Apoptosis-inducing factor has two roles: acting as an apoptosis effector and maintaining mitochondrial metabolism. A recent study uncouples these roles and finds that only the mitochondrial functions are required for tumorigenesis in mice.

Hallmarks of malignancy include evasion of apoptotic signals and altered cellular metabolism to support growth and proliferation. However, there is significant overlap between metabolic and apoptotic functions in the mitochondria, and untangling the precise contribution of each to tumorigenesis poses a significant challenge because several proteins contribute to both pathways.

Apoptosis-inducing factor (AIF) was first described as a caspase-independent apoptosis effector. During induction of apoptosis, AIF translocates from the mitochondria to the nucleus, triggering chromatin condensation and DNA fragmentation.1 Early studies also found that reduction of AIF protein renders cells susceptible to peroxide-induced apoptosis,2 suggesting a functional role in oxidative phosphorylation (OXPHOS) and redox control.3 Multiple models have since elucidated the roles of AIF in mitochondrial respiration. Most consistently, genetic deletion of AIF (or its binding partner CHCHD4) results in decreased protein levels of respiratory complex I,4 and consequently impaired OXPHOS.

Which biological functions of AIF are important for tumor growth and development? Rao et al. explore the role of AIF in lung cancer using genetic mouse models in which KRASG12D drives tumorigenesis in the presence or absence of AIF.5 They began by noting increased AIF protein levels in non-small cell lung cancer (NSCLC) compared to the adjacent non-involved lung tissues. However, in contrast to what the authors expected to find after removing AIF (increased tumor formation), they surprisingly observed that loss of AIF actually impaired lung tumor initiation, resulting in decreased tumor burden and prolonged survival of the mice. Thus, AIF was acting like an oncoprotein rather than a tumor suppressor. AIF loss was characterized by decreased OXPHOS and abnormal mitochondrial morphology. This phenotype is consistent in multiple settings and cells, including primary and transformed pneumocytes in vitro, ex vivo tumor cells and in established NSCLC cell lines. The authors then generated knock-in mice to re-express either wild-type AIF or a Δ96–110 AIF mutant. This mutant is incapable of translocating to the nucleus, thereby nullifying its role as an apoptotic effector. Ahead of time the “bet” was probably that nullifying the apoptotic effector function would cause AIF to lose its tumor suppressive effect. However, the second surprise was that expression of either wild-type or mutant AIF restored the oxygen consumption rates and rescued the ability to form tumorspheres. In vivo, re-expression of wild-type or mutant AIF restored KRASG12D tumorigenesis and increased tumor burden compared to AIF knock-out (Fig. 1).

Fig. 1
figure 1

Outcomes of AIF manipulation in KRAS-driven tumor growth. Mice expressing a mutant KrasG12D allele in the presence (a) or absence (b) of AIF have different levels of tumor growth. Re-expression of mutant AIF (Δ96–110) (c), which restores the metabolic role of AIF but is unable to translocate to the nucleus to induce apoptosis, rescues the tumor growth defect of AIF knock-out tumors. WT, wild-type; KO, knock-out; CHCHD4, Coiled-Coil-Helix-Coiled-Coil-Helix Domain Containing 4; OXPHOS, oxidative phosphorylation. (This figure was created using Servier Medical Art templates, which are licensed under a Creative Commons Attribution 3.0 Unported License; https://smart.servier.com)

Full size image

A common but oversimplified interpretation of the high rates of glucose uptake in many tumors is that the malignant cells within these tumors have defective OXPHOS. However, a number of recent studies have emphasized the importance of OXPHOS and other aspects of mitochondrial metabolism in tumor growth, including in tumors with oncogenic KRAS. In cells expressing KRASG12D, abolishing OXPHOS by deleting mitochondrial transcription factor A severely abrogated tumor cell growth.6 The work presented by Rao et al. extends these previous observations into genetic models and establishes a requirement for AIF-induced OXPHOS in tumor initiation. The new data serve to highlight the essentiality of OXPHOS in lung cancer. Importantly, a propensity for lung tumors to engage in high levels of oxidative metabolism in vivo has also recently been demonstrated in human patients,7,8 suggesting that targeting OXPHOS in tumors may yield clinical benefits. Lessanu Deribe et al. found that in preclinical genetically engineered mouse models (GEMMs) and human NSCLC cell line and xenograft models, NSCLC mutants for SMARCA4, KRAS, and TP53 had both enhanced OXPHOS and marked sensitivity to inhibition of OXPHOS by the small molecule IACS-010759, providing specific “Precision Medicine” biomarker predictions for selection of patients to use OXPHOS targeting agents in NSCLCs.9 In addition, it appears that multiple tumor types as diverse as glioblastomas, acute myeloid leukemias, and mantle cell lymphomas may require OXPHOS, indicating the need for clinical lab certified biomarkers to identify patients across tumor types that might benefit from such targeted therapy.10,11 Indeed, clinical trials with respiratory chain Complex I inhibitors are already underway in cancer (NCT03291938, NCT02882321). Finally, given the current GEMM results, we also need to consider the possibility of targeting OXPHOS as a cancer prevention strategy.