Back to the Future: Therapeutic Targeting of Cancer Cell Metabolism

In simple terms, cancer can be defined as the uncontrolled growth and division of cells, leading to tumor formation, invasion, and metastases. Unlike normal cells that require growth factor signals, tumor cells often have mutations that result in constitutively active (“always on”) signaling pathways that drive aberrant cell growth and division. As a result of the altered metabolism in tumor cells, they have a high demand for nutrients that drive anabolic growth including glucose, glutamine, and serine. As discussed in a recent review article focused on the MYC oncogene, ¹ cancer cells are “addicted” to nutrients. It is this liability of nutrient dependence that makes tumor metabolism an attractive target for both basic research and drug discovery efforts. In this special collection of SLAS Discovery, four papers are presented that focus on recent advances in metabolic assay development, metabolomics of patient tumors, and identification of novel inhibitors.

The observation of an altered metabolic state in human tumors dates back almost a century to the work of Dr. Otto Warburg, ² who observed that tumors preferentially utilize glycolysis over oxidative phosphorylation even in the presence of oxygen, a process named oxidative glycolysis. As a consequence of converting most of the glucose into lactate, lactate secretion has become a useful surrogate biomarker for the glycolytic potential of a cancer cell. The detailed mechanism for Warburg’s observation is still unclear; however, recent data suggest that enhanced aerobic glycolysis provides key building blocks needed for tumor cell growth. ³ In addition to glucose, glutamine and serine have both been identified as critical mediators of tumor cell proliferation. Genes coding for proteins involved in glutaminolysis, including the glutamine transporter SLC1A5 and glutaminase GLS1, are direct targets of the oncogene MYC. ⁴ Glutamine provides a source of nitrogen for nucleotide biosynthesis, as well as contributes to glutathione biosynthesis, a molecule necessary to protect tumor cells from redox stress. The importance of glutaminolysis in driving tumor growth has led to the development of glutaminase inhibitors that are now advancing in clinical trials in acute myeloid leukemia (AML) and breast cancer. As the primary source of one-carbon units required for nucleotide biosynthesis, serine is critical to support the proliferation of tumor cells and has become a recent area of active research in cancer metabolism. ⁵

The detailed regulation of tumor cell metabolism has only recently started to become understood following the advent of genomic and metabolomic technologies. With the development of next-generation sequencing and The Cancer Genome Atlas projects, mutations in oncogenes and tumor suppressors have been found in almost all tumor types. Studies over the last decade have demonstrated that many oncogenes (e.g., MYC, HIF1a, PI3K, and RAS) act directly to upregulate nutrient transporters and enzymes in key anabolic growth pathways, including glycolysis and glutaminolysis, thus directly linking genetic drivers with metabolic reprogramming. ⁶ In addition to global transcriptional upregulation, overexpression of metabolic enzymes via gene amplification has also been demonstrated. 3-Phosphoglycerate dehydrogenase (PHGDH), the first enzyme in the serine biosynthesis pathway, is upregulated in a number of estrogen receptor (ER)–negative breast cancers as a result of gene amplification. An in vivo siRNA screen demonstrated that PHGDH was essential for growth in the gene-amplified tumors. ⁷ Possibly the most compelling evidence that metabolism is a driver of tumorigenesis has come from the finding that metabolic enzymes themselves can act directly as oncogenes or tumor suppressors. Somatic point mutations in isocitrate dehydrogenase (IDH) isoforms 1 and 2 have been identified in various tumor types. Utilizing untargeted metabolomics combined with enzymology of the purified enzymes, Dang et al. ⁸ demonstrated that the IDH mutations imparted a neo-enzymatic activity that generated a novel metabolite, 2-hydroxyglutarate (2-HG). Subsequent mechanistic studies demonstrated that 2-HG acted as an oncometabolite in part through an epigenetic reprogramming of the cell, resulting in impaired differentiation and accumulation of tumorigenic progenitor cells. Inhibitors of the mutated IDH enzymes have advanced to clinical development (e.g., AG-120, ClinicalTrials.gov NCT02074839), with evidence of clinical benefit being observed in patients. ⁹ Serum levels of 2-HG have been shown to be a robust quantitative biomarker of disease burden and treatment follow-up in IDH-mutated AML patients. ¹⁰ Deletion of metabolic genes in certain tumors has led to the hypothesis that these enzymes act as tumor suppressors. Loss of succinate dehydrogenase and fumarate hydratase has been observed in some tumor types, leading to the accumulation of succinate and fumarate, respectively, in these tumors.^11,12 As with IDH, part of the mechanism by which these metabolites enhance tumorigenesis is thought to be through epigenetic modulation, although activation of the oncogene HIF has also been implicated as a potential contributing mechanism. ¹³ In addition to mutations, the tissue of origin and tumor microenvironment can also dictate the metabolic profile of tumors. In a recent study from Mayers et al., ¹⁴ it was shown that identical genetic lesions can lead to different metabolic programs, depending on whether tumors originated in the lung or pancreas. In murine lung tumors, mutation of Kras and TP53 resulted in tumors that require uptake of branched-chain amino acids for growth. In contrast, pancreatic tumors with the same genetic mutations did not utilize branched-chain amino acids. Changes in expression of the branched-chain aminotransferase (BCAT) were shown to be responsible for these differences. ¹⁴ These observations add an additional layer of complexity, and opportunity, for targeting cancer metabolism.

Another recent approach to targeting tumor metabolism has utilized synthetic lethality as a method to attack a tumor’s metabolic weakness. One example of this approach came from the observation that the ENO1 gene is in a chromosomal region that is frequently deleted in a subset of glioblastoma tumors. ¹⁵ ENO1 codes for isoform 1 of enolase, an enzyme involved in glycolysis. Considered a passenger deletion, loss of ENO1 does not impair tumor growth; however, it does make the tumors completely dependent on the alternate enolase isoform ENO2. As a consequence, ENO1-deleted glioblastomas are more sensitive to ENO2 inhibition than normal cells, which still retain functional ENO1. ¹⁵ Synthetic lethality has also been used to identify potential therapeutic agents for IDH1 mutant cholangiocarcinoma. Utilizing a chemical biology approach, Saha et al. ¹⁶ demonstrated that IDH1 mutant cholangiocarcinoma cells are hypersensitive to tyrosine kinase inhibitor dasatinib due to a dependence on the enzyme SRC. Additional studies are required to elucidate the essential role of SRC kinase in the survival of IDH1 mutant cholangiocarcinomas.

During periods of stress or in response to insults such as chemotherapy, tumors can engage catabolic pathways as a survival mechanism. One such catabolic pathway that has been proposed to play a role in tumor cell survival is autophagy. ¹⁷ Literally meaning “self-eating,” autophagy is a coordinated process that leads to the degradation of large macromolecules, with recycling of metabolic intermediates to be used as an energy source. Some of the key genes in the autophagy pathway have been identified, and their roles evaluated via genetic manipulation in tumor models. Some of the key autophagy enzymes are “druggable,” including the protease ATG4B and the kinase ULK1, and inhibitors are being actively pursued for drug development. Mammalian cells can express several genes with the same functional activity of different autophagy enzymes, for example, the ATG4 and ATG8 family members that function during autophagy at different stages. This redundancy/promiscuity in the different family members is a necessary consideration in the screening for targeted compounds. In this issue, Xu et al. ¹⁸ developed enzyme and cell-based assays for a key enzyme in the autophagy pathway, the cysteine protease ATG4B. The elegant fluorescence resonance energy transfer (FRET)-based enzyme assay was used for high-throughput screening (HTS) of a large collection of compounds to identify inhibitors of the proteolytic activity of the enzyme. From this, the authors identified a peptide covalent inhibitor that shared similarity to the substrate recognition sequence of ATG4B. Optimization of this hit led to an analogue (compound 2) that had more potent activity (IC50 = 1.1 µM) in the enzyme assay and displayed similar activity in cells. The authors demonstrated that the ATG4B time-resolved (TR) FRET assay is a robust platform that could be further applied for the screening of large compound libraries to identify novel chemical scaffolds. Molecular proximity and specificity are important aspects of many metabolic pathways in cells, such as the one-carbon pool that has both a cytoplasmic and a mitochondrial version.^19,20 In addition to dual location, the one-carbon pool has added layers of complexity in cancer due to isoform switching (MTHFD2 in tumor vs. MTHFD2L in normal cells), as well as enzyme overexpression, especially of the mitochondrial enzyme SHMT2. ¹⁹ Furthermore, the potential for multienzymatic complexes with metabolite channeling makes this a very complex network to dissect. Recent advances in deuterium labeling ²¹ have enhanced our ability to study the dynamics of the one-carbon pool, but techniques such as TR-FRET could provide additional valuable proximity data in cells in real time.

In recent years, immuno-oncology has emerged as a very promising approach to cancer therapy. ²² It had been known through experiments in mice that the immune system can recognize tumors as nonself due to the presence of tumor neo-antigens generated through mutation and chromosomal rearrangement. Although upregulation of immune checkpoints is known to play a major role in tumor-mediated immune suppression, changes in metabolism within the tumor microenvironment have also been shown to be critical. For example, local depletion of tryptophan with concomitant accumulation of kynurenine in the tumor microenvironment has been implicated in cell cycle arrest of CD8+ T cells and generation of CD4+/FoxP3+ Treg cells. ²³ A key enzyme in the tryptophan catabolic pathway, indoleamine 2,3-dioxygenase (IDO1), has emerged as a therapeutic target designed to stimulate the immune response. IDO1 inhibitors have shown efficacy in preclinical models and are advancing in clinical trials. Another metabolite with potent immune-suppressive activity is adenosine. Adenosine accumulates in the tumor microenvironment through the action of CD39- and CD73-mediated hydrolysis of ATP and AMP, respectively. In preclinical models, it has been shown that inhibition of CD73 alone or in combination with the adenosine 2A receptor can result in potent antitumor responses. ²⁴ CD73 has been targeted with monoclonal antibodies that are now in clinical development as agents in combination with checkpoint antagonists. The role of additional metabolites in modulating the tumor immune microenvironment is an area of intense research activity.

All the observations described above indicate that metabolic reprogramming is the result of a complex network of different interacting metabolic pathways, in many cases regulated by oncogenes, other signaling networks, or the microenvironment. This makes different members of these networks potentially druggable targets. As with any other enzymatic reaction, the direct substrate or the end product is the ideal readout for screening purposes. Thus, accurate techniques that measure the levels of metabolites, such as glucose, lactate, glutamine, glutamate, serine, and many other small metabolites, are of particular interest for both basic and applied research. Many of these techniques so far are based on cutting-edge analytic platforms that include liquid chromatography, mass spectrometry, and nuclear magnetic resonance for metabolic profiling, and stable isotope labeling to study metabolic fluxes of labeled metabolites such as glucose, glutamine, or serine. Further sophisticated statistical software (chemometrics) has enabled meaningful information extraction from the global metabolomic data. Untargeted metabolic profiling involves the measurement of hundreds of metabolites with different levels of sensitivity and accuracy for each metabolite. Metabolite identification from untargeted assays is a labor-intensive task, and although it could reveal distinct metabolic signatures of cancer cells in a single experiment, it is a complex and resource-intensive method for HTS. Other simpler techniques, such as colorimetric, fluorometric, and luminescence assays, have also been used with different levels of sensitivity and accuracy. In vitro experiments are also very sensitive to experimental conditions, and thus many different culture conditions should be included in the design of an HTS. The amount of sample and complex sample manipulation are other factors to consider in the design of HTS. In this issue, Leippe et al. ²⁵ describe the development of bioluminescent assays that allow for rapid detection of key cancer metabolites (glucose, lactate, glutamine, and glutamate) in tumor cells and growth media. The high sensitivity, wide linear range, and multiplexing of these assays make them very useful in screening campaigns. The authors make use of the Library of Pharmacologically Active Compounds (LOPAC) cassette to demonstrate the usefulness of these assays for HTS.

Complementary to the mechanistic understanding of tumor metabolism and the development of therapeutics, techniques have evolved to monitor tumor metabolism in patients, opening the door for patient tailoring and determination of pharmacodynamic markers. ¹⁸ F-Fluorodeoxyglucose positron emission tomography (FDG-PET) is widely used in clinical practice and is a direct assessment of tumor glucose uptake. Studies have suggested that a lack of change in FDG-PET can be used as an early predictor of nonresponders to treatment. ²⁶ This provides a powerful tool for clinicians to quickly evaluate whether a particular therapy might be beneficial. More recently, newer PET tracers have been developed to monitor additional metabolites, including glutamine and choline. 2-HG levels in patients with gliomas harboring an IDH1 mutation can be quantitated noninvasively using magnetic resonance spectroscopy. ²⁷ This is an important advance since IDH mutant brain tumors represent a distinct disease from primary glioblastoma multiforme (GBM) and may respond differently to targeted or classical chemotherapeutic agents. ²⁸

In addition to imaging, recent studies have demonstrated the ability to monitor large-scale metabolomic profiles of patients, either from plasma, from tumor biopsies, or directly during surgery. This metabolomic approach has the potential to provide a wealth of quantitative data that can be used to identify key metabolic pathways in patients before and after treatment, as well as at relapse. The less invasive metabolomics analysis of human serum is a rich source of potential biomarkers that could be used as pathological indicators for tumor staging and patient therapeutic tailoring. In this issue, Ríos Peces et al. ²⁹ utilized the power of untargeted metabolomics to evaluate the metabolite profiles of patients with pancreatic ductal adenocarcinoma (PDAC) in comparison with healthy controls. The authors utilized both hydrophilic interaction liquid chromatography (HILIC) and reverse-phase chromatography, coupled to mass spectrometry ionization, to obtain a metabolomic fingerprint from the serum of each patient. Following statistical analysis, four lipid metabolites representing two phospholipid classes (LPC and LPE) were found to be significantly altered (decreased) in patient versus control serum. ²⁹ The authors speculate that phospholipases expressed by the tumor cells might be responsible for the observed lipid changes. These studies represent a starting point (hypothesis generation) to better understand the impact of altered metabolism on PDAC development and progression, with the goal ultimately being to modulate these for therapeutic benefit. Other studies have also shown that serum concentrations of glutamate/glucose and creatine/glutamine ³⁰ or elevated serum levels of branched-chain amino acids ³¹ could be used as early markers for the development of PDAC. These observations emphasize the importance of precise biomarker identification. In many instances, these studies have also been corroborated in animal models, ³² which are in many cases an important tool for the study of tumor metabolic reprogramming using both immunocompetent and immunocompromised mice.

Identification of molecules that target metabolic pathways affected in obesity and other diseases may also be of interest for cancer research since drugs such as metformin and statins have also been found to be therapeutic tools for various types of cancer. ³³ In this issue, Adachi et al. ³⁴ describe a high-throughput assay for intestinal monoacylglycerol acyltransferase (MGAT) activity. The rapid-fire mass spectrometry–based assay is a powerful technique to monitor (MGAT) enzymatic activity. The authors use ¹³C-labeled substrate and can detect both ¹³C-diacylglycerol (DAG) and ¹³C-triacyglycerol (TAG) as products. Utilizing this assay to screen a large collection of small molecules, the authors identified an aryl sulfonamide (T1) inhibitor with potent IC50 against intestinal microsomal MGAT (21 nM), as well as purified MGAT2 enzyme (83 nM). Identification of this small-molecule probe offers the opportunity to characterize the role of MGAT activity in driving obesity and potentially obesity-related tumorigenesis.

With new tools and technologies in place, cancer metabolism has become an area of renewed focus for both basic research and drug discovery efforts. The ability to link genetics with metabolomics and integrate in vivo biomarker assays offers the opportunity to target cancer metabolism in the era of personalized medicine.