energetic and substrate demands of the tumor, but this rewiring also create new opportunities for therapy

• Significant advances have been made in understanding cancer cell intrinsic metabolism, the metabolism of non-tumor cells in the tumor microenvironment, and the metabolic interactions between them.

• Rising interest in the metabolic vulnerabilities of cancer has led to the development of novel therapies targeting diverse aspects of nutrient transport and utilization

• Application of metabolic therapies has been hindered by context-dependentvariations in substrate utilization and by effects on normal cells both within and outsideof the tumor microenvironment.