Tumor start‐up: mitochondrial fusion makes it happen

Whether changes in cellular metabolism precede tumor formation and trigger malignant properties or simply serve as a bioenergetic adaptation of cancer during disease progression remains debated. Bonnay et al (2020) now show that a metabolic reprogramming toward increased oxidative phosphorylation is required for irreversible cell immortalization and subsequent tumor formation.

Recent work reports aberrant mitochondrial fusion and increased oxidative phosphorylation as potent metabolic drivers of brain tumors in Drosophila.

Otto Warburg described that cancer cells preferentially convert glucose to lactate, regardless of oxygen availability, a phenomenon known as the Warburg effect. This observation was intriguing since anaerobic glycolysis produces less ATP than complete oxidation of glucose via tricarboxylic acid (TCA) cycle/electron transport chain (ETC) in oxidative phosphorylation (OxPhos). Since then, studies revealed that not all tumors have a preference for glycolysis; in fact, many have high mitochondrial activity, being vulnerable to OxPhos inhibition. The balance between anaerobic glycolysis and OxPhos in cancer cells is quite dependent on the tumor type and the stage of tumor progression (Ashton et al, 2018). In addition, many tumors are composed by highly heterogeneous populations with distinct molecular and metabolic signatures, caused by microenvironmental influences and (epi)genetic plasticity. Nonetheless, one of the key hallmarks of cancer cells is their capacity to reprogram metabolism to support the energetic demands of high rates of proliferation (Ashton et al, 2018). Whether metabolic rewiring occurs at the first steps of tumor initiation, providing tumorigenic properties to cancer cells or if it only reflects metabolic adaptation to the tumor context, is one of the most seminal questions in current cancer research. Due to the complexity of higher animals, the initial steps of tumor formation have been hard to dissect. In this work, Bonnay et al (2020) used a Drosophila neural stem cell brain tumor model to investigate in vivo if tumor metabolic reprogramming has an active role in tumor formation and cell immortalization

The Drosophila brain is generated by neural stem cells, the neuroblasts (NBs), that divide asymmetrically to self‐renew and give rise to neural progenitors more committed to differentiation, as intermediate neural progenitors (INPs). Eventually, these cells can generate neurons and glia; however, mutations in asymmetrically segregated differentiation factors result in INPs that retain NB‐like self‐renewal capacity and become tumor‐like NBs (tNBs). tNBs do not respond to regulatory mechanisms and proliferate indefinitely, forming malignant brain tumors (Caussinus & Gonzalez, 2005). Generation of brain tumors by Brat (Brain tumor) depletion is one of the most widely used Drosophila cancer models. Upon division, Brat‐depleted INPs are cell cycle arrested during 24–48 h and only after this period re‐enter cell cycle, initiating brain overgrowth (Bowman et al, 2008). Although the biological reason for this lag period is not understood, it has been hypothesized to be important for these abnormal INPs to adopt tumorigenic properties. Therefore, this time window provides a very convenient system to study whether there are any metabolic alterations in tumor‐initiating cells and whether these changes are required for malignancy.

As a proof of principle Bonnay et al (2020) started by investigating if Brat tumors have a metabolic profile that is different from normal brains. Using transcriptome data of both Brat‐depleted and normal NBs they found that in tNBs most glycolytic enzymes were highly upregulated, whereas OxPhos‐related enzymes were unaltered or downregulated. Although at face value these results indicated that tNBs were mainly glycolytic, a direct assay of their metabolic profile by seahorse and metabolomic analysis revealed the opposite. Surprisingly, the authors showed that Brat tumors had high oxygen consumption rates and increased abundance of TCA cycle intermediates, characteristic of oxidative metabolism. These observations emphasize the fact that the transcriptional profile of cells not always recapitulates their metabolic profile.

Although globally Brat tumors have an oxidative metabolism, through the analysis of a genetically encoded fluorescent NAD⁺/NADH sensor (SoNar), Bonnay et al (2020) detected that these tumors are metabolically heterogeneous. Through single‐cell RNA sequencing of tNBs, they were able to identify tumor cell populations with distinct characteristics, including a small population of highly proliferative cells with a signature of enhanced OxPhos gene expression. This diversity agrees with Genovese et al (2019), which demonstrated that a different Drosophila NB‐derived brain tumor model contains a vast cellular heterogeneity, capable of conferring different metabolic strategies and proliferative properties to tumors. Moreover, the presence of tumor metabolic diversity is conserved through many human tumors, as in small cell lung cancer. Hensley et al (2016) beautifully showed in vivo that these tumors displayed enhanced glucose oxidation, although being capable of using alternative nutrients, as lactate, depending on tumor perfusion.

To test the functional role of OxPhos in tumor growth, Bonnay et al (2020) performed a genetic screen to reveal the requirement of major metabolic pathways as glycolysis, OxPhos, pentose phosphate pathway and gluconeogenesis in tumor growth. Interestingly, only OxPhos inhibition, by knock down of ETC/TCA cycle enzymes, was able to rescue Brat tumorigenesis. Since wild‐type NBs rely mainly in glycolysis to proliferate (Homem et al, 2014), this study suggests that tNBs suffer a metabolic reprogramming from glycolysis to OxPhos. This also adds to the increasing number of tumors reported to not have a classical Warburg metabolism (Ashton et al, 2018).

To dissect at what stage of tumor development is OxPhos required, the authors analyzed the metabolic profile of NB lineages in vivo in the early steps of tumor formation. After the oncogenic signal, caused by depletion of Brat, INPs suffer an initial lag period before tumor initiation. Through a series of genetic rescue experiments, the authors showed that, during this lag period, these abnormal INPs become irreversibly transformed and suffer a strong increase in NAD⁺ levels, indicative of OxPhos upregulation. Hence, surprisingly, the switch to OxPhos is coincident with tNB immortalization.

Bonnay et al (2020) also discovered that mitochondria fusion, often associated with an increase in OxPhos, is a crucial event for metabolic reprogramming and subsequent Brat tumor formation. The fusion regulator marf was specifically upregulated in tumor‐initiating cells with high levels of OxPhos and its depletion avoided tNB immortalization. This characteristic was shared with a different NB‐brain tumor model, suggesting that tumor formation, through a different oncogene, may also require the same metabolic reprogramming. Moreover, mechanistically, Bonnay et al (2020) demonstrated that high OxPhos levels are required for NB immortalization through the generation of NAD⁺, and not by affecting ATP or ROS levels. It is now vital to dissect how the increase in NAD⁺ pool can promote cell immortalization.

Altogether, Bonnay et al (2020) unraveled a new fundamental role of OxPhos and mitochondrial dynamics in the initial steps of tumorigenesis (Fig 1). Interestingly, the interplay between mitochondria fusion and OxPhos in cancer cells seems to be conserved. In liver cancer, it was recently reported an increase in mitochondria fusion, capable of reprogramming metabolism to support tumor growth (Li et al, 2020). Also, in breast cancer, the highly expressed oncogene MYC was shown to correlate with an increase in mitochondria fusion to promote enhanced mitochondria metabolism (von Eyss et al, 2015). As Brat tumors were also shown to upregulate the levels of the Drosophila Myc (Betschinger et al, 2006), it would be interesting to understand whether Myc is one of the mechanisms triggering mitochondria fusion and metabolic changes. Also fascinating would be to understand whether the tumor microenvironment can influence these changes. Besides opening a new avenue of research questions, this work is likely to be a game changer in the view of tumor biology, as it describes metabolic reprogramming, not as a consequence, but as a cause for malignant transformation.

Figure 1. Upregulation of OxPhos and mitochondrial fusion drives tumorigenesis in Drosophila brain tumors.

Drosophila neuroblasts (NB) divide asymmetrically to self‐renew and to give rise to more committed cells, as intermediate neural progenitors (INP). RNAi of brain tumor (Brat) results in INPs that retain the NB self‐renewal capacity and become tumor‐like NBs (tNBs) leading to brain overgrowth. Bonnay et al (2020) now show that during the immortalization period, tNBs rewire their metabolism from glycolysis to OxPhos by upregulating mitochondrial fusion. An increase in mitochondrial NAD⁺ levels and upregulated mitochondrial fusion are the triggers for the tumorigenic behavior of tumor‐initiating cells (TIC). Despite having an overall oxidative metabolism, Brat tumors are composed of highly heterogeneous populations with distinct metabolic profiles, recapitulating the heterogeneity observed in mammalian tumors.

The EMBO Journal (2020) 39: e106927.