Altered energy metabolism is a hallmark of malignancy that can be harnessed to detect and treat cancer1. But tumours are metabolically diverse, so, for treatments to be targeted effectively, non-invasive methods are needed that can observe metabolic characteristics in vivo. Writing in Nature, Momcilovic and colleagues2 report that an imaging agent called 4-[18F]fluorobenzyl triphenylphosphonium (18FBnTP) can be used to identify tumours in mice that could be targeted by inhibitors of oxidative phosphorylation, one of the cellular pathways involved in energy production.
Mitochondria are cellular organelles that integrate essential metabolic functions in the cell, including energy production, the synthesis of biological molecules and signal transduction. These organelles use oxidative phosphorylation to produce the molecule adenosine triphosphate (ATP), the major energy currency of the cell. Oxidative phosphorylation capitalizes on an electrochemical gradient, known as the membrane potential (∆Ψm), that is established by the oxidation of glucose- or nutrient-derived ‘fuel’ molecules in mitochondria (Fig. 1a). Mitochondrial metabolism supports cell proliferation and cancer progression in some malignancies3–5. However, tools have been lacking to observe oxidative phosphorylation directly in tumours, or to predict which tumours rely on this pathway as opposed to other energy-generating pathways — such as glycolysis, which occurs outside mitochondria, in the cytoplasm.
Momcilovic et al. used 18FBnTP to study oxidative phosphorylation in mouse models of lung cancer. This tracer is a positively charged ion that localizes to the negatively charged inner membrane of mitochondria, where oxidative phosphorylation occurs6 (Fig. 1a). The inclusion of fluorine-18 atoms in the molecule provides a radioactive signal that allows accumulation of the tracer to be observed with positron emission tomography (PET), an imaging technique that is commonly used to monitor cancer in the clinic.
In mice genetically engineered to develop non-small cell lung cancer (NSCLC), Momcilovic et al. observed 18FBnTP retention in a subset of tumours. They validated the specificity of the probe by showing that drugs that increase or decrease ∆Ψm had the expected effect on the 18FBnTP signal monitored by PET.
The authors found that 18FBnTP uptake by the various tumour subsets differed, even though they contained the same cancer-promoting ‘drivers’ (mutations in genes encoding the cancer-promoting protein KRas and the tumour-suppressor protein Lkb1). As in human NSCLC, tumour subsets in this mouse model are histologically diverse; that is, they have different tissue architectures when viewed under the microscope. One tumour subtype, called adenocarcinoma, tended to take up 18FBnTP, whereas another subtype, squamous cell carcinoma, did not (Fig. 1b). This was true even for adenocarcinomas and squamous cell carcinomas in the same mouse. Further evidence of the metabolic variability of these tumours came from studies using [18F]fluoro-2-deoxyglucose (18F-FDG) — a PET tracer that is commonly used in the clinic to detect glucose uptake in tumours. Squamous cell carcinomas that were exposed to both 18F-FDG and 18FBnTP took up 18F-FDG but not 18FBnTP, whereas for adenocarcinomas exposed to both tracers, some took up just 18FBnTP and others took up both tracers. Together, the findings reveal that lung tumours have diverse metabolic ‘personalities’ that are amenable to non-invasive monitoring.
Can the metabolic characteristics monitored by these probes be used to predict whether a tumour will respond to treatment? This question is timely because many metabolic inhibitors — including the oxidative-phosphorylation inhibitor IACS-010759 — are undergoing evaluation in clinical trials7. Most of these trials do not have the benefit of being able to non-invasively assess the activity of the pathway targeted by the drug. Momcilovic et al. found that, when mouse tumours were grouped according to their 18FBnTP signal, only those with a high uptake of 18FBnTP (generally adenocarcinomas) showed marked growth suppression when treated with IACS-010759. Inhibiting oxidative phosphorylation rapidly suppressed 18FBnTP uptake, even before the onset of tumour-cell death, indicating that monitoring the tracer provides a way of predicting whether the tumour will respond to the drug. These findings speak to the potential of 18FBnTP to group tumours according to whether or not they are likely to respond to such inhibitors, and to detect early therapeutic responses non-invasively.
The authors’ findings resonate with emerging themes in cancer-metabolism research. Previous studies have generally focused on glycolysis — specifically, the rapid, non-mitochondrial conversion of glucose to the molecule lactate — as a major energy source for cancer cells grown in vitro. By contrast, the role of oxidative phosphorylation has been under-appreciated. But studies of human NSCLC in vivo have revealed metabolic heterogeneity and prominent metabolism using pathways linked to oxidative phosphorylation (such human studies lacked the tools to monitor oxidative phosphorylation directly)8,9.
Metabolic reprogramming in cancer has commonly been considered to involve a ‘switch’ from oxidative phosphorylation to glucose uptake, leading instead to glycolysis, but the findings reported by Momcilovic et al. support the idea that this is an oversimplification. In their models, some of the adenocarcinomas take up both 18F-FDG and 18FBnTP, and glucose uptake as revealed by 18F-FDG with PET neither rules out oxidative phosphorylation nor predicts whether tumours will respond to or resist oxidative-phosphorylation inhibitors. The use of 18FBnTP instead provides a more direct assessment of oxidative phosphorylation by providing a way of monitoring ∆Ψm, and might help to interpret the metabolic significance of other tracers.
With respect to the particular utility of 18FBnTP in NSCLC, it is notable that histological classification of these tumours in humans so far requires invasive tissue sampling, usually from a single site that might not reflect the complexity across the entire tumour. This is important, because the distinction between adenocarcinoma and squamous cell carcinoma influences the choice of chemotherapy10. An imaging technique such as the one proposed by Momcilovic and colleagues might therefore help to guide therapeutic decision-making.
The challenges of assessing disease-associated metabolic perturbations in vivo have hampered our ability to translate mechanistic insights into therapies. Imaging agents such as 18FBnTP might help by enabling key aspects of tissue metabolism to be observed non-invasively. Altered mitochondrial function, including enhanced or suppressed ∆Ψm, has been observed in diverse human diseases, including cancer, neurodegeneration and heart dysfunction, not to mention normal ageing11. It will be fascinating to see whether 18FBnTP can help us to understand such conditions in experimental organisms, and eventually in humans.
Nature 575, 296-297 (2019)