Megamitochondria in alcohol-associated liver disease and cancer: a friend or foe?

Dated back to the 1970s, megamitochondria (giant mitochondria) were discovered in liver biopsy specimens from patients with alcohol-associated liver disease (ALD) using light and electron microscopy. Later, it was shown that patients with megamitochondria appear to be a subgroup of patients with ALD who have milder clinical symptoms, lower rates of cirrhosis, fewer complications and better long-term survival.¹ In line with these observations in human ALD, megamitochondria were also found in experimental ALD models in rodents and in ethanol-treated precision-cut human liver slice cultures. However, the mechanisms by which alcohol induces megamitochondria formation and how megamitochondria impact the pathogenesis of ALD remain obscure, and this mystery has only begun to be gradually revealed recently.

Mitochondria are highly dynamic organelles that constantly undergo fission and fusion, a process known as ‘mitochondrial dynamics’.² Mitochondrial fusion merges two neighbouring mitochondria, regulated by GTPase proteins mitofusin 1 (MFN1) and MFN2 for the outer mitochondrial membrane and optic atrophy 1 (OPA1) for the inner membrane, forming a larger organelle and preserving mtDNA integrity. In contrast, mitochondrial fission splits one mitochondrion into two smaller ones, mediated by the GTPase protein dynamin-related protein 1 (DRP1, gene name DNM1L). In a recent study, we observed increased megamitochondria in human liver samples from individuals with alcohol-associated hepatitis (AH) and in livers from chronic-plus binge (Gao-binge) alcohol-fed mice (figure 1A,B).³ Mechanistically, DRP1 protein levels, as well as those of the fusion proteins MFN1 and MFN2, significantly decreased in human AH liver samples and in Gao-binge alcohol-fed mouse livers.³ The decrease in MFN1 and MFN2 is likely a compensatory response to slow fusion processes, given the accumulation of both elongated and large sphere-shaped megamitochondria. These findings suggest that decreased mitochondrial fission, rather than increased fusion, contributes to the formation of megamitochondria in ALD.

Figure 1. Megamitochondria and a proposed model of mitochondrial dynamics in regulating cell death, pyrimidine metabolism and mtDNA-mediated cGAS–STING innate immune responses in ALD and liver tumourigenesis. (A) Representative H&E staining of human alcohol-associated hepatitis liver and (B) electron microscopy images from a Gao-binge alcohol-fed mouse liver (see Ma et al³). Arrows indicate megamitochondria. Scale bars: 100 µm in (A) and 500 nm in (B). (C) Alcohol consumption reduces hepatic DRP1 levels, leading to decreased mitochondrial fission and the formation of megamitochondria. The lower-left grey module in (C) illustrates mitochondrial stasis during chronic alcohol exposure: the curved arrows denote the normal cycling between DRP1-mediated fission and MFN1/MFN2-mediated fusion, whereas reductions in both processes suppress mitochondrial dynamic remodeling. Newly formed megamitochondria show increased oxygen consumption, NAD⁺ production and fatty acid beta-oxidation to protect against ALD; however, chronic megamitochondria lose these protective responses, resulting in increased release of mitochondrial apoptotic factors and mtDNA. This causes heightened apoptosis and a cGAS–STING-mediated inflammatory response. Megamitochondria also demonstrate increased pyrimidine synthesis, which³ may promote cell proliferation alongside increased cell death and innate immune responses, ultimately contributing to liver injury and liver tumour development. Created with BioRender.com. ALD, alcohol-associated liver disease; ALDH2, aldehyde dehydrogenase 2; cGAS, cyclic GMP-AMP synthase; DRP1, dynamin-related protein 1; IFN, interferon; IRF, interferon regulatory factor; LD, lipid droplet; MFN, mitofusin; NAD⁺, nicotinamide adenine dinucleotide; NADH, nicotinamide adenine dinucleotide-hydrogen; STING, stimulator of interferon genes; TBK1, TANK-binding kinase 1.

Mitochondria play a key role in alcohol metabolism. Alcohol is primarily metabolised in the liver through cytosolic alcohol dehydrogenase and mitochondrial aldehyde dehydrogenase 2 (ALDH2), both of which consume nicotinamide adenine dinucleotide (NAD⁺), causing an initial decrease in hepatic NAD⁺ and an increase in nicotinamide adenine dinucleotide-hydrogen (NADH). This supports the well-established notion that alcohol metabolism decreases the NAD⁺/NADH ratio. Moreover, alcohol consumption is known to cause mitochondrial dysfunction, increase mtDNA degradation and raise oxidative stress in the liver. Using Seahorse-based bioenergetics and liver metabolomic analyses, we unexpectedly observed increased mitochondrial oxygen consumption, higher hepatic NAD⁺ levels and improved mitochondrial fatty acid beta-oxidation in Gao-binge alcohol-fed mice.³ These findings support the idea that alcohol might enhance mitochondrial bioenergetics and both morphological and metabolic flexibility as adaptive responses.

It should be noted that only a small percentage of alcohol drinkers develop more severe fibrosis, cirrhosis or AH. While genetic factors, such as ALDH2 polymorphisms and mutations, may explain the diverse responses to alcohol, cellular adaptive responses, including increased mitochondrial plasticity (elongation and megamitochondria), bioenergetics and NAD⁺ production, also play a vital role in protecting the liver by improving alcohol detoxification and reducing hepatotoxicity. However, these beneficial adaptations, such as megamitochondria, are transient. In alcohol-fed liver-specific DRP1 knockout (L-DRP1 KO) mice, these responses become maladaptive over time, leading to chronic liver injury, inflammation and fibrosis.

Mechanistically, it is well known that damaged mitochondria are removed via selective mitophagy and other lysosomal mitochondrial quality-control mechanisms.⁴ Nonetheless, mitochondrial fragmentation is necessary for efficient removal of damaged mitochondria via mitophagy and other lysosomal-related processes. While megamitochondria initially serve as a protective response, spontaneous damage to mtDNA and proteins within megamitochondria also occurs over time, and megamitochondria cannot be easily cleared by mitophagy-lysosomal pathways due to their large size. Indeed, we found that mitophagy was reduced in L-DRP1 KO mouse livers using the cytochrome c oxidase subunity 8 (COX8)-GFP-mCherry mitophagy reporter assay.³ Chronically accumulated ‘old’ megamitochondria contain higher levels of damaged mtDNA and proteins compared with newly formed ones. The increased respiration observed in alcohol-fed wild-type mice is likely due to the induction of new megamitochondria, whereas most megamitochondria in alcohol-fed L-DRP1 KO mice are ‘chronic’ (or old), showing maladaptation with impaired oxygen consumption and increased release of mitochondrial apoptotic factors and mtDNA, resulting in more severe liver injury.

Clinically defining the transition from old maladaptive megamitochondria to human ALD would be challenging, as early human ALD livers are not feasible for biopsy. Currently, we still lack experimental data to clearly define when the ‘newly formed adaptive megamitochondria’ become ‘old maladaptive megamitochondria’ in alcohol-fed mice. A more detailed time-course study measuring mitochondrial bioenergetics and NAD⁺ production, together with monitoring changes in mitochondrial dynamics and liver pathogenesis, would help define a clear time window for when the megamitochondria become maladaptive. It should be noted that ethanol feeding induces hepatic mitochondrial dysfunction and impairs mitochondrial bioenergetics in rats. Why the alcohol-induced adaptive mitochondrial response was not observed in rats remains to be investigated.

While the evidence above suggests that megamitochondria (newly formed vs old) may have both beneficial and harmful effects on hepatocyte functions, our more recent findings indicate that the balance between mitochondrial fission and fusion, rather than merely increasing megamitochondria formation or mitochondrial fragmentation, is more crucial for maintaining cellular homeostasis and functions in the liver.⁵ To explore how mitochondrial dynamics regulate hepatocyte and liver functions, we generated a series of genetically engineered mice that either lack mitochondrial fission, fusion or both. Mice with hepatic deletion of L-DRP1 exhibited elevated serum alanine aminotransferase levels and fibrosis, with spontaneous hepatocellular adenomas developing in mice by 12–18 months of age. In contrast, L-MFN1 KO and L-MFN2 KO mice showed no significant liver damage or tumours, although a small percentage of L-MFN1, L-MFN2 double KO mice developed tumours later. Interestingly, mice with simultaneous deletion of L-DRP1, L-MFN1 and L-MFN2 (triple KO (TKO) mice) had significantly reduced liver injury, fibrosis and both spontaneous and oncogene-induced (c-Myc and YAP) tumour development. L-DRP1 KO mice showed increased mtDNA release, leading to marked activation of the cyclic GMP-AMP synthase (cGAS)–stimulator of interferon genes (STING)–interferon pathway, which is almost completely suppressed in TKO mice. Both the tumour-suppressing and tumour-promoting roles of cGAS–STING have been documented. Prolonged activation of cGAS–STING inflammatory signalling depletes effector immune cells and attracts immunosuppressive cells, collectively creating an immunosuppressive tumour microenvironment that supports tumour survival and immune evasion. Indeed, deletion of hepatic cGAS reduced spontaneous and oncogene-induced liver tumour development in L-DRP1 KO mice, supporting the idea that the activated cGAS–STING pathway promotes liver tumourigenesis in L-DRP1 KO mice.⁵ In addition to glucose and fatty acid metabolism, mitochondria also play a vital role in pyrimidine metabolism by providing nucleotide precursors for mtDNA replication and supporting cell growth. Dihydroorotate dehydrogenase (DHODH), a mitochondrial inner membrane protein, catalyses the oxidation of dihydroorotate to orotate, which is a rate-limiting step in de novo pyrimidine synthesis. Metabolomics analysis revealed a marked increase in the levels of dihydroorotate and orotate in L-DRP1 KO mice, both of which were blunted in TKO mice.⁵ Because mitochondrial stasis is beneficial for liver function and for preventing hepatocellular carcinoma (HCC), pharmacological inhibition of DRP1 by Mdivi or of MFN1 and MFN2 by MFI8 or SAMβA to create hepatic mitochondrial stasis, as well as pharmacological inhibition of the cGAS–STING pathway, may be beneficial for treating ALD and HCC.

In summary, these new data show that transient induction of the hepatic megamitochondria is a protective metabolic adaptive response against ALD. However, this adaptive response diminishes over time, contributing to the development of ALD. Loss of mitochondrial fission protein and accumulation of megamitochondria are enough to cause liver injury, activate the cGAS–STING innate immune response, disrupt pyrimidine metabolism and result in spontaneous liver tumour formation. These effects can be reversed by restoring ‘mitochondrial stasis’ by simultaneously removing both fission and fusion machinery proteins (figure 1C). These groundbreaking findings advance our understanding of how altered mitochondrial dynamics influence mitochondrial adaptation, development and the progression of ALD and liver tumours, potentially opening new avenues for preventing and treating ALD and liver cancer.

Data availability statement

Data are available upon reasonable request.