Mitochondrial biogenesis (also called mitobiogenesis or mitogenesis), i.e., the synthesis of new mitochondria, can be stimulated by different endogenous agents, including adenosine monophosphate (AMP), oxidized nicotinamide adenine dinucleotide (NAD⁺), cyclic AMP (cAMP), and calcium ions (Ca²⁺), among other signals (Pfanner et al., 2019) (Figure 2). In that regard, mitochondrial biogenesis is part of the cellular adaptation to exercise, caloric restriction, temperature decline, and diseases, among other physiological contexts (Jornayvaz and Shulman, 2010; Wang et al., 2023). Increased AMP amounts, which can indicate low ATP levels due to decreased production and/or increased consumption, promote the activation of AMP-activated protein kinase (AMPK), that phosphorylates the peroxisome proliferator-activated receptor-gamma coactivator-1α (PGC-1α), a transcription coactivator that translocates to the cell nucleus and stimulates the expression of the nuclear respiratory factors 1 and 2 (NRF1 and NRF2, respectively) (Jäger et al., 2007; Fernandez-Marcos and Auwerx, 2011). In the nucleus, NRF1 and NRF2 promote the expression of mitochondrial transcription factor A (TFAM), a protein that induces the transcription of specific targets in the mitochondria, as well as promotes mitochondrial DNA (mtDNA) replication and condensation (Farge et al., 2014; Kukat et al., 2015). PGC-1α can be activated also by SIRT1, a deacetylase, which is upregulated by NAD⁺ (Gerhart-Hines et al., 2007; Rodgers et al., 2005). AMPK also upregulates SIRT1 by increasing the levels of NAD⁺ (Cantó et al., 2009; Fulco et al., 2008). On the other hand, Ca²⁺ modulate PGC-1α by a mechanism dependent on Ca²⁺/calmodulin-dependent protein kinase (CaMK) and p38 protein kinase, also leading to mitochondrial biogenesis (Wright et al., 2007). The modulation of mitochondrial biogenesis by cAMP initiates with the activation of protein kinase A (PKA), followed by the phosphorylation of cAMP response element-binding protein (CREB), that leads to the expression of PGC-1α (Bouchez and Devin, 2019). Mitochondrial biogenesis alters the number, size and mass of the organelles, allowing bioenergetic improvements to respond to stressful conditions (Jornayvaz and Shulman, 2010).

Mitochondrial fusion (combination of two separate mitochondria leading to the formation of a large organelle) and fission (mitochondrial fragmentation leading to the formation of separate organelles) lead to alterations in the number, form, and size of mitochondria (Shen et al., 2023) (Figure 3). Mitochondrial fusion initiates after the interaction of two mitochondria and the fusion of the OMM in an event that is coordinated by mitofusins 1 and 2 (MFN1 and MFN2, respectively), which are dependent of guanosine triphosphate (GTP) (Dorn, 2020). After the fusion of the OMM, the IMM of each mitochondria initiates their combination in a process dependent on optic atrophy 1 protein (OPA1), another GTPase (Tilokani et al., 2018; Shen et al., 2023). Cardiolipin, a phospholipid present at high concentrations in the IMM, cooperates with the OPA1-dependent fusion of the IMM (Ban et al., 2018). Mitochondrial fusion is stimulated, for example, when a mitochondrion begins to present dysfunctions (e.g., decreased ATP production, increased production of reactive species) or increase in the levels of markers of chemical damage (such as mtDNA damage and lipid peroxidation) (Chen et al., 2023). Thus, a healthy mitochondrion (which does not present high levels of markers of molecular damage, therefore) fuses with a dysfunctional one, distributing diverse molecular components and preventing the progression of organellar dysfunction (Chen et al., 2023).

On the other hand, mitochondrial fission leads to the separation of one mitochondrion in two mitochondria (Shen et al., 2023) (Figure 3). The endoplasmic reticulum (ER) cause the constriction of the mitochondrion by a mechanism involving nucleation and polymerization of actin at the contact sites between the organelles (de Brito and Scorrano, 2010; Friedman et al., 2011; Ji et al., 2015; Manor et al., 2015). After this step, mitochondrial fission 1 protein (FIS1) recruits dynamin-related protein 1 (DRP1), which oligomerizes and forms a ring-shaped structure around the mitochondrion (Sun et al., 2024). GTP is cleaved and dynamin 2 (DNM2) is recruited to the constriction site, leading to the fission of the OMM (Sun et al., 2024). The fission of the IMM seems to be dependent on Ca²⁺ and cardiolipin and is mediated by DRP1 in a GTP-dependent manner (Paradies et al., 2019). Two smaller mitochondria are generated after the mitochondrial fission process is completed. Stimulating mitochondrial fission is important because less functional portions of a mitochondrion are eliminated, preventing them from remaining in the organelle, which would cause, among other effects, an increase in the production of reactive species and, consequently, in the levels of redox stress markers (Chen et al., 2023). Smaller and dysfunctional mitochondria, originating from mitochondrial fission, can be degraded in the phenomenon of mitophagy (Kraus et al., 2021).

Mitochondrial fission may or not be followed by mitophagy, a phenomenon in which damaged mitochondria is digested by the cell (Wang et al., 2023) (Figure 4). Loss of MMP, mtDNA damage, augmented reactive species production, and accumulation of misfolded proteins, for example, trigger mitophagy by a mechanism associated with the accumulation of serine/threonine PTEN-induced putative kinase 1 (PINK1) in the OMM, followed by its transautophosphorylation at Ser228 (Gan et al., 2022). Then, PINK1 phosphorylates Parkin, an E3 ubiquitin protein ligase, and monomeric ubiquitin, which, in turn, promotes amplification of the activation of Parkin (Harper et al., 2018). Activated Parkin ubiquitinates proteins located in the OMM, including the voltage-dependent anion channel 1 (VDAC1), mitochondrial Rho-GTPase 1 (Miro1), and proteins involved in the control of mitochondrial fusion, among others (Birsa et al., 2014; Tanaka et al., 2010; Wang et al., 2023). This event leads to phosphorylation of these proteins by PINK1 and to the amplification of the recruitment of Parkin to the mitochondria, causing an increase in the number of ubiquitin chains in the organelles (Harper et al., 2018). This sequence of chemical modifications allows the interaction of the mitochondria with the autophagy adapters, including sequestosome 1 (p62/SQSTM1), nuclear dot protein 52 (NDP52/CALCOCO2), and optineurin (OPTN), to cite a few (Matsumoto et al., 2011; Thurston et al., 2009; Wong and Holzbaur, 2015). The adapters mediate the interaction of ubiquitinated mitochondria with the phagophores by a mechanism related to microtubule-associated protein 1A/1B-light chain 3 (LC3) present in the membranes of these structures (Johansen and Lamark, 2020). The formation of the autophagosomes, which are double-membrane structures, leads to the fusion of the lysosomes with ubiquitinated mitochondria and the digestion of the organelles (Wang et al., 2023). Alternatively, mitophagy can be triggered by mechanisms independent on the PINK1/Parkin pathway (Wang et al., 2023).

In addition to being the most important source of ATP for animal cells, mitochondria constantly produce reactive oxygen species (ROS) through ETC (Masuda et al., 2024). Among the ROS, the most prominent are superoxide radical (O₂ ^−•), hydrogen peroxide (H₂O₂), and hydroxyl radical (^•OH) (Lambert and Brand, 2009; Murphy, 2009; Sies et al., 2022). Furthermore, there is data showing that mitochondria can generate reactive nitrogen species (RNS), such as nitric oxide (NO^•) and peroxynitrite (ONOO⁻), for example, (Lacza et al., 2009; De Armas et al., 2019). Mitochondrial ROS and RNS exert physiological roles in modulating signaling pathways associated with the control of bioenergetics, stress response, hypoxia adaptation, immune response, and cell death (Sena and Chandel, 2012; Sies et al., 2022; 2024). Elevated levels of ROS and/or RNS can not only cause mitochondrial dysfunction, but also widespread damage to cellular components, including proteins, lipids, and nucleic acids (Sies et al., 2017; Cardozo et al., 2023). The antioxidant defenses present in mitochondria can be of the non-enzymatic (such as glutathione - GSH) and enzymatic types (Mailloux, 2018; Marí et al., 2020; Chen et al., 2024). The latter include the enzymes manganese-dependent superoxide dismutase (Mn-SOD), catalase (CAT), different isoforms of glutathione peroxidase (GPx), and peroxiredoxin (PRx) (Cao et al., 2007; Holley et al., 2011; Huang et al., 2019).

Mitochondria control cell death primarily by integrating stress signals into regulated changes in mitochondrial membrane integrity, metabolism, and signaling output. The central mitochondrial mechanism governing apoptosis is mitochondrial outer membrane permeabilization (MOMP), a process tightly regulated by the B-cell lymphoma 2 (BCL-2) protein family (Bock and Tait, 2020; Czabotar and Garcia-Saez, 2023; Glover et al., 2024). In response to genotoxic, metabolic, or oxidative stress, pro-apoptotic BH3-only proteins activate the multidomain effectors BCL-2-associated X protein (BAX) and BCL2 homologous antagonist/killer (BAK), promoting their oligomerization within the outer mitochondrial membrane and formation of proteolipidic pores (Li et al., 2024; Vogler et al., 2025). MOMP enables the release of cytochrome c and other intermembrane space proteins, including second mitochondria-derived activator of caspase/direct inhibitor of apoptosis-binding protein with low pI (SMAC/DIABLO, respectively), which together activate the apoptosome [apoptotic protease activating factor 1 (APAF1)/caspase-9 complex) and downstream executioner caspases-3 and -7 (Figure 5). While caspases orchestrate the biochemical dismantling of the cell, mitochondrial dysfunction following widespread MOMP (including loss of membrane potential, impaired OXPHOS, and ATP depletion) can itself drive caspase-independent cell death, underscoring mitochondria as the point of irreversible cellular collapse (Bock and Tait, 2020; Dho et al., 2025).

Beyond canonical apoptosis, mitochondria modulate cell fate through quantitative control of MOMP. Sublethal or “minority” MOMP affects only a subset of mitochondria, generating limited caspase activity insufficient to cause immediate death (Liu et al., 2025). This partial mitochondrial permeabilization can induce DNA damage, cellular senescence, or stress-adaptive phenotypes, highlighting mitochondria as regulators of death signaling intensity rather than simple on/off switches (Hall-Younger and Tait, 2025).

Mitochondria also control inflammatory forms of cell death by releasing mitochondrial-derived damage-associated molecular patterns (DAMPs). Following extensive membrane remodeling, mitochondrial DNA, cardiolipin, and ROS can escape into the cytosol, activating innate immune pathways such as cyclic GMP-AMP synthase/stimulator of interferon genes (cGAS/STING, respectively), inflammasomes, and type I interferon signaling (Lin et al., 2022; Glover et al., 2024). Thus, mitochondrial permeabilization links cell death execution to inflammatory signaling.

Collectively, mitochondria govern cell death through coordinated regulation of membrane permeabilization, bioenergetic collapse, caspase activation, and immunogenic signaling, positioning them as central decision-making hubs in cellular fate control.

The evidence summarized in Table 1 and in Figure 6 highlights the multifaceted actions of DMF on mitochondrial function and redox biology, with implications for neuroprotection and the treatment of neurodegenerative disorders. Early cellular studies, such as those in dopaminergic CATH. a neurons exposed to tetrahydrobiopterin [BH4, which undergoes autoxidation and enhances the rate of dopamine oxidation and the formation of quinone proteins, which are toxic to brain cells by a mitochondria-dependent manner (Berman and Hastings, 1999; Jana et al., 2011)] demonstrated that DMF enhances respiratory chain activity, particularly Complexes I and IV, while suppressing cytochrome c release (Choi et al., 2006). These findings suggest a dual role in preserving bioenergetic output and preventing apoptosis. The similarity of the effects of DMF with those of N-acetylcysteine (NAC) raises the possibility that its protective mechanism involves redox-sensitive signaling, possibly mediated by GSH metabolism. However, these results are limited by the acute exposure design, which may not fully capture the dynamics of chronic neurodegenerative processes. Moreover, it can be speculated that DMF promoted mitochondrial protection by a mechanism associated with the upregulation of NQO1, a quinone reductase that can metabolize toxic quinones into less reactive agents (Ross and Siegel, 2021). Nonetheless, future research should combine NQO1 knock down with quinone flux analysis to confirm this speculation.

Studies using disease-relevant models further advance the understanding of the mechanisms induced by DMF. In Machado-Joseph disease (MJD) neuroblastoma cells, DMF improved O₂ flux, ATP-linked respiration, and maximal uncoupled capacity while reducing mitochondrial superoxide production (Wu et al., 2022). These improvements were accompanied by upregulation of the Nrf2/NQO-1 pathway, reinforcing the notion that DMF acts as a pharmacological activator of antioxidant defenses. Yet, the comparison with the curcumin analog JM17, which exerted stronger antioxidant effects at lower concentrations, suggests that DMF may not be the most potent modulator of mitochondrial oxidative stress, highlighting the need for structure-activity studies to optimize therapeutic analogs.

Patient-derived iNeurons carrying C9orf72 mutations, a model of ALS and frontotemporal dementia, responded to DMF with reduced mitochondrial reactive oxygen species (mtROS) and upregulation of Nrf2 (Au et al., 2024). While these findings align with previous results, they leave unresolved whether Nrf2 activation is causal or correlative in mitochondrial protection. The absence of functional bioenergetic assessments in this study limits its interpretability, particularly since redox improvement does not always equate to enhanced mitochondrial output.

In fibroblasts from Friedreich’s ataxia patients, DMF not only increased Nrf2, NQO-1, and GSH-related enzymes but also induced frataxin expression (Petrillo et al., 2019). The restoration of frataxin, a key protein in mitochondrial iron-sulfur cluster assembly (Campbell C. J. et al., 2021), suggests a unique disease-modifying potential of DMF in this context. Importantly, DJ-1 (Parkinsonism associated deglycase, also known as PARK7), another redox-sensitive protein whose levels are decreased in Friedreich’s Ataxia and contribute to a defective Nrf2-dependent signaling (Smith and Kosman, 2020), remained unaffected, indicating that protective scope stimulated by DMF may be selective rather than global across redox modulators. Nonetheless, the study lacked direct measurements of respiratory chain function, leaving unclear whether frataxin induction translates into functional mitochondrial recovery.

Animal models of Friedreich’s ataxia further support the benefits caused by DMF, with treatment increasing frataxin, aconitase activity, and respiratory chain components in multiple tissues (Hui et al., 2021). Interestingly, while some models showed increases in mitochondrial DNA (mtDNA) copy number, others did not, suggesting a context-dependent regulation of mitochondrial biogenesis. Moreover, the failure to consistently modulate cytochrome c levels across tissues underscores that the effects promoted by DMF are not uniform. This variability raises important questions about tissue-specific responses, which may have therapeutic implications given the differential vulnerability of brain regions in neurodegenerative diseases.

In a model of X-linked adrenoleukodystrophy (Abcd1^−/− mice), DMF exhibited a broader scope of action, not only upregulating Nrf2 and mitochondrial biogenesis-related genes (Sirt1, Ppargc1a, Nrf1, Tfam) but also increasing ATP production and attenuating inflammatory gene expression (Ranea-Robles et al., 2018). This dual effect on mitochondrial protection and inflammation is particularly relevant, as neurodegenerative disorders are increasingly recognized as disorders of immunometabolic dysfunction (Lonkar et al., 2025). However, it remains unclear whether mitochondrial preservation drives anti-inflammatory effects or whether both outcomes occur in parallel through independent signaling pathways.

Across acute toxicological insults and chronic degenerative milieus, DMF has been consistently associated with improvements in cellular redox balance and selected mitochondrial-related parameters, alongside modulatory effects on inflammatory gene networks. While activation of the Nrf2 pathway represents a well-characterized mechanism in several experimental settings, its direct involvement is not uniformly demonstrated across all models, and in many cases mitochondrial functional endpoints are only partially assessed. Accordingly, reported effects of DMF on mitochondrial homeostasis often rely on indirect redox-sensitive readouts or transcriptional changes rather than comprehensive bioenergetic or dynamic measurements. In this context, auxiliary pathways, including PGC-1α-, SIRT1-, and antioxidant enzyme–related networks (e.g., HO-1 and glutathione metabolism), may contribute to the observed phenotypes in a model- and tissue-dependent manner. The heterogeneity of experimental designs and outcome measures underscores the need for more integrative approaches to delineate whether these pathways operate in parallel, sequentially, or independently of canonical Nrf2 signaling.

Across neural and peripheral systems, DMF is recurrently associated with the activation of redox-sensitive transcriptional programs that include gene networks classically linked to mitochondrial biogenesis and, in specific contexts, to quality control (Figure 7; Table 2). Rather than uniformly inducing demonstrable expansion of mitochondrial mass, turnover, or function, these responses more consistently reflect engagement of upstream regulatory circuits (i.e., transcriptional or copy-number readouts), most notably those centered on Nrf2, that are commonly interpreted as preparatory or permissive for mitochondrial adaptation. This coherence, therefore, becomes appreciable only when disparate experimental models are interpreted along a shared interpretative framework, rather than a fully unified mechanism.

For instance, in rat oligodendrocyte progenitors, DMF elicited a transient pro-oxidant pulse, characterized by increased mitochondrial O₂ ^−• production and reduced MMP, followed by induction of Nrf2, PGC-1α, TFAM, COX1, Mn-SOD, and HO-1 (De Nuccio et al., 2020). The fact that 4-HT (Mn-SOD mimetic) and ZnPP-IX (HO-1 inhibitor) attenuate cytoprotection suggests that low-level oxidative cues are not collateral stress but may function as instructional signals that engage redox-sensitive transcriptional regulators linked to mitochondrial adaptation, rather than unequivocal drivers of mitochondrial biogenesis per se. While this coordinated gene induction is compatible with activation of a biogenic program, the available data primarily substantiate transcriptional engagement rather than direct increases in mitochondrial content or sustained bioenergetic capacity. Such a signaling-based interpretation aligns with the established sensitivity of oligodendroglial cells to redox fluctuations during differentiation and myelin maintenance (Magalhães et al., 2017; Narine and Colognato, 2022).

When extended to human fibroblasts, mouse skeletal muscle and cerebellum, and peripheral blood mononuclear cells (PBCMs) from patients with multiple sclerosis, DMF treatment is similarly associated with increased mtDNA copy number, coordinated induction of nuclear- and mitochondrial-encoded genes (including TFAM, MT-ND2/ND6, SDHA/B, MT-CYB, MT-CO1/2, ATP5B, and MT-ATP6), and enhanced O₂ consumption (Hayashi et al., 2017). Collectively, these endpoints are frequently interpreted as evidence of mitochondrial biogenesis; however, in the absence of ultrastructural quantification (e.g., transmission electron microscopy - TEM) or direct assessment of mitochondrial protein synthesis, they are more conservatively viewed as reflecting transcriptional and genomic remodeling. Such changes may represent mitochondrial priming or compensatory gene-expression responses rather than unequivocal expansion of the mitochondrial network.

A similar pattern emerges in the monosodium iodoacetate (MIA)-induced spinal cord injury model, in which DMF increased PGC-1α, NRF1, TFAM, and mtDNA in an Nrf2-dependent manner (Gao et al., 2022). These findings indicate that the transcriptional architecture classically associated with mitochondrial biogenesis remains responsive, even under inflammatory and metabolic stress. Whether these transcriptional changes translate into sustained increases in mitochondrial content or functional throughput under physiological load, however, remains unresolved. Given that Nrf2 regulates HO-1, an enzyme implicated in both mitochondrial signaling (Hull et al., 2016; Navarro et al., 2017) and anti-inflammatory responses (Campbell N. K. et al., 2021), clarifying the role of the Nrf2/HO-1 axis becomes not merely an adjunct question but a necessary step in determining whether DMF’s effects are mechanistically unified across neural injury paradigms.

A similar integrative perspective bridges findings from aging-related paradigms, DMF increased mtDNA copy number, reduced damaged mtDNA, and activated SIRT1, Nrf2, and SOD2 across multiple brain regions (Sadovnikova et al., 2021). Rather than directly establishing de novo mitochondrial biogenesis, these outcomes support a role for DMF in reinforcing mitochondrial genome integrity and antioxidant defenses. Such effects are consistent with modulation of mitochondrial maintenance and fidelity, processes that may indirectly favor functional resilience without necessarily increasing mitochondrial abundance. Yet, without functional assays (e.g., respiration, coupling efficiency, and ROS yield), such structural and genomic improvements remain inferential; thus, morphological quantification and performance testing remain indispensable to confirm mitochondrial biogenesis.

Insights into mitochondrial turnover arise from dopaminergic injury models, where DMF upregulated Nrf2, PINK1, Parkin, BNIP3, Beclin-1, LC3A/B-II, and BCL-2 in the absence of NRF1 induction (Pinjala et al., 2024). This expression profile is compatible with preferential engagement of signaling nodes associated with mitophagy-related pathways rather than coordinated activation of canonical biogenic machinery, i.e., initial removal of dysfunctional mitochondria may precede or condition any subsequent biogenic response (mitochondrial re-population). However, without time-resolved analyses or direct measurements of mitophagic flux, these molecular changes are more appropriately interpreted as modulation of mitophagy-associated signaling rather than confirmation of enhanced mitochondrial turnover.

The dissociation between transcriptional or proteomic remodeling and functional recovery is particularly evident in severe stress models. In piglet cortex following in-hospital cardiac arrest (IHCA), DMF increased subunits of mitochondrial respiratory complexes and morphology-related proteins without restoring mitochondrial function (Piel et al., 2024). In contrast, porcine cardiac tissue from the same paradigm exhibited improvements in mitochondrial network organization and ultrastructural features. This dissociation underscores that proteomic enrichment of mitochondrial components does not necessarily equate to functional recovery, particularly under conditions of severe energetic collapse.

Translational complexity is further underscored by studies in Friedreich’s ataxia. In patient-derived cells, DMF increases frataxin expression and mtDNA content (Jasoliya et al., 2019), findings that are consistent with transcriptional or genomic modulation but insufficient to confirm induction of mitochondrial biogenesis or involvement of Nrf2 signaling. However, short-term in vivo treatment fails to reproduce these effects in YG8LR mice (Abeti et al., 2022), highlighting the influence of pharmacokinetics, tissue context, and temporal dynamics on the engagement of nuclear-encoded mitochondrial programs.

Taken together, these heterogeneous findings support an interpretative framework in which DMF preferentially engages transcriptional and signaling programs associated with mitochondrial biogenesis, antioxidant defense, and turnover in a context-dependent manner. Yet functional rescue remains variable across models, emphasizing that durable neurological benefit will depend on mechanistic clarification, exposure optimization, and adoption of rigorous flux-based methodologies capable of distinguishing molecular signatures from true mitochondrial recovery. Progression beyond priming toward fully realized mitochondrial remodeling appears variable and frequently unverified by direct functional or structural criteria, underscoring the need for integrative, flux-based validation strategies.

Considering the conceptual continuity that emerges across these diverse models, future research priorities can be articulated in an equally integrated manner. Multi-omics profiling should be coupled with ultrastructural analyses (e.g., TEM) to differentiate genuine mitochondrial biogenesis from transcriptional priming. Mechanistic causality should be tested through genetic (e.g., CRISPR-Cas9 knockouts) or pharmacologic disruption of Nrf2, HO-1, and SIRT1, thereby elucidating whether the shared signatures observed across tissues indeed converge upon a unified regulatory axis. Moreover, defining DMF’s hormetic window through dose-time matrices quantifying ROS dynamics, MMP fluctuations, and downstream transcription would anchor mitochondrial adaptation within measurable signaling thresholds. Flux-level assays, including mt-Keima or mito-QC for mitophagy, MitoTimer for turnover, ^35S-methionine or puromycin labeling for nascent mitochondrial translation, and TFAM-mtDNA occupancy (ChIP), should replace static proxies wherever possible. Complementary high-resolution respirometry under substrate-inhibitor titrations will further resolve how DMF remodels ATP production, proton leak, and redox balance.

Finally, in vivo studies must incorporate longitudinal designs, BBB-exposure measurements, and cell-type–resolved analyses, especially given the broad preclinical dose range (10–300 mg kg⁻¹). Bayesian pharmacokinetic modeling, informed by fumarate hydrolysis and distribution kinetics, may refine dose–response extrapolation to humans. Comparative analyses of DMF, MMF, and next-generation fumarate esters may ultimately reveal whether mitochondrial benefits and tolerability can be decoupled or optimized for distinct diseases such as MS, PD, ischemia-reperfusion injury, and FA.

Building on the evidence that DMF reshapes mitochondrial quantity and quality through tightly regulated programs of biogenesis and turnover, a broader conceptual role for mitochondria emerges. Rather than acting as passive targets of stress, mitochondria function as integrative signaling hubs that decode redox, metabolic, and inflammatory cues into adaptive cellular decisions. Transient oxidative signals are not merely buffered but translated into coordinated transcriptional, structural, and quality-control adjustments that recalibrate mitochondrial performance under stress. Crucially, these adaptive programs position mitochondria at the crossroads between survival and death. The same organelle network that expands or renews itself to sustain bioenergetic and redox homeostasis also harbors the molecular machinery governing mitochondrial outer membrane permeabilization, cytochrome c release, and caspase activation. Thus, mitochondrial remodeling and apoptotic control are not separable phenomena but interdependent layers of a unified stress-response architecture. Within this framework, mitochondrial signaling emerges not only as a determinant of cellular resilience, but also as a decisive regulator of intrinsic apoptotic pathways, directly linking adaptive mitochondrial states to the molecular execution of cell death.

As observed across neural and peripheral models, DMF consistently attenuates mitochondria-dependent apoptosis by stabilizing mitochondrial function and reprogramming intrinsic death checkpoints. At the molecular level, this protection is reflected in preservation of mitochondrial membrane potential, rebalancing of BCL-2 family proteins, limitation of cytochrome c release, and suppression of downstream caspase activation. These effects position mitochondria not merely as targets of injury, but as active arbiters of whether stressed cells undergo recovery or engage the intrinsic apoptotic cascade (Figure 8; Table 3).

In 158N oligodendrocytes challenged with oxysterols, DMF stabilized MMP, curtailed O₂ ^−• and H₂O₂ production, limited poly (ADP-ribose) polymerase (PARP) and caspase-3 activation, and reduced nuclear fragmentation (Zarrouk et al., 2017; Sghaier et al., 2019). In human dopaminergic SH-SY5Y cells exposed to Aβ_1-42, DMF stimulated Akt signaling, restrained cytochrome c release, downregulated BAX, and blocked caspases-9 and -3, despite minimal impact on stress kinases p38 and c-Jun N-terminal kinase (JNK), underscoring pathway selectivity upstream of mitochondria (Rajput et al., 2020). These findings highlight selective modulation of mitochondrial checkpoints upstream of executioner caspases. In vivo, DMF lessened apoptotic signaling after spinal cord injury (SCI) (Cordaro et al., 2017), traumatic brain injury (TBI) (Gao et al., 2024), pentylenetetrazol (PTZ)-induced seizures (Singh et al., 2019), rotenone toxicity (Khot et al., 2023), and chronic cerebral hypoperfusion (CCH) (Shavakandi et al., 2022), with convergent decreases in BAX, cleaved caspase-3, AIF, and Fas ligand, and increases in BCL-2, collectively indicating modulation of intrinsic apoptotic thresholds.

Redox regulation emerges as a central modulatory theme. DMF stimulated antioxidant defenses (SOD, CAT, GPx-1, and Mn-SOD), restores cardiolipin, and reduces mitochondrial O₂ ^−• generation, thereby limiting MOMP and caspase engagement (Sghaier et al., 2019; Cordaro et al., 2017). The increase in cardiolipin is mechanistically notable because it anchors respiratory chain complexes and modulates cytochrome c mobilization, offering a direct route to constrain apoptosis (Li et al., 2025). Contextual variability remains evident, however, as DMF reduces the LC3-II/LC3-I ratio and mitochondrial fragmentation in oxysterol-treated oligodendrocytes, a pattern compatible with suppression of autophagy- or mitophagy-associated signaling under acute oxidative stress (Sghaier et al., 2019). Such suppression could be protective during acute oxidative insults but might prove maladaptive if damaged mitochondria require clearance, highlighting the need for precise temporal mapping of mitophagy flux.

Nrf2 signaling repeatedly emerges as a central, though not exclusive, modulatory axis linking redox balance, inflammation, and mitochondrial survival. DMF-induced increases in Nrf2 and downstream effectors such as HO-1 and NQO-1 are accompanied by suppression of NF-κB-dependent cytokines (TNF-α, IL-1β) and inflammatory mediators [myeloperoxidase (MPO)/iNOS] (Cordaro et al., 2017; Gao et al., 2024; Shavakandi et al., 2022). Inhibition of Nrf2 nuclear translocation abrogates several of these effects (Shavakandi et al., 2022), supporting a contributory role for this pathway. Nonetheless, Nrf2 signaling should be viewed as a central modulatory axis rather than a singular causal determinant of mitochondrial protection. Additional associations with TP53-induced glycolysis and apoptosis regulator (TIGAR) and lysosome-associated membrane protein 2 (LAMP2) expression (Khot et al., 2023), as well as increased neurotrophic factor [brain-derived neurotrophic factor (BDNF), glial cell line-derived neurotrophic factor (GDNF), neurotrophin-3 (NT-3)] levels and benefits from DMF-preconditioned mesenchymal stem cells (Cordaro et al., 2017; Babaei et al., 2023), further suggest coordinated modulation of metabolic, lysosomal, and survival pathways that converge on mitochondrial stability.

Despite these convergent trends, substantial heterogeneity persists across models with respect to dose (1–50 µM in vitro; 12.5–100 mg/kg in vivo), timing (pretreatment vs. post-injury), cell type (oligodendrocytes vs. dopaminergic-like vs. mixed brain regions), and outcome measures. For instance, in hippocampal PTZ experiments, executioner caspase activity was reduced without clear shifts in upstream BCL-2 family balance (Singh et al., 2019), whereas in others (SH-SY5Y cells) mitochondrial protection occurs independently of stress-kinase modulation (Rajput et al., 2020). These differences emphasize that DMF biases intrinsic apoptotic thresholds through multiple, context-dependent nodes rather than through a single invariant mechanism.

Collectively, the available evidence positions DMF as a potent modulator of mitochondria-associated apoptotic signaling in neural systems, acting through integrated redox, inflammatory, and intrinsic death-regulatory pathways. While the protective signature is compelling, precise delineation of causal mechanisms and temporal dynamics remains essential to optimize therapeutic translation in neurodegenerative and acute CNS injury settings.

Regarding future investigations, rigorous causality tests are needed, such as genetic Nrf2 (and Keap1) loss-of-function across neural subtypes, live-cell MOMP imaging and cardiolipin tracking, and true mitophagy-flux assays (e.g., mito-Keima) to define when DMF preserves or impairs mitochondrial quality control. Mechanistic dissection of TIGAR and LAMP2 as mediators rather than correlates, and of the role of Akt in blocking cytochrome c release, is warranted. Region-resolved in vivo studies combining neuroinflammation readouts with mitochondrial endpoints, alongside pharmacokinetics/brain penetration and dose–response mapping, will sharpen translational predictions. Finally, testing DMF with pro-mitophagic or neurotrophic co-therapies may reconcile protection with necessary mitochondrial turnover.

DMF demonstrates how controlled mitochondrial dysfunction can be therapeutically harnessed to reshape immune cell fate. In RRMS, clinical benefit correlates with a mitochondria centric shift in T cell metabolism: routine dosing decreases relapse frequency while dampening oxidative phosphorylation (Gold et al., 2012; Liebmann et al., 2021). High resolution respirometry in patient derived CD4⁺ and CD8⁺ lymphocytes showed that DMF lowers basal and maximal respiratory capacity, collapses spare glycolytic reserve, and remodels cristae architecture, culminating in cytochrome c release and caspase −9 and −3 activation (Liebmann et al., 2021) (Figure 9; Table 4).

Paradoxically, DMF hyperpolarized mitochondria while shrinking organelle mass, implying a compensatory tightening of the proton gradient despite curtailed electron flux. This redox imbalanced state amplified O₂ ^−• production, yet it is readily reversed by GSH, NAC or mitoquinone mesylate (mitoQ). Rescue of redox poise not only reinstated oxidative phosphorylation but also blocked DMF induced lymphocyte quiescence, highlighting a critical role for redox and/or thiol dependent signaling (Zhang and Forman, 2012; Kukulage et al., 2022; Lin et al., 2024; Vašková et al., 2023). Besides, this data implicates mitochondrial oxidative damage as key mechanistic features of DMF toxicity in immune cells. Measuring GSH pools before therapy may therefore forecast patient sensitivity to DMF.

An analogous metabolic checkpoint operates in innate immunity. In LPS stimulated human microglial clone 3 cells, DMF suppressed iNOS and the pro-inflammatory cytokines IL-1β, IL-6 and TNF-α, aligning with its anti-inflammatory profile, while preventing the surge in both glycolytic flux and O₂ consumption (Sangineto et al., 2024). This anti-inflammatory effect coincided with impaired mitochondrial respiration, reduced Complex I activity, and decreased reactive species production and lipid peroxidation markers, reflecting a broad dampening of mitochondrial bioenergetics. Notably, no compensatory activity was observed in Complexes II or V, suggesting a non-redundant impairment in mitochondrial electron transport chain function–effects that may be exploited to temper microglial overactivation in neurodegeneration.

Systems level evidence positions the transcription factor Nrf2 at the apex of this response. Beyond antioxidative gene induction, Nrf2 regulates mitochondrial biogenesis, fusion and mitophagy, thereby synchronizing organelle architecture with fluctuating energetic demand (Luchkova et al., 2024). Whether DMF signals through Nrf2 to orchestrate mitochondrial quality control alongside redox homeostasis remains a pressing mechanistic question.

Clinically, the lymphopenia occasionally observed during DMF therapy most likely reflects the same mitochondrial stress response that underpins efficacy (Dinoto et al., 2022). Stratification based on respiratory reserve, GSH/GSSG ratio or Nrf2 dependent gene signatures could predict benefit to risk ratios and guide adjunctive thiol supplementation. In neurodegenerative settings, carefully titrated DMF might recalibrate microglial metabolism without broad immunosuppression, offering neuroprotection through metabolic modulation.

In summary, DMF illustrates that calibrated impairment of mitochondrial bioenergetics can attenuate immune cells activation. In that regard, off-target effects on mitochondrial function merit careful consideration, since the redox-sensitive nature of mitochondria renders them particularly vulnerable to electrophilic stressors such as DMF, especially in non-pathological tissues. Future works should quantify drug induced shifts in mitochondrial GSH, delineate Nrf2 centric transcriptional circuits, and map complex specific respiratory changes across immune subsets, thereby refining the therapeutic window of DMF and inspiring novel mitochondria targeted interventions.

Martinez et al. (2024) reported the effects caused by DMF in mouse primary cerebral vascular smooth muscle cells, alone and with TNF-α, revealing a coordinated shift from mitochondrial to glycolytic ATP production, increased lactate output under inflammatory stimulus, heightened mitochondrial ROS, apoptosis, and glutathione depletion (higher GSSG/GSH) (Table 5). Notably, pharmacologic inhibition of Nrf2 blunts the glycolytic ATP increase, indicating that part of the metabolic rewiring is Nrf2-dependent. These findings align with an electrophile-driven mechanism in which DMF perturbs redox buffering, transiently activating Nrf2 yet simultaneously imposing mitochondrial impairment, manifesting as ROS elevation and apoptosis, particularly when inflammatory signaling is present. The convergence of redox imbalance (GSH consumption), mitochondrial dysfunction (impaired oxidative ATP synthesis, ROS production), and inflammatory cues (TNF-α-augmented glycolysis) supports a model of stress-adaptation failure, where Nrf2-mediated glycolytic compensation is insufficient to prevent cell death under combined insults.

Methodologically, the use of primary cerebrovascular smooth muscle cells strengthens physiological relevance for the neurovascular unit. Besides, translation requires caution, since exposure windows are acute, concentrations may exceed steady-state levels achieved in vivo, and outcomes were captured at the population level without single-cell resolution. The explicit link between mtROS and downstream apoptosis, while plausible, is correlative here and would benefit from targeted rescue experiments (e.g., mitochondria-directed antioxidants or genetic modulation of mitochondrial quality control).

Future work includes, for example, in vivo validation in neuroinflammation models to test cerebrovascular toxicity and barrier integrity, metabolomics/fluxomics to map the Nrf2-dependent rerouting of carbon and nitrogen metabolism, causal tests of mtROS using site-specific ROS modulators and redox-insensitive Nrf2 constructs, dissection of mitochondrial dynamics and mitophagy as determinants of cell fate, and stratification of inflammatory contexts (e.g., TNF-α versus mixed cytokines) to define interaction surfaces with DMF.

Together, these results position DMF as a dual-edged electrophile, since it triggers Nrf2-linked glycolytic compensation yet concurrently compromises mitochondrial function and redox homeostasis, tipping cells toward apoptosis under inflammatory stress. This mechanistic tension (adaptation versus injury) should guide risk-benefit assessments for DMF in neurovascular settings.

According to the data previously published by different research groups, it is possible to make the following suggestions as future directions in research related to DMF-induced effects on the mitochondrial physiology of CNS cells: -

Therapeutic window mapping by cell type and inflammatory state. Implement dose–time–inflammation matrices across neurons, astrocytes, oligodendroglia, microglia, and cerebrovascular cells, integrating OCR and extracellular acidification rate, ATP production routes, and apoptosis/mitophagy flux assays (PINK1/Parkin, BNIP3/LC3 turnover). Co-administration with redox modulators should be systematically tested to define rescue ranges and liabilities.

Clinical disappointments with generic scavengers (vitamins C/E) highlight a core lesson: CNS redox pathology is not a homogeneous excess of “free radicals” but a compartmental, pathway-specific breakdown involving mitochondria-immune crosstalk. Emerging directions (e.g., mitochondria-targeted antioxidants, Nrf2 activation, modulation of NOX/NOS, nano-delivery, and gene/epigenetic interventions) aim to restore networked redox homeostasis rather than merely quench radicals. Therapeutics that reprogram the Nrf2 axis and mitochondrial resilience should be evaluated with attention to subcellular targeting, patient-specific redox signatures, and inflammatory tone, a precision-redox framework that aligns with current artificial intelligence/multi-omics strategies. DMF exerts a bidirectional influence on mitochondria in CNS-related systems: it frequently activates Nrf2-dependent programs that favor redox resilience, biogenesis markers, and anti-apoptotic signaling in neurons and oligodendroglia, yet it can blunt respiration or promote apoptosis in immune and vascular cells or under heightened inflammation. The totality of evidence argues for a nuanced, context-aware deployment of DMF in neurological disorders, anchored in cell type-specific dosing, stringent functional readouts, and, where appropriate, adjunctive redox buffering to preserve mitochondrial benefits while minimizing liabilities.