How Changes in Cell Energy Factories Relate to Fatty Liver Disease Linked to Metabolism and How Exercise May Help

Mitochondrial fusion, primarily mediated by outer membrane proteins (MFN1, MFN2) and the inner membrane protein (OPA1), is essential for maintaining mitochondrial network connectivity and integrity. In pathological states of MASLD and MASH, fusion protein expression is generally suppressed, although subtle discrepancies exist across different models.

Multiple studies have demonstrated significant downregulation of MFN1 in MASLD and MASH (Table 1). Specifically, MFN1 expression was significantly reduced in the MASLD model [16,17], with the same phenomenon observed in the MASH model reported by Gonçalves et al. [18] and in the type 2 diabetes mellitus (T2DM) co-morbid MASH model studied by Wang et al. [19]. However, certain models, such as the gestational diabetes mellitus (GDM)-associated MASLD model described by Stevanović-Silva et al. [20] and the 8-week Western diet (WD)-induced model by Rosa-Caldwell et al. [21], did not exhibit significant alterations in MFN1. Regarding intervention efficacy, aerobic exercise demonstrated significant ameliorative effects. Deng et al. [17] reported that both HIIT and MICT reversed the downregulation of MFN1 protein in MASLD. Similarly, Gonçalves et al. [18] and Stevanović-Silva et al. [20] found that MICT significantly increased MFN1 protein levels. In contrast, VWR failed to induce significant changes in MFN1 protein across multiple studies [18,21]. Notably, da Costa Fernandes et al. [16] observed that RE resulted in a further suppression of Mfn1 mRNA expression.

MFN2, the most extensively studied fusion protein, demonstrated consistent downregulation across various MASLD induction protocols [17,20,21,22,23], including T2DM-comorbid MASH [19]. However, no significant changes in MFN2 were observed in some MASH models induced by high-fat diets or pharmacological agents [18,24,25]. Interestingly, da Costa Fernandes et al. [16] reported a significant increase in Mfn2 mRNA in a 14-week high fat diet (HFD)-induced MASLD model in Swiss mice. MICT has been confirmed to significantly upregulate MFN2 protein expression across multiple models [17,18,20,23,24], and combined training (RE + MICT) also showed efficacy in increasing MFN2 protein [22]. While Deng et al. [17] and Wang et al. [19] found that HIIT significantly increased MFN2 expression, Andani et al. [25] reported no significant changes following either MICT or HIIT in a MASH model. VWR alone had no significant impact on MFN2 protein expression [18,21]. Furthermore, in the study by da Costa Fernandes et al. [16] RE was found to inhibit Mfn2 mRNA expression, contrasting with the disease-induced upregulation observed in their specific model.

OPA1 expression is typically reduced in MASLD/MASH models [20,21,22,23], although some studies reported no significant changes [18,26]. The regulation of OPA1 by exercise in rodent models exhibits intensity dependence and heterogeneity. A number of studies have indicated that MICT and VWR do not induce significant alterations in OPA1 expression [18,20,21]. Li et al. [26] demonstrated that MICT exerted no appreciable effect, whereas HIIT markedly elevated OPA1 protein levels. Hu et al. [22] found that combined training also increased OPA1 protein. In a zebrafish model, MICT was sufficient to promote OPA1 protein expression [23].

In summary, MASLD/MASH pathology leads to the downregulation of the fusion proteins MFN1 and OPA1. The expression of MFN2 appears to be disease-stage dependent, being downregulated in MASLD but potentially unchanged in MASH. Aerobic exercise, including MICT and HIIT, is the most effective intervention for restoring fusion markers. Notably, the improvement of OPA1 may be contingent upon high-intensity exercise. In contrast, VWR exerts a weaker regulatory effect on fusion proteins, while resistance training may inhibit the expression of fusion genes under specific conditions.

Mitochondrial fission is primarily mediated by Drp1, Fis1, and mitochondrial fission factor (Mff), with Drp1 and Fis1 being the focal points of research. Unlike the relatively consistent downregulation of fusion proteins, alterations in fission proteins during MASLD/MASH present a complex pattern.

Drp1 exhibits high heterogeneity in disease models. Zou et al. [23] and Rosa-Caldwell et al. [21] found decreased Drp1 protein in MASLD, whereas da Costa Fernandes et al. [16] and Hu et al. [22] reported elevated expression. No significant changes were observed in other MASH or GDM-MASLD models [18,19,20]. This heterogeneity in baseline expression dictates the directional outcomes of exercise interventions. In models where Drp1 may be overexpressed or unchanged, high-intensity or greater exercise volume has been observed to reduce its expression. Combined training, HIIT, and RE each led to a significant decrease in Drp1 protein expression [16,19,22]. Stevanović-Silva et al. [20] found that MICT combined with daily VWR reduced Drp1 protein, despite an opposing trend in mRNA. It appears that low-intensity exercise is inadequate in inducing substantial changes in Drp1. Neither MICT nor VWR have been shown to induce significant protein alterations [18,21]. However, in a zebrafish model, moderate-intensity swimming significantly increased Drp1 protein expression [23].

In contrast to Drp1, Fis1 shows a consistent trend. HFD interventions consistently led to significantly increased Fis1 expression [16,17,26]. Exercise interventions targeting Fis1 were highly effective and consistent. MICT, HIIT, and RE all significantly downregulated Fis1 expression [16,17,26]. However, VWR failed to induce significant changes in Fis1 mRNA in a WD-induced model [21]. Data on Mff is scarce; only Rosa-Caldwell et al. [21] reported a significant decrease in Mff mRNA in MASLD, which was unaffected by VWR.

In summary, the imbalance of mitochondrial fission in MASLD/MASH is characterized by a consistent elevation of Fis1. Exercise interventions primarily inhibit excessive pathological fission by downregulating Fis1. The regulation of Drp1 appears to be dependent on exercise intensity and volume. Notably, resistance training demonstrates efficacy similar to other modalities in inhibiting fission proteins, contrasting with its unique suppressive profile in fusion protein regulation.

Most studies assessed biogenesis via peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) and mitochondrial transcription factor A (TFAM) with fewer measuring nuclear respiratory factor 1/2 (NRF1/2). In the majority of models [17,18,19,21,23], protein or mRNA levels of biogenesis markers were significantly reduced. Stevanović-Silva et al. [20] found no significant changes in these proteins in a GDM-associated MASLD model. Conversely, da Costa Fernandes et al. [16] reported that HFD intervention paradoxically led to significant increases in PGC-1α protein and NRF2 mRNA/protein.

Exercise modalities exert divergent effects on biogenesis. MICT and HIIT showed consistent efficacy; multiple studies confirmed that both modalities significantly induced the expression of PGC-1α and TFAM [17,18,19,20,23]. The effects of VWR were inconsistent, with Gonçalves et al. [18] reporting increased PGC-1α protein and Rosa-Caldwell et al. [21] finding no change in mRNA. Notably, RE exhibited a potential inhibitory effect; da Costa Fernandes et al. [16] showed that RE significantly decreased the expression of PGC-1α, TFAM, and NRF1/2.

Mitophagy was assessed via the PTEN-induced kinase 1 (PINK1)/E3 ubiquitin-protein ligase Parkin (Parkin) pathway and flux markers. In MASLD/MASH models, the suppression of the canonical PINK1/Parkin pathway (reduced protein expression) was the most consistent finding [17,18,20,23,26]. Aerobic exercise demonstrated strong consistency in restoring PINK1/Parkin expression; most MICT and HIIT protocols significantly increased their levels [17,18,20,23,26], whereas VWR did not [18]. In the GDM-associated model, Stevanović-Silva et al. [20] observed that MICT failed to reverse the reduction in Parkin.

Regarding autophagic flux, baseline alterations varied across models. Zou et al. [23] and Li et al. [26] observed elevated sequestosome-1 (P62) protein or increased Microtubule-associated protein 1 light chain 3 (LC3)-II/LC3-I ratios in MASLD, suggesting blocked flux. However, other studies found no significant changes [18,20,21]. In terms of intervention, moderate-to-low intensity exercise tended to unclog autophagic flux. MICT significantly reduced accumulated P62 or elevated LC3 ratios in MASLD models [23,26], though it had no significant effect in a MASH model [18]. In the WD-induced MASLD model, Rosa-Caldwell et al. [21] found that VWR significantly increased the LC3-II/LC3-I ratio and decreased P62, whereas VWR had no effect in the MASH model [18]. HIIT showed no significant impact on the LC3-II/LC3-I ratio in one MASLD study [26].

Additionally, Li et al. [26] found that MASLD led to elevated BCL2/adenovirus E1B 19 kDa protein-interacting protein 3 (BNIP3) protein, which was significantly downregulated by both MICT and HIIT. Conversely, Rosa-Caldwell et al. [21] reported decreased BNIP3 protein in MASLD, which was unaffected by VWR. MICT was also found to significantly increase damage-regulated autophagy modulator (DRAM) mRNA expression in a MASH model by Andani et al. [25].

In summary, in MASLD/MASH models, the canonical mitophagy pathway (PINK1/Parkin) is consistently suppressed. HIIT and MICT show strong consistency in restoring PINK1 and Parkin expression. While baseline changes in autophagic flux markers vary across models, exercise generally tends to facilitate the unclogging or enhancement of autophagic flux.

To systematically summarize and compare the regulatory patterns of different exercise interventions on hepatic mitochondrial indices, we integrated results from all 11 included studies to cross-compare the effects of five exercise modalities—MICT, HIIT, VWR, combined training, and RE—on the expression of 16 key molecules involved in mitochondrial fusion, fission, biogenesis, and quality control (Figure 3). The overall analysis reveals that the effects of exercise interventions exhibit significant "marker specificity" and "modality dependence."

Specifically, three distinct regulatory characteristics can be identified from the landscape: First, consistency in improvement: Regardless of the exercise modality, the suppression of Fis1 overexpression and the upregulation of MFN2 appear as the most universal beneficial mechanisms, highlighting the core commonality of exercise in restoring mitochondrial network connectivity and inhibiting pathological fragmentation. Second, heterogeneity in regulation: The alteration of the fission protein Drp1 exhibits drastic fluctuations across studies (increase, decrease, or no change). This high degree of heterogeneity visually reflects the sensitive dependence of this marker on the stage of disease progression and the specific intervention protocol. Finally, modality specificity: Aerobic exercise (MICT and HIIT) demonstrates broad and stable positive regulatory effects on activating biogenesis (PGC-1α, TFAM) and restoring mitophagy pathways (PINK1/Parkin). In contrast, resistance training and voluntary wheel running display unique molecular fingerprints, such as the potential suppression of certain fusion genes by resistance training and the ineffectiveness of voluntary exercise on specific markers. This visual summary compellingly suggests that while exercise is generally beneficial, the remodeling of its molecular targets is highly dependent on the type and intensity of the exercise.

Mitochondrial fusion facilitates the homogenization of trans-inner membrane potential, mitochondrial DNA (mtDNA) complementation, and the sharing of tricarboxylic acid cycle intermediate pools by merging adjacent mitochondria. This fusion machinery is primarily orchestrated by key regulators, including MFN1 and MFN2 anchored to the outer mitochondrial membrane, and OPA1 residing in the inner membrane [30] (Figure 5).

Beyond its pivotal role in mitochondrial fusion, MFN2 also tethers mitochondria to the endoplasmic reticulum (ER), facilitating lipid transfer and insulin signaling to maintain energy homeostasis [31]. However, in the context of MASLD, MFN2 expression is significantly downregulated [32,33]. This downregulation compromises the integrity of ER-mitochondria contact sites, impairs mitochondrial membrane phosphatidylethanolamine biosynthesis, and reduces oxidative phosphorylation efficiency, thereby exacerbating intracellular lipid accumulation [32]. Furthermore, MFN2 suppression triggers robust ER stress and inflammatory responses, driving the pathological progression from steatosis to fibrosis and hepatocellular carcinoma [34]. Similarly, MFN1 is markedly downregulated in steatohepatitis models, an effect further exacerbated by proinflammatory cytokines [35].

Within mitochondria, OPA1 preserves inner membrane fusion and the tightness of cristae junctions, a function critical for sequestering pro-apoptotic factors such as cytochrome c within the cristae lumen to prevent their release and for optimizing the assembly of respiratory supercomplexes [36,37]. The functionality of OPA1 hinges on a precise stoichiometric balance between its long and short isoforms, a dynamic equilibrium governed by the proteases Mitochondrial inner membrane zinc-dependent metalloprotease (OMA1) and Yeast mitochondrial escape 1-like ATPase (YME1L) [38]. In MASLD, this balance is frequently disrupted [22,23]. The consequent OPA1 dysfunction not only induces excessive mitochondrial fragmentation but also dismantles cristae architecture, impairs ATP synthesis, and heightens susceptibility to apoptosis, thereby constituting a core driver of hepatocyte injury [39].

The central regulator governing mitochondrial fission is Drp1, encoded by the DNM1L gene in humans [40]. Structurally, Drp1 lacks a pleckstrin homology domain capable of directly recognizing and binding lipid membranes. Consequently, its recruitment from the cytosol to prospective fission sites on the outer mitochondrial membrane relies on specific receptors, such as MFF, Fis1, and MiD49/51, to initiate the subsequent membrane constriction and scission [40] (Figure 5).

In sharp contrast to the suppression of mitochondrial fusion, mitochondrial fission processes are aberrantly hyperactivated in MASLD [22,23,41]. The aberrant activation of Drp1 exacerbates hepatic lipid metabolic imbalance through multiple pathways. It not only disrupts mitochondrial homeostasis, leading to diminished ATP synthesis efficiency and cellular energy metabolic defects [42], but also reprograms lipid metabolism towards a pathological state characterized by "increased synthesis and decreased lipolysis" [43]. Furthermore, aberrant Drp1 activation precipitates functional defects in succinate dehydrogenase, thereby interfering with tricarboxylic acid cycle flux and insulin signal transduction, which further promotes abnormal hepatic lipid accumulation [44,45]. Ultimately, the excessive ROS generated by mitochondrial fragmentation trigger oxidative stress and inflammatory cascades, establishing a vicious cycle where metabolic dysregulation and inflammatory injury mutually reinforce one another [42].

Notably, the role of Drp1 in MASLD is characterized by distinct stage-specificity. During the early phase of hepatic steatosis, Drp1 hyperactivation serves as a pivotal pathogenic driver. Consequently, inhibiting its activity can effectively ameliorate mitochondrial dynamic imbalances and attenuate the progression of lipid deposition in hepatocytes [41,46,47]. Conversely, as the disease progresses to MASH, the functional role of Drp1 shifts, becoming indispensable for maintaining mitochondrial quality control. At this stage, Drp1 is critical for the clearance of dysfunctional mitochondria; excessive inhibition leads to the accumulation of damaged organelles, thereby exacerbating cellular stress, inflammatory responses, and the overexpression of profibrotic genes [48]. Therefore, therapeutic strategies targeting Drp1 must be precisely tailored to the stage of disease progression. In advanced stages, simplistic 'one-size-fits-all' inhibitory approaches may yield counterproductive outcomes.

Our analysis reveals that MFN2 is the most sensitive target of hepatic mitochondrial fusion in response to exercise. In various MASLD/MASH models, MFN2 is significantly downregulated, and aerobic exercise can significantly upregulate its expression via PGC-1α-dependent pathways [18,19]. In contrast, the restoration of the inner membrane fusion protein OPA1 exhibits significant "intensity dependence". Li et al. [26] noted that only HIIT significantly increased OPA1 protein levels, whereas MICT was ineffective. The mechanism behind this phenomenon may relate to the activation threshold of the cellular energy sensor AMP-activated protein kinase (AMPK). High-level mechanistic studies indicate that AMPK can directly phosphorylate OPA1 or regulate mitochondrial membrane potential to maintain cristae tightness [50]. The high metabolic stress generated by HIIT may more effectively activate this pathway, thereby specifically driving OPA1-mediated inner membrane remodeling.

Notably, RE exhibits a unique pattern of impact on fusion proteins. da Costa Fernandes et al. [16] found that while RE increased bioenergetic markers such as ATP synthase subunit alpha (ATP5), it suppressed MFN1 mRNA expression. This may be linked to the competitive inhibition between the mTORC1 signaling pathway activated by RE and the PGC-1α-driven biogenesis pathway in the liver [51], suggesting potential limitations of resistance training alone in improving hepatic mitochondrial morphology.

Drp1 exhibits high model-dependent heterogeneity in MASLD/MASH. First, in rodent models where Drp1 is overactivated or even unchanged, exercise tends to decrease its expression to inhibit pathological fission. For instance, both RE and combined training significantly reduced elevated Drp1 levels [16,22]. Wang et al. [19] found that in T2DM combined with MASH models, although baseline Drp1 changes were not significant, HIIT still induced a significant downregulation of Drp1. The molecular mechanism underlying this inhibitory effect has been elucidated by Hu et al. [22]: exercise activates the nicotinamide adenine dinucleotide (NAD)+-dependent deacetylase Sirt1, which directly deacetylates Drp1 at the K642 site, thereby inhibiting its translocation to mitochondria and excessive fission activity. Furthermore, da Costa Fernandes et al. [16] discovered that the suppression of Drp1 by RE might be linked to its epigenetic regulation of MTCH2 (Mitochondrial Carrier Homolog 2) promoter methylation, subsequently remodeling the metabolic environment of the outer mitochondrial membrane.

Second, the regulation of Drp1 appears to exhibit a distinct "intensity threshold." While HIIT significantly downregulated pathologically elevated Drp1 levels, lower-intensity interventions such as VWR and standalone MICT failed to induce significant changes in Drp1 protein expression. However, augmenting the exercise load proved effective [18,21]. Stevanović-Silva et al. [20] found that combining MICT with daily VWR significantly reduced Drp1 protein, a beneficial effect also observed with combined aerobic and resistance training strategies. Notably, in a zebrafish model, Zou et al. [23] reported that MICT conversely led to an increase in Drp1 protein expression. This discrepancy may reflect an adaptive response specific to lower vertebrates or early disease stages, where a moderate increase in fission facilitates necessary mitochondrial turnover and biogenesis.

Unlike the heterogeneity of Drp1, another fission receptor protein, Fis1, exhibits high consistency. HFD typically leads to a significant increase in Fis1 expression, indicating a generalized activation of mitochondrial fission mechanisms in pathological states [16,17]. Various forms of exercise interventions (MICT, HIIT, and RE) consistently and significantly downregulate Fis1 expression, effectively curbing excessive mitochondrial fission. Regarding Mff, Rosa-Caldwell et al. [21] reported that its mRNA was decreased in a WD model and insensitive to voluntary exercise.

In summary, the exercise-mediated regulation of mitochondrial fission is multidimensional: it inhibits pathologically activated Drp1 and Fis1 through Sirt1-mediated deacetylation and epigenetic modifications, but in specific contexts (such as adaptive compensation phases), it may also maintain or fine-tune Drp1 levels to coordinate with biogenesis, a process highly dependent on sufficient exercise intensity or load.

Based on the synthesized evidence, we propose the existence of a potential "intensity threshold" governing hepatic mitochondrial dynamics. Although direct comparisons within the same study are limited, a cross-study analysis suggests a distinct pattern: the restoration of MFN2-mediated outer membrane fusion and the suppression of aberrant Fis1 activity appear achievable via MICT. In contrast, profound structural remodeling of the inner membrane (OPA1) and the reversal of pathological Drp1 activation seem to necessitate higher-intensity stimuli (e.g., HIIT) or a greater cumulative exercise load. Furthermore, future investigations should focus on post-translational modifications (e.g., acetylation, SUMOylation) to bridge the gap between transcriptional profiles and protein abundance.

In most studies, exercise-induced dynamic remodeling is highly synchronized with the restoration of mitochondrial biogenesis. Both MICT and HIIT can significantly upregulate PGC-1α and its downstream transcription factor TFAM [19,23]. The mechanism of this synergistic effect lies in PGC-1α acting not only as the "master regulator" of biogenesis but also as a transcriptional co-activator for MFN2, ensuring that newly generated mitochondria can rapidly fuse into the network. However, da Costa Fernandes et al. [16] pointed out the distinctiveness of RE, which showed a suppressive trend for PGC-1α mRNA. This suggests that when prescribing exercise for MASH patients with sarcopenia, caution is needed regarding the potential interference of resistance training alone on hepatic mitochondrial turnover.

Exercise plays a crucial "scavenger" role in regulating mitophagy. Rosa-Caldwell et al. [16] found that even low-intensity VWR significantly enhanced basal autophagic flux (increased LC3-II/I, decreased p62). This suggests that enhancing autophagy may be one of the earliest adaptive responses of exercise improving liver health. Mechanistically, HIIT showed stronger consistency than MICT in restoring the PINK1/Parkin pathway [26], which may be related to high-intensity exercise more effectively inducing transient mitochondrial depolarization, thereby triggering the stabilization of PINK1 on the outer mitochondrial membrane [52]. Additionally, Andani et al. [25] found that MICT specifically upregulated DRAM, indicating that different exercise modalities may activate distinct autophagy receptor pathways to clear damaged mitochondria.

Ultimately, the orchestrated restoration of mitochondrial dynamics and quality control machinery by exercise is not an end in itself, but a means to re-establish hepatic bioenergetic homeostasis. The exercise-induced restoration of mitochondrial dynamics proteins represents more than a mere morphological adaptation; it functionally translates into enhanced hepatic metabolic capacity. Specifically, the upregulation of fusion proteins extends beyond facilitating outer membrane fusion; it plays a pivotal role in reconstituting the physical tethers at mitochondria-associated endoplasmic reticulum membranes (MAMs) and lipid droplet-mitochondria contact sites [24,53]. The re-establishment of these organelle interfaces is indispensable for optimizing calcium signaling and facilitating the direct transfer of fatty acids into the mitochondrial matrix, thereby potentiating β-oxidation rates, maximizing ATP production, and suppressing lipotoxicity-induced apoptosis [24,54]. Concurrently, the exercise-mediated suppression of pathological fission directly alleviates reactive oxygen species (ROS) generation and mitochondrial depolarization, which functionally manifests as improved hepatic insulin sensitivity and augmented whole-body fat oxidation [22,55]. Together, these molecular adaptations forge a robust link between mitochondrial structural remodeling and the clinical alleviation of hepatic steatosis and inflammation.