Targeting mitochondrial homeostasis as a cancer treatment strategy: current status and future prospects.

Background
Cancer remains a leading cause of mortality worldwide, with therapeutic efficiency hindered by drug resistance and tumor progression that demand more innovative therapy strategies [1]. Mitochondria, the central regulators of cellular metabolism, activity, and fate, play critical roles in cancer progression, metastasis, and relapse, making them promising targets for anti-cancer therapy [2].
As key hubs of cellular metabolism, signaling, and fate decisions, mitochondria maintain cellular homeostasis through quality control (QC) mechanisms (e.g., fission, fusion, and mitophagy), metabolic regulation, energy production, and essential processes such as apoptosis [3]. Numerous steps in glucose, lipid, and protein metabolism, including the tricarboxylic acid cycle (TCA), occur within mitochondria, which also house the respiratory complexes that generate an electrochemical gradient across the inner mitochondrial membrane to produce adenosine triphosphate (ATP) [4, 5]. The electron transfer chain especially the complex I in also play pivotal role in regulating mitochondrial homeostasis revealed by mitochondrial proteome [6]. Mechanistically, mitochondrial dysfunction and diseases can be broadly classified into several categories: mitochondria-mediated apoptosis, disrupted mitochondrial metabolism, imbalanced mitochondrial dynamics associated with reactive oxygen species (ROS) and Ca2+ overload, and instability of the mitochondrial genome. Apoptosis is tightly regulated by the Bcl-2 protein family, which includes anti-apoptosis proteins (e.g., Bcl-2, Bcl-xl), pro-apoptosis proteins (e.g., BAX, BAK), and BH3-only proteins (e.g., BID, BAD) [4, 5, 7]. Interactions among these proteins influence the formation of macropores in the mitochondrial outer membrane, triggering cytochrome c release, caspase activation, and apotosis [8, 9]. Mitochondria also regulate ion homeostasis, including calcium, iron, and copper homeostasis, thereby influencing apoptosis, ferroptosis, and cuproptosis [10, 11]. Following publication of the mitochondrial genome in 1981, extensive research has examined mitochondrial DNA (mtDNA), which replicates independently within mitochondria encodes essential respiratory complex subunits such as cytochrome b [12]. Instability of mtDNA contributes to numerous diseases, including Leigh syndrome. mtDNA also functions as a damage-associated molecular patterns (DAMPs) recognized by pattern-recognition receptors (PRRs) such as inflammasomes, cGAS-STING, and TLR9, initiating innate immune repsonses [13]. Mitochondrial clearance through the NF-κB-p62-mitophagy pathway limits inflammasome activation and inflammation, highlighting the role of mitochondria in innate immunity [14]. Intracellular mitochondrial transfer also exists to rescue impaired mitochondrial functions [15, 16]. Notably, mitochondrial-mediated cellular activities intersect with each other. For example, MCL1, a member of the Bcl-2 protein family, together with the MITOK protein and ATP synthase subunit, constitutes an ATP-sensitive potassium channel (mitoKATP), which connects cell apoptosis, ion homeostasis, and energetic status, revealing the complex mitochondrial function in regulating cellular homeostasis [17]. Therefore, mitochondria are deeply involved in cellular activities, and their dysfunction is implicated in several diseases, including mitochondrial myopathy, neurodegenerative diseases, and metabolic syndrome [18–20]. As illustrated in Fig. 1, complex mitochondrial functions in health and diseases have been gradually revealed in the past decade.
In cancer, mitochondrial functions are not inactivated but reprogrammed (e.g., TCA cycle rewiring) to fuel tumor initiation, progression, and metastasis [21]. While previous research mainly focuses on aerobic glycolysis in cancer cells known as “Warburg effect”, not all tumor shares this property [22]. Some tumor possesses high level of oxidative phosphorylation (OXPHOS), other cancer that relatively rely on glycolysis also retain mitochondrial function [23]. Therefore, the metabolism traits in cancer are heterogeneous, which requires to be analyzed specifically based on the cancer subtype and cellular context. Besides, abnormal QC, mtDNA damage, and ion disturbances related to apoptosis and ferroptosis widely exist in cancer cells [24, 25]. Antioxidant systems in mitochondria and adaptive control of cell death also contribute to the survival of cancer cells [26, 27]. In addition to cancer cell itself, mitochondrial dysfunction of cells in the tumor microenvironment (TME) such as immune and cancer stem cells (CSCs) also contributes to tumor growth [28, 29]. Mitochondria dysfunction is closely related to the exhaustion of CD8 + T cells in the TME, which undermines anti-tumor immunity [30]. Multiple traits such as impaired mitochondrial biogenesis, and disturbed metabolism in exhausted CD8 + T cells provide promising targets for anti-cancer therapy [30]. Furthermore, mitochondria are central to chemotherapy resistance, with mechanisms ranging from anti-ROS accumulation (e.g., CHK1-SSBP1 as a ROS-sensing pathway in platinum resistance) to altered death signaling [31, 32].
Given the central roles of mitochondria in cancer cells, the modulation of mitochondrial functions is a promising anticancer treatment strategy. Therapeutic methods range from dietary interventions to small-molecule drugs that regulate mitochondrial homeostasis in cancer or surrounding cells [33, 34]. While innovative concepts such as horizontal mitochondrial transfer and gene editing offer future promise, most strategies remain in preclinical stages and face significant translational challenges due to mtDNA complexity and on-target toxicity [35–37].
In this review, we provide a comprehensive overview of the multifaceted roles of mitochondrial homeostasis and dysfunction in cancer from the perspectives of cell metabolism, QC, mtDNA, ion homeostasis, and the TME. We also summarize the current landscape of mitochondria-targeted anticancer treatment before discussing the translational challenges and future directions necessary to realize the potential of mitochondria in oncology.

Mitochondrial dysfunction and cancer
Mitochondria carry out tightly coordinated functions, including metabolic regulation, QC processes, mtDNA maintenance, and ion homeostasis. Under normal conditions, these systems preserve mitochondrial homeostasis. When this balance is disrupted, cellular bioenergetics and signaling are impaired, promoting malignant transformation (Fig. 2).

Mitochondrial metabolism and cancer

Mitochondrial energy metabolism and signal transduction
Mitochondria act as central hubs linking cellular energy metabolism with signal transduction; they process glucose, lipid, and amino acid substrates to produce energy and generate signaling molecules that regulate cell fate, death, and gene expression [23].
Glycolysis, a key metabolic pathway, converts glucose into pyruvate, producing ATP and nicotinamide adenine dinucleotide (NADH). Anaerobically, pyruvate is reduced to lactate, and aerobically, pyruvate enters the mitochondria, forming acetyl-coenzyme A (CoA) for TCA cycle usage [38]. Pyruvate can be converted to alanine, thereby connecting sugar and protein metabolisms [39]. The eight enzymatic reactions in the TCA cycle generate metabolic intermediates and high-energy electron carriers. Specifically, the cycle produces NADH and FADH2, which donate electrons to the electron transport chain (ETC; complexes I–IV) to drive oxidative phosphorylation (OXPHOS). Additionally, the cycle generates GTP (which can be converted to ATP) through substrate-level phosphorylation [40–44]. Fatty acids are converted to acetyl-CoA via beta-oxidation and enter the TCA cycle. Excess acetyl-CoA can also be used to achieve lipid synthesis for energy storage [45]. Amino acids can enter the TCA cycle through various intermediates, such as oxaloacetate and α-ketoglutarate [46]. For example, glutamate can be converted to α-ketoglutarate, which either enters the TCA cycle or is reconverted to glutamate and subsequently to glutamine, which then plays a critical role in cancer metabolism [47]. The mitochondrial serine-glycine-one-carbon pathway complements the TCA cycle by supplying one-carbon units for nucleotide synthesis and methionine-driven methylation, supporting proliferative signaling [23, 48, 49]. Mitochondria translate this metabolic flux into crucial cellular signals for life-and-death decisions, serving as central hubs in the intrinsic apoptosis pathway. The commitment to mitochondrial apoptosis is precisely regulated by the dynamic equilibrium between pro-apoptotic (e.g., BAX, BAK) and anti-apoptotic (e.g., BCL-2, BCL-XL) proteins of the BCL-2 family. Oligomerization of activated BAX/BAK on the outer mitochondrial membrane leads to mitochondrial outer membrane permeabilization, the irreversible “point of no return” in apoptosis, which results in the irreversible release of cytochrome c and other factors to activate caspases [50, 51]. Bcl-xl can also retro-translocate BAX from mitochondrial to cytosol preventing BAX accumulation on the mitochondrial membrane and subsequent apoptosis [52]. Alternatively, mitochondria undergoing BAX/BAK1/BID-dependent oxidative stress may maintain prolonged plasma membrane contact, leading to local oxidative damage—a process termed mitoxyperiosis. This process can subsequently cause membrane lysis and cell death, referred to as mitoxyperilysis [53]. Furthermore, the mode of release of this mitochondrial content is critical for immune surveillance. In immunogenic cell death, mitochondrial DAMPs, such as cytochrome c, ATP, and mtDNA, are released in a spatiotemporally coordinated manner. These molecules act as potent “danger signals” that recruit and activate antigen-presenting cells, thereby bridging the gap between tumor cell death and the induction of adaptive antitumor immunity [54, 55]. DAMPs also play a key role in the activation of innate immune pathways [56]. In adaptive signaling, ROS generated by the ETC act as redox signals that oxidize thiol groups in regulatory proteins, modulating gene expression under stress [11]. In directing cell fate, mitochondrial-derived acetyl-CoA is a crucial substrate for histone acetylation, establishing an epigenetic landscape that favors an anabolic state and promotes cancer development [40].

Mitochondrial metabolic reprogramming
Mitochondrial metabolic reprogramming is a hallmark of cancer, enabling tumor cells to meet the bioenergetic and biosynthetic demands of rapid proliferation while surviving in harsh microenvironmental conditions like hypoxia [57]. The reprogramming is primarily driven by activated oncogenes (e.g., MYC, PIK3CA, RAS) and mutated tumor suppressor genes (e.g., TP53, PTEN, LKB1) [58–63], which converge on key signaling pathways such as phosphatidylinositol 3-kinase (PI3K)/protein kinase B (Akt) [51], mTOR [50], AMP-activated protein kinase (AMPK) [64, 65], and hypoxia-inducible factor (HIF) [66, 67] to rewire cellular metabolism.

Oncogenic drivers of metabolic reprogramming
The transcription factor MYC acts as a master regulator, broadly enhancing mitochondrial metabolism by stimulating mitochondrial biogenesis, glutamine catabolism, and the expression of glycolytic and TCA cycle genes [58, 68]. In stark contrast, the tumor suppressor TP53 plays a complex, context-dependent role. Wild-type p53 promotes mitochondrial respiration and integrity, while its frequent mutation in cancer leads to a loss of these functions and a shift toward glycolytic dependency [69, 70]. This phenomenon is clinically exemplified by Li-Fraumeni syndrome (LFS), a hereditary cancer predisposition syndrome caused by germline TP53 mutations. Cells derived from patients with LFS display enhanced mitochondrial metabolism, which establishes a permissive bioenergetic environment for tumor initiation. Inhibition of aberrant mitochondrial activity can prolong cancer-free survival in preclinical models. However, while several LFS models demonstrate increased mitochondrial respiration that can be therapeutically targeted, the underlying mechanisms appear to be mutation-specific [71]. Notably, mutations within the p53 oligomerization domain, such as p53(A347D), yield a stable dimeric form that localizes to mitochondria. This mutant p53 confers a gain-of-function, inducing mitochondrial network aberrations and promoting transcription-independent apoptosis under stress. Together, these findings reveal a non-canonical, mitochondria-centered role for mutant p53 that contributes to tumorigenesis yet presents a potential therapeutic vulnerability [72].

Diverse metabolic phenotypes: from glycolysis to OXPHOS
The Warburg effect describes the preference of cancer cells for glycolysis rather than OXPHOS even under oxygen-rich conditions where oxidative OXPHOS would be more efficient. This approach for producing ATP is counterintuitive because glycolysis yields significantly less ATP per glucose molecule than OXPHOS and, therefore, requires markedly increased glucose uptake––a process that is facilitated by the upregulation of GLUT1 expression in cancer cells [57, 73]. This reliance on glucose can be visualized clinically via radiolabeled glucose analogs 18F-fluorodeoxyglucose-positron emission tomography imaging [74, 75]. Lactate, a glycolysis end product, is not merely a waste product but can be used as a fuel source by other cancer cell subpopulations or to stimulate mitochondrial ETC activity, illustrating metabolic symbiosis [57, 76, 77].
While the Warburg effect highlights the role of glycolysis in many cancers, some cancer cell types (e.g., ovarian cancer and glioblastoma) generate most of their energy through OXPHOS [78–81]. This demonstrates that the metabolism of cancer cells is not monolithic but exhibits great diversity and adaptability [82]. This metabolic plasticity can involve reliance on glycolysis, enhanced lipogenesis, altered fatty acid oxidation (FAO), and rewiring of amino acid and nucleotide metabolism, all of which are orchestrated by key signaling pathways such as AMPK, mTOR, and MYC [64, 82, 83]. Beyond these core pathways, higher-order systems, such as the circadian clock, also regulate cellular metabolism. Mitochondrial OXPHOS and ATP generation exhibit circadian rhythms [84, 85]. Oncogenic signals can disrupt core clock genes, thereby dysregulating these metabolic cycles and creating a permissive environment for tumor growth [86]. This layer of temporal regulation further underscores the sophisticated adaptability of cancer cells, allowing them to fine-tune their metabolic output in response to both intrinsic and extrinsic cues.
With the advent of single-cell sequencing technology and the growing recognition of tumor heterogeneity, distinct metabolic preferences among cells within the TME have been revealed. Immune cells, particularly macrophages and T cells, exhibit the highest glucose uptake, whereas tumor cells preferentially use glutamine to sustain their survival and proliferation [83, 87, 88].

Metabolites as signaling and biosynthetic molecules
The use of extracellular nutrients is initiated by their specific uptake into the cell. For instance, glutamine is primarily imported via transporters such as ASCT2 [89]. Once inside the cytosol, glutamine is converted to glutamate, which is then transported across the inner mitochondrial membrane by carriers like the aspartate-glutamate carrier (AGC) to enter the TCA cycle [90]. Within the mitochondria, glutamine-derived carbon skeletons are metabolized to α-ketoglutarate, which not only provides energy but also generates several intermediates that influence cancer progression [23, 40, 76, 91]. For example, α-ketoglutarate, acting as an essential substrate under normoxic conditions, promotes HIF-1α hydroxylation and its subsequent ubiquitin–proteasome degradation by regulating the activity of prolyl hydroxylase domain proteins (PHDs). Accumulation of succinate can inhibit the hydroxylation and degradation of HIF-1α, thereby stabilizing and activating HIF-1α. In the α-ketoglutarate-dependent hydroxylation reaction, P4HA1 (prolyl 4-hydroxylase subunit alpha 1) is another hydroxylase that promotes collagen hydroxylation and extracellular matrix remodeling, playing an important role in cancer metastasis. It can compete with PHD to release the hydroxylation of HIF-1α, thereby stabilizing HIF-1α [76, 92]. Other TCA intermediates also play an important role in tumorigenesis. The mitochondrial pyruvate carrier imports pyruvate derived from glycolytic flux, which is then decarboxylated to acetyl-CoA—a metabolite that can lead to changes in gene expression that support cancer cell proliferation, growth, and migration [40, 75, 93]. Fumarate inhibits DNA and histone demethylation, stabilizes HIF-1α, supports pseudohypoxia, causes protein succination, and promotes epithelial-mesenchymal transition (EMT) in cancer [40, 94]. TCA cycle metabolites and nucleotide synthesis can limit cell proliferation; however, whether ATP and NADPH are such metabolites remain controversial [91].
Beyond their signaling roles, these mitochondrial metabolites provide essential precursors for biosynthetic pathways, including citrate-derived lipid synthesis, aspartate-mediated nucleotide biosynthesis, and amino acid replenishment, highlighting the anabolic functions of mitochondrial metabolism in cancer cells [95, 96]. Citrate exported from mitochondria via the citrate shuttle is cleaved by ATP citrate lyase to generate cytosolic acetyl-CoA, which fuels fatty acid and membrane lipid synthesis required for rapid tumor cell proliferation [95]. Mitochondrial oxaloacetate and glutamine-derived aspartate serve as crucial carbon and nitrogen donors for nucleotide synthesis, linking mitochondrial metabolism to DNA replication and RNA production [96, 97]. Collectively, these processes illustrate how mitochondrial metabolism not only supports energy generation but also supplies the necessary anabolic building blocks for cancer growth, survival, and adaptation to the TME.

ROS and hypoxia: interconnected drivers of metabolic dysregulation
The rewired metabolism of cancer cells, particularly the inefficiency and altered flux of the mitochondrial ETC, is a major source of ROS [98]. ROS are unstable, reactive, and partially reduce oxygen derivatives. Mitochondrial ETCs, especially complexes I and III, are primary sites of ROS generation [99]. Thus, oncogene-driven metabolic reprogramming directly contributes to an altered redox state. In terms of their role, ROS represent a double-edged sword in cancer development [99], which depends primarily on ROS stress. Low ROS levels are important for signal transduction [100, 101], while increased ROS levels activate cancer cell proliferation, migration, invasion, angiogenesis, and drug resistance through HIF and other pathways [99]. For example, elevated ROS levels in liver cancer cells activate mitophagy and PI3K/AKT signaling pathways, thereby promoting cancer progression [102]. Extremely high ROS levels cause DNA damage and genomic instability, leading to neoplastic transformation or cell death [103]. Cancer cells employ sophisticated sensing mechanisms to maintain the precarious balance. For instance, VPS35 can sense chemotherapeutic-induced ROS and dampen mtROS generation by inhibiting mitochondrial protein translation, reducing ROS toxicity in cancer cell nuclei, thereby conferring treatment resistance [354]. Under hypoxic conditions, mitochondrial Complex III–derived ROS stabilize HIF-1α, which dimerizes with HIF-1β to activate hypoxia-responsive genes, such as GLUT1, which enhance glycolysis and glucose uptake for cancer cell survival [104, 105]. Concurrently, ROS induce chromosomal collapse and micronuclear instability, thereby linking tumor hypoxia to genomic instability and cancer progression [106]. Remarkably, certain cancers, such as pancreatic ductal adenocarcinoma, demonstrate extreme metabolic plasticity by maintaining OXPHOS, even under situations of severe hypoxia, to sustain key metabolites for proliferation, contributing to their notorious treatment resistance [107, 108] (Fig. 3).

Mitochondrial quality control (MQC) in cancer
MQC is not a series of isolated events but an integrated network of processes that collectively maintain mitochondrial health and function. This network encompasses mitochondrial biogenesis, fusion, fission, mitophagy, intercellular mitochondrial transfer, and the mitochondrial unfolded protein response (UPRmt). This delicate balance is co-opted in cancer to fuel tumor progression, resist stress, and adapt to the TME.

Core components of MQC machinery
Mitochondrial biogenesis is a process by which cells synthesize new mitochondrial components and integrate them into the existing mitochondrial network to maintain mitochondrial quantity, function, and adaptability in response to energy demands by developmental signals and environmental stressors [109]. Peroxisome proliferator-activated receptor gamma coactivator 1-alpha (PGC-1α) mainly regulates mitochondrial biogenesis [110].
Mitochondrial fusion and fission are important regulatory mechanisms in MQC. Mitochondrial fusion, which is mediated by MFN1/2 (outer membrane) and OPA1 (inner membrane), allows content mixing to dilute damage [111, 112]. Conversely, fission, executed by DRP1, facilitates the segregation of damaged components and the distribution of mitochondria during cell division [113, 114].
Mitophagy is the selective autophagic clearance of damaged mitochondria. It is activated when damage surpasses the capacity of other QC mechanisms or to meet metabolic demands [115]. There are two mitophagy pathways: 1) ubiquitin-dependent, which includes two subtypes, PINK1-Parkin-mediated and Parkin-independent mitophagy (PINK1 directly recruits autophagy receptors, including OPTN and NDP52, to the mitochondria via ubiquitin phosphorylation, initiating mitophagy); and 2) ubiquitin-independent, which utilizes receptors such as BNIP3, NIX, and FUNDC1 to directly bind LC3 without requiring ubiquitination [116]. Furthermore, a new mitophagy mechanism was recently discovered, which functions under oxidative stress and involves lysosome-mediated selective clearance of inner mitochondrial membrane (IMM) for QC [117].
Intercellular mitochondrial transfer extends MQC beyond cell-autonomous boundaries. Mitochondria can be transferred via nanotubes, microvesicles, or direct extrusion to support recipient cell metabolism, regulate immunity, or dispose of damaged organelles [36, 118, 119].
The UPRmt is an adaptive stress response activated by mitochondria in response to protein homeostasis imbalances (including unfolded or misfolded protein accumulation, oxidative damage, and mtDNA mutations) to restore mitochondrial protein homeostasis and function. The UPRmt upregulates chaperones and proteases to restore protein folding and function, which is crucial for maintaining cellular health under stress [120].

MQC as a nexus of oncogenic rewiring
In cancer, the homeostatic MQC network is hijacked to support tumorigenesis. In the following sections, we dissect how individual processes are rewired; however, it is critical to appreciate that their functions are highly context-dependent and influenced by cancer type, stage, and genetic landscape.
PGC-1α and mitochondrial biogenesis in cancer
PGC-1α, a key protein, promotes mitochondrial biogenesis and OXPHOS, increases oxygen consumption in invasive cancer cells, provides ATP to cancer cells, and promotes hypoxia adaptation and metastasis, as evidenced in breast cancer and cholangiocarcinoma [121–124]. Paradoxically, impaired biogenesis can also be selected for, leading to therapy resistance in some contexts, highlighting a complex trade-off [125, 126].

Mitochondrial fission and fusion in cancer
Mitochondrial dynamics, which are governed by the balance between fission and fusion, play a pivotal role in tumor metabolism and progression. Fission, which is often driven by elevated DRP1 (a key GTPase), is generally pro-tumorigenic and has been observed in metastatic breast cancer, lung cancer, colorectal cancer (CRC), pancreatic cancer (PC), glioblastoma, neuroblastoma, and squamous cell carcinoma (SCC) [127]. Elevated DRP1 activity disrupts mitochondrial function and induces mtDNA release, causing cellular stress that drives tumorigenesis [128, 129]. DRP1-induced fission facilitates metabolic reprogramming and chemotherapy resistance, whereas its inhibition can block cell-cycle progression and suppress metastasis [122, 130–132]. By fragmenting mitochondria, DRP1 enables their redistribution to regions of high energy demand and ensures their inheritance during rapid cell division, thereby fulfilling the metabolic and biosynthetic needs of proliferating cancer cells [127].
Conversely, mitochondrial fusion, mediated by MFN1/2 and OPA1, preserves mitochondrial integrity by diluting oxidized components and mutant mtDNA, maintaining membrane potential, and buffering tumor-promoting stress signaling [127, 133]. Most cancers exhibit decreased fusion-related protein expression, and an imbalanced DRP1/MFN1-2 ratio correlates with enhanced proliferation, metastasis, and poor prognosis [122, 134, 135].
However, fusion is not universally tumor-suppressive; in specific contexts, fusion proteins can sustain cancer metabolism. For instance, OPA1 supports metabolic adaptation in residual triple-negative breast cancer cells (TNBCs) after chemotherapy, while MFN2 interacts with PKM2 to regulate the mTOR pathway and promote cancer cell growth [136, 137].

Controversial role of mitophagy in cancer
Mitophagy serves as a prime example of MQC context-dependency, acting as either a tumor suppressor or promoter depending on the cancer type, stage, and microenvironment [138]. Mitophagy eliminates damaged mitochondria to prevent excessive ROS accumulation and inflammasome activation, thereby suppressing tumorigenesis. For example, impaired mitophagy in breast cancer activates the NLRP3 inflammasome, fostering a pro-tumorigenic microenvironment that enhances proliferation and metastasis [139].
However, excessive or dysregulated mitophagy can also sustain cancer progression by maintaining mitochondrial fitness under metabolic stress, enhancing OXPHOS, and conferring drug resistance [140–142]. Mechanistically, mitophagy proceeds through both ubiquitin-dependent (PINK1–Parkin) and ubiquitin-independent (BNIP3, NIX, FUNDC1) pathways, with each exerting distinct effects in cancer [116, 143, 144].
PINK1-mediated mitophagy promotes metastasis and chemoresistance in multiple tumors, including hepatocellular carcinoma and PC, by enhancing mitochondrial bioenergetics and inhibiting apoptosis [140–142]. Conversely, inhibition of PINK1 signaling sensitizes tumor cells to chemotherapy [145, 146]. Although PINK1-mediated mitophagy generally sustains tumor metabolism and survival, it paradoxically promotes PD-L1 degradation in TNBC, reducing immune evasion and improving the efficacy of PD-1/PD-L1 blockade therapy [147].
Parkin, which acts downstream of PINK1, generally functions as a tumor suppressor [148]. Loss or mutation of this protein, which is frequently observed in cancers such as hepatocellular carcinoma and ovarian cancer, results in impaired mitophagy, enhanced PI3K/Akt signaling, and mitotic instability [138, 149]. The restoration of Parkin expression reverses these phenotypes and promotes apoptosis [148].
The ubiquitin-independent mediators BNIP3, NIX, and FUNDC1 exhibit similar context-specific effects. Hypoxia-induced BNIP3 or NIX expression via HIF-1α can drive mitophagy-dependent metabolic adaptation and metastasis in melanoma and PC [150–152], yet in other contexts, their loss promotes tumor progression and therapeutic resistance [153–155]. FUNDC1 displays similarly dual roles, which promote proliferation in endometrial and breast cancers while exerting tumor-suppressive functions in hepatocellular carcinoma by enhancing apoptosis and inhibiting sorafenib resistance [156–159].
In summary, mitophagy exerts both oncogenic and tumor-suppressive effects in cancer, depending on the cellular context, metabolic demands, and tumor microenvironmental cues. Understanding the molecular determinants that dictate this switch remains critical for exploiting mitophagy as a therapeutic target.

Intercellular mitochondrial transfer in cancer
Intercellular mitochondrial transfer has emerged as a critical mechanism shaping tumor progression and antitumor immunity. Cancer cells can acquire intact mitochondria from T cells, macrophages, and even neurons within the TME, thereby boosting their own bioenergetic capacity, while simultaneously impairing immune cell function [36, 160]. The transfer of mitochondria in this way enhances cancer cell metabolic fitness, stress resistance, and invasive potential; for instance, neuronal-to-cancer mitochondrial transfer supports survival during transendothelial migration and facilitates metastasis. This process also contributes to chemoradiotherapy resistance and multidrug tolerance [160, 161]. Conversely, artificially transferring mitochondria from bone marrow mesenchymal cells to CD8 + T cells can enhance the metabolic adaptability and antitumor functions of T cells [162]. Because the transfer is largely mediated by tunneling nanotubes, inhibiting nanotube assembly suppresses aberrant mitochondrial acquisition by cancer cells, thereby preventing immune cell exhaustion. Combining nanotube assembly inhibitor with PD-1 immune checkpoint inhibitor (ICI) improves antitumor efficacy in an aggressive immunocompetent breast cancer model [118].
Notably, mitochondrial transfer is not strictly unidirectional. Beyond acquiring functional mitochondria, cancer cells can transfer mutated mtDNA-carrying mitochondria to tumor-infiltrating T cells. Mitochondria evade mitophagy by delivering inhibitory factors, leading to T-cell metabolic dysfunction, senescence, and impaired antitumor activity, ultimately reducing the efficacy of ICIs [163].

UPRmt in cancer
The UPRmt is conserved and activated in cancer cells in response to mitochondrial stress to maintain mitochondrial integrity and support tumor growth and resistance. Under mitochondrial stress, the UPRmt limits excessive ROS accumulation and prevents apoptosis; however, failure to activate this pathway often results in cell death [164]. In addition to restoring proteostasis, the UPRmt also reprograms cellular metabolism; its translational control mechanisms divert glycolytic intermediates toward serine-glycine-one-carbon metabolism through key regulatory enzymes, thereby enhancing biosynthetic capacity and conferring resistance to folate-targeting chemotherapies such as methotrexate [165]. The UPRmt is not only a local response of mitochondria but also a process that sends signals to the cell nucleus, activating transcription factors such as ATF4, ATF5, and CHOP, thereby driving the expression of pro-survival genes [166, 167]. Moreover, UPRmt-induced mitokines, such as GDF15, can reshape the TME and contribute to immune evasion, further highlighting its multifaceted role in cancer progression and therapeutic resistance [167, 168]. Hence, studies targeting the UPRmt in cancer have been developed [169]. For example, HSP60 interacts with ClpP (a mitochondrial protease) to maintain mitochondrial ATP production and promote prostate cancer cell survival, whereas the UPRmt inhibitor suppresses prostate cancer growth in mice via HSP60 [170].

mtDNA
Human mtDNA is a small, circular, double-stranded DNA molecule that encodes 13 essential subunits of the OXPHOS system, along with 22 tRNAs and two rRNAs required for their translation within the organelle [171]. Therefore, mtDNA integrity is crucial for maintaining normal OXPHOS productivity. However, mtDNA is inherently vulnerable: it lacks the protective histone proteins that shield nuclear DNA and possesses limited DNA repair capacity, primarily relying on base excision repair [171, 355]. These features, combined with its proximity to the primary site of ROS production, render it highly susceptible to mutation. Furthermore, upon cellular damage, mtDNA can be released into the cytosol or extracellular space, where it acts as a DAMP to activate innate immune signaling [40, 172]. This unique combination of vital function and structural vulnerability positions mtDNA as a key player in the pathogenesis of cancer.

Mechanisms underlying mtDNA-driven oncogenesis
The critical role of mtDNA in tumorigenesis is evidenced by its depletion, which severely impairs cancer cell growth and tumorigenicity [173]. mtDNA contributes to cancer progression through the following interconnected mechanisms:
Respiratory dysfunction and the ROS-mutation cycle. As mtDNA encodes core respiratory chain proteins, mutations directly cause OXPHOS dysfunction. This dysfunction often leads to increased electron leakage and ROS production. In a vicious cycle, these ROS molecules further damage the already vulnerable mtDNA, introducing additional mutations and creating a self-perpetuating feed-forward loop that fuels genomic instability and tumor progression [103].
Nuclear epigenetic remodeling. Beyond bioenergetics, mtDNA mutations can reshape the nuclear epigenome. By altering the flux of TCA cycle metabolites—such as α-ketoglutarate, acetyl-CoA, and succinate—that serve as cofactors or substrates for chromatin-modifying enzymes, dysfunctional mitochondria can induce global changes in histone and DNA methylation, thereby driving pro-tumorigenic gene expression programs [174, 175].
Activation of innate immune and inflammatory pathways. The release of mtDNA into the cytosol is a potent trigger of the cGAS-STING pathway, leading to the production of type I interferons and pro-inflammatory cytokines [176]. While this can theoretically mount an anti-tumor immune response, tumors often co-opt this pathway for their benefit. The mitochondrial chaperone Lon, when overexpressed, promotes mtDNA release and STING-dependent upregulation of PD-L1, thereby inhibiting T-cell activation and facilitating immune evasion [177]. Similarly, hypoxic stress can induce the release of mtDNA, which binds to HMGB1 and activates the TLR9 pathway, ultimately promoting tumor cell proliferation [178]. The interplay between mitochondrial dynamics and innate immunity is further highlighted by the finding that the fission protein DRP1 can interact with and sequester cGAS on the mitochondrial membrane, paradoxically suppressing mtROS and ferroptosis to support cancer cell survival [179].
Variations in copy number and content. The quantitative aspects of mtDNA also contribute to tumorigenesis. For example, the variation in mtDNA copy number between cells is associated with metabolic flexibility, allowing cancer cell subpopulations to adapt to varying TME conditions [180]. The ratio of mtDNA to nuclear DNA, rather than the absolute copy number per se, is a critical metric governing downstream phenotypic outputs in tumors [181]. In addition to sequence mutations and copy number changes, the potential transfer of mtDNA sequences into the nuclear genome may contribute to genomic instability in certain cancers [182].

Mitochondrial ion homeostasis: intersection of metabolic signaling and cancer cell death
Mitochondria are central hubs for ion homeostasis, dynamically regulating the flux of calcium (Ca2+), iron (Fe), and copper (Cu) to coordinate energy production, metabolism, and cell survival. The precise regulation of these ions is frequently disrupted in cancer, creating vulnerabilities that tumor cells exploit for growth while simultaneously exposing them to novel forms of cell death.

Calcium signaling pathway
Calcium is a pivotal mitochondrial signaling molecule that regulates the activity of TCA cycle enzymes to support the biosynthetic and energetic demands of tumor growth [183]. Tumor cells often exhibit constitutive Ca2+ transfer from the endoplasmic reticulum (ER) to mitochondria, a process mediated by the mitochondria-associated ER membrane complex comprising IP3R, VDAC1, and Grp75 [183, 184]. This sustained Ca2+ signaling not only fuels anabolism but also inhibits apoptosis, conferring a survival advantage [183].
The mitochondrial calcium uniporter (MCU) is the primary channel for Ca2+ uptake into the mitochondrial matrix, and its dysregulation is a common feature in human cancers [183]. MCU-mediated Ca2+ uptake promotes mitochondrial biogenesis and tumor growth in CRC by activating PGC-1α and transiently facilitating mitotic progression [185, 186]. However, this reliance on Ca2+ signaling is a double-edged sword. Disruption of MAMs, such as that induced by metformin in hepatocellular carcinoma, can induce an energy crisis and inhibit tumor growth [187]. Moreover, mitochondrial calcium overload can trigger the opening of the mitochondrial permeability transition pore(mPTP), leading to apoptosis or necrosis [11, 188, 189]. This delicate balance positions calcium channels as promising therapeutic targets.

Iron and the induction of ferroptosis.
Iron is essential for mitochondrial function, serving as a cofactor in Fe-S cluster synthesis, heme production, and OXPHOS [190, 191]. However, its redox activity also places it at the center of ferroptosis, an iron-dependent form of regulated cell death driven by catastrophic lipid peroxidation [192, 193]. Mitochondria are critical sites for executing ferroptosis. The ETC of the organelle can generate ROS that initiate lipid peroxidation, a process influenced by dietary lipids such as diacyl-polyunsaturated fatty acid phosphatidylcholines (PC-PUFA2s) [194]. Morphologically, ferroptotic cells exhibit shrunken mitochondria with condensed and ruptured outer membranes [192, 193]. Key mitochondrial defenders against ferroptosis include GPX4 and DHODH, which work to suppress lethal lipid peroxides [195].
Therapeutically, inducing ferroptosis is a promising strategy, particularly for overcoming therapy resistance. Treatment-resistant, mesenchymal cancer cells are often highly susceptible to ferroptosis [196–198]. Conversely, cancer cells can evade this fate through mechanisms such as mitochondrial recruitment of cGAS or enhanced mitophagy [179, 199, 200]. Therefore, targeting ferroptosis defenses, such as by triggering GPX4 ubiquitination, represents a viable approach to inhibiting tumor growth [201].

Copper: a double-edged sword in tumorigenesis
Copper is an essential nutrient, and its levels are frequently elevated in tumors to support proliferation [202, 203]. Copper promotes tumorigenesis by activating oncogenic signaling pathways such as RAS-ERK and PI3K-AKT-mTOR, stabilizing HIF-1α, and facilitating immune evasion [203].
However, this dependence creates a vulnerability, in that excess copper can induce a novel form of cell death, cuproptosis, which depends on mitochondrial respiration and involves the direct binding of copper to lipoylated enzymes in the TCA cycle. This binding triggers the aggregation of acylated proteins and the loss of iron-sulfur cluster proteins, leading to proteotoxic stress and cell death [204]. This mechanistic insight underpins therapeutic strategies; while mitochondrial copper depletion may be effective against OXPHOS-dependent cancers [205, 206], the targeted delivery of copper ionophores to induce cuproptosis is an emerging antitumor strategy that can synergize with immunotherapy [207, 208].
In summary, dysregulation of mitochondrial ion homeostasis creates a landscape of unique vulnerabilities and oncogenic dependencies. Targeting these ion pathways, such as by modulating calcium signaling, inducing ferroptosis, or triggering cuproptosis, holds significant promise for developing novel cancer therapeutics.

Epigenetic regulation of mitochondrial function in cancer
Epigenetic modifications serve as a critical interface between the nuclear genome, mitochondrial homeostasis, and oncogenic transformation. By regulating gene expression without altering the DNA sequence, epigenetic mechanisms, including DNA and RNA modification, histone alteration, and mitochondrial-specific epigenetics, orchestrate the metabolic and functional reprogramming of mitochondria to drive cancer progression.

Nuclear DNA and RNA methylation
DNA methylation typically suppresses gene expression at the nuclear level. Methylation of the POLGA, which encodes a core subunit of mtDNA polymerase, is a prime example. This epigenetic silencing reduces mtDNA copy number, locking cancer cells into a glycolytic state to support their growth [209]. Beyond DNA, N6-methyladenosine (m6A) modification on RNA is a key regulator of cancer metabolism. The m6A methyltransferase METTL3 stabilizes HK2 mRNA in cervical cancer, enhancing the Warburg effect [210]. Conversely, in clear cell renal cell carcinoma (RCC), the m6A demethylase FTO erases m6A marks on PGC-1α mRNA, restoring mitochondrial oxidative metabolism and generating tumor-suppressive oxidative stress [211]. These opposing roles highlight the context-dependent nature of epigenetic regulation in fine-tuning the metabolic balance between glycolysis and OXPHOS.

Histone modification
Histone acetylation dynamically regulates nuclear-encoded mitochondrial gene expression. In prostate cancer, arginine-induced histone acetyltransferase activity facilitates TEAD4 recruitment to OXPHOS gene promoters, sustaining mitochondrial function and tumor cell survival [212]. Conversely, histone deacetylation silences crucial genes. In specific ovarian and lung cancers, epigenetic silencing of the chromatin remodeler SMARCA4 leads to reduced IP3R calcium channel expression, impairing mitochondrial calcium signaling and blunting apoptosis, thereby fostering chemoresistance [213].

Mitochondrial-specific epigenetics
The epigenetic landscape directly extends to the mitochondrion. Hypomethylation of the mtDNA D-loop region enhances the transcription of mtDNA-encoded genes and increases mtDNA copy number, boosting the capacity of OXPHOS in cancers such as colon cancer and osteosarcoma [356]. At the RNA level, the mitochondrial methyltransferase METTL17 mediates mitochondrial RNA methylation; its loss in CRC disrupts mitochondrial protein translation, leading to lethal lipid peroxidation and ferroptosis [214]. Furthermore, metabolite-induced signaling within the mitochondria is subject to epigenetic control. In acute myeloid leukemia (AML), fumarate binding promotes mitochondrial translation, but this pro-tumorigenic signal can be suppressed by arginine methylation of the enzyme ME2 [215] (Fig. 4).

Mitochondria and the TME
The TME is an active ecosystem composed of cancer cells, diverse immune populations (T cells, B cells, NK cells, and macrophages), cancer-associated fibroblasts (CAFs), and other stromal components [57]. The function of each cellular constituent in this ecosystem is critically governed by mitochondrial homeostasis, which acts as a central processing unit by (1) furnishing the metabolic energy and precursors for proliferation and effector functions; (2) maintaining functional integrity through dynamic QC; and (3) releasing signaling molecules, such as mtDNA and mtROS, which dictate cell fate and intercellular communication. As discussed in the following sections, mitochondrial perturbations, whether in respiration, MQC, or ion homeostasis, within specific TME cells can decisively shift the balance between tumor suppression and promotion, underscoring the role of the organelle as a master regulator in the TME.

CD8 + Tcells
CD8 + T cells play a central role in antitumor immunity, and mitochondrial dysfunction impairs CD8 + T cell antitumor activity through various pathways. At the mitochondrial respiration level, CD8 + T cells rely on fatty acid-binding protein 5 (FABP5) to coordinate effective lipid transport, provide energy for mitochondrial respiration, and maintain CD8 + T cell bioenergetic needs. TAGLN2, an essential molecule that interacts with FABP5 to perform this function, is extremely important in mediating normal cellular immunity. In ovarian cancer, TAGLN2 inhibition impairs mitochondrial respiration, reducing the energy supply to CD8 + T cells and weakening the antitumor immune function [216]. Metabolic interventions that promote pyruvate entry into mitochondria can reverse T-cell exhaustion by enhancing OXPHOS and restoring effector functions [217]. Moreover, mitochondrial metabolism sustains CD8 + T-cell migration for efficient infiltration into solid tumors to exert its antitumor effect [218]. At the MQC level, reduced mitophagy in CD8 + T cells leads to depolarized mitochondrial accumulation, epigenetic metabolic reprogramming, and reduced response to anti-PD-1 therapy [219]. Furthermore, impaired MQC can decrease the number of mitochondria, resulting in decreased T-cell activity, while mitochondrial transfer can directly supplement CD8 + T cells with functional organelles, enhancing their anti-tumor capacity [162]. At the mtDNA level, PKM2 activation enhances nuclear and mtDNA hypomethylation in CD8 + T cells by altering one-carbon metabolism and enhancing mitochondrial biogenesis and function, thereby enhancing the recall response and antitumor function of CD8 + T cells [357].
At the level of ion homeostasis, MFN2 strengthens ER–mitochondria contacts by interacting with the ER-resident Ca2+-ATPase SERCA2, thereby precisely regulating Ca2+ transfer to mitochondria. This controlled Ca2+ influx supports optimal mitochondrial metabolism while preventing Ca2+ overload-induced apoptosis in tumor-infiltrating CD8 + T cells [220]. Similarly, linoleic acid promotes ER-mitochondria tethering and fine-tunes Ca2+ homeostasis, thereby sustaining mitochondrial bioenergetics and enhancing CD8 + T cells [221].

CD4 + T cells
CD4 + T cell subsets exhibit distinct functional and metabolic profiles. Anti-tumor subsets, such as Th1 and Th17 cells, require mitochondrial competence; switching Th17 cell metabolism from glycolysis to OXPHOS enhances their persistence and anti-tumor efficacy [222]. Tumor-derived TGF-β inhibits mitochondrial respiration by inhibiting mitochondrial complex V activity in these anti-tumor CD4 + T cells, thereby reducing anti-tumor factor IFN-γ production by these CD4 + T cells [358].
In contrast, immunosuppressive Tregs depend on mitochondrial respiration for their function, and inhibition of mitochondrial electron transport by agents such as atovaquone can abrogate their suppressive activity and enhance the efficacy of checkpoint blockade [223, 224]. The TME further shapes these responses through metabolites such as lactate, which promotes the proliferation and immunosuppression of Tregs [359].
mtDNA release from different CD4 + T cells may induce opposing tumor effects. In a TNBC mouse model, low-dose radiation has been shown to induce mitochondrial damage and mtDNA release, activate the STING pathway, promote Th1 differentiation and tumor infiltration, and enhance Th1 anti-tumor immunity [225]. FABP5 inhibition has been shown to impair lipid metabolism and cristae structure, trigger mtDNA release, activate cGAS-STING-dependent type I interferon signaling, enhance IL-10 production, and reinforce the suppressive function of Tregs in mice models [226].

B cells
B cells are effector cells that mediate humoral immunity and exert antitumor effects within the TME in various cancers [227, 228]. Enhancing B cell mitochondrial OXPHOS has been shown to reduce lung metastasis [228]. Mechanistically, this is consistent with OXPHOS-dependent plasmablast differentiation facilitating anti-tumor antibody production [229]. Furthermore, mitochondrial QC and mtROS drive the fate differentiation of activated B cells, including class-switch recombination (CSR) and plasma cell differentiation [230]. However, certain B cell subpopulations contribute to immune evasion in CRC by reprogramming mitochondrial NAD + regeneration and oxidative metabolism [231]. Regulatory B cells, which constrain autoimmune response and restrict inflammation via interleukin (IL)−10 expression, also rely on OXPHOS to maintain their function and phenotype [232], illustrating the dual role of mitochondrial pathways in shaping B cell responses in cancer.

Macrophages
Macrophage polarization and function within the TME are governed by mitochondrial metabolism. Pro-inflammatory M1 macrophages typically rely on glycolysis, whereas anti-inflammatory, pro-tumorigenic M2 macrophages and tumor-associated macrophages (TAMs) depend on OXPHOS and mitochondrial fatty acid oxidation (FAO) [233, 234]. Lipid accumulation in TAMs promotes FAO, driving OXPHOS, ROS production, and M2 activation [235]. Metabolites such as itaconate and fructose can modulate this balance by inhibiting succinate dehydrogenase or altering calcium signaling, respectively, to favor an M2-like, pro-tumor state [236, 237]. Therapeutic strategies aim to reprogram TAMs toward an M1 phenotype; for instance, by using photodynamic therapy to release oxidatively damaged mtDNA from tumor cells, which converts TAMs to a tumor-suppressive M1 state [238].

Natural killer (NK) cells
NK cells, which are part of the innate immune system, can quickly recognize and kill virus-infected or tumor cells without prior sensitization. NK cell cytotoxicity is critically dependent on mitochondrial integrity. In the hypoxic TME of liver cancer, sustained Drp1 activation causes excessive mitochondrial fission, leading to fragmented organelles and diminished NK cell cytotoxicity, enabling immune evasion [239]. Conversely, tumor metabolism can also influence NK cell recruitment; for instance, in nutrient-deficient environments, tumor mitochondrial ACAT1 can acetylate NF-κB p50, enhancing the expression of immune factors that promote NK cell recruitment and activation [240].

CSCs
CSCs are a subpopulation of cancer cells with similar functions to normal stem cells. CSCs are important in tumor occurrence and development and are closely related to tumor invasion, metastasis, drug resistance, and recurrence after treatment [241]. CSCs use mitochondrial pathways to maintain stemness and resist treatment [29]. In neoadjuvant chemotherapy-resistant TNBC, co-amplification of MYC and MCL1 promotes CSC-driven resistance by enhancing mitochondrial OXPHOS and ROS production [242]. In terms of MQC, PGC-1α is instrumental in maintaining CSC stemness across multiple cancer types [122, 124]. DRP1 regulates asymmetric division of CSCs, a core characteristic of stem cells [113]. In breast cancer, inhibition of the transcriptional coactivator YAP/TAZ induces DRP1 expression and impairs OXPHOS, significantly reducing chemoresistant CSCs [243]. Mitophagy protects CSCs from damage by clearing ROS, thereby promoting CSC survival [29]. Moreover, mtDNA depletion is associated with enhanced CSC stemness and increased treatment resistance [29]. Targeting mitochondrial homeostasis in CSCs has great potential for treating drug-resistant tumors and preventing recurrence. Using copper ion carriers combined with photothermal therapy to deliver copper ions to the mitochondria to induce cuproptosis in CSCs can effectively inhibit tumor recurrence and distant metastasis [244].

CAFs
CAFs, which are common cells in the TME stroma, promote tumor progression by enhancing the ability of cancer cells to undergo EMT and promoting stem cell properties and metastatic dissemination through mitochondrial-mediated metabolic coupling. [245] CAFs exhibit a glycolytic phenotype, producing lactate, which is exported to and utilized by cancer cells for OXPHOS, fueling cancer cell proliferation in oral and breast cancers [246, 247]. The mDia2-MIRO1 axis in CAFs controls mitochondrial localization and OXPHOS, and is essential for secreting tumorigenic factors [248]. Notably, CAFs can directly transfer mitochondria to cancer cells, enhancing recipient OXPHOS and promoting breast cancer metastasis [360]. Common mtDNA deletions in skin basal and squamous cell carcinomas serve as biomarkers to distinguish CAFs from normal fibroblasts [249]. This suggests that mtDNA loss is important in CAF phenotypic transformation. In cholangiocarcinoma, inducing the release of mitochondrial pro-apoptotic factors to induce CAF apoptosis can inhibit tumor growth [250]. Moreover, CAFs can regulate the mitochondrial function of tumor cells. For instance, in prostate cancer, paracrine ANGPTL4 in CAFs promotes tumor cell mitochondrial biogenesis and OXPHOS, thereby promoting cancer cell growth and chemotherapy resistance [361].

Others
Endothelial cells play a pivotal role in tumor tissue angiogenesis and are critically involved in hematogenous metastasis. Metabolites such as R-2-HG, which is produced by isocitrate dehydrogenase (IDH)1 mutant tumor cells, flow from the TME into endothelial cells, activating the mitochondrial respiratory chain and promoting endothelial cell migration during tumor angiogenesis [251]. Endothelial cell OPA1 is particularly important for tumor angiogenesis and can promote tumor growth and metastasis through NF-κB [252]. Thus, blocking mitochondria-mediated angiogenesis in endothelial cells may have anticancer effects.
Mesenchymal stem cells (MSCs), a type of multipotent stromal cell distributed across various human tissues, have been shown to promote chemoresistance in the TME through the transfer of mitochondria to cancer cells via tunneling nanotubes. The transferred mitochondria restore mitochondrial function, attenuate mitophagy, and support ATP-dependent drug efflux mediated by ATP-binding cassette(ABC) transporters [253, 254]. Paradoxically, under chemotherapy-induced stress, T-cell acute lymphoblastic leukemia cells can donate mitochondria to MSCs, a process that contributes to the cancer cells’ own chemoresistance [255]. This bidirectional and context-dependent mitochondrial transfer underscores the complex role of MSCs in shaping the survival of tumor cells under therapeutic pressure (Fig. 5).

Therapeutic strategies targeting mitochondrial homeostasis
Conventional cancer therapies (e.g., surgery, chemotherapy, and radiotherapy) are challenged by toxicity and multidrug resistance, whereas biotherapies, including immunotherapy and gene therapy, are limited by drug resistance and insufficient delivery of biomacromolecules. Therefore, developing novel therapeutic strategies is crucial. Mitochondria, as essential providers of cellular energy, play key roles in cell differentiation, signaling, ROS production, apoptosis, and calcium homeostasis and are promising cancer treatment targets. We previously discussed the imbalances in mitochondrial homeostasis, including MQC, metabolism, mtDNA damage, and ion homeostasis, in various cancers. In this section, we explore the therapeutic strategies targeting mitochondria, categorized by specific mitochondrial alterations and treatment objectives. Existing studies have established systematic classifications for antineoplastic agents, such as the “mitocans” taxonomy [256]. Here, we focus on the modulation of mitochondrial homeostasis by these compounds. Accordingly, the classification adopted here is primarily based on the specific molecular targets through which antineoplastic drugs exert their regulatory effects on mitochondrial homeostasis (Fig. 6 and Table 1).

Targeting alterations in mitochondrial metabolism

OXPHOS
The Warburg effect is the primary mechanism that underlies glucose metabolism in cancer cells. However, inhibiting glycolysis does not always prevent tumorigenesis. For example, blocking the pyruvate kinase M2 isoform, which catalyzes the final glycolysis stage, has been shown to promote tumorigenesis in a breast cancer model [271]. Many cancer cells can utilize OXPHOS for glucose oxidation in their functional mitochondria [272], suggesting that OXPHOS represents an important therapeutic target.
The TCA cycle, which occurs in the mitochondrial matrix, is the main component of OXPHOS. Mutations in TCA cycle enzymes, particularly succinate dehydrogenase, fumarate hydratase, and IDH, are linked to cancer progression [273]. Studies have targeted the TCA cycle for OXPHOS tumor interventions, especially on reversing the outcomes of these mutations. For example, wild-type IDH catalyzes the conversion of isocitrate to α-ketoglutaric acid(α-KG) in normal cellular metabolism. In contrast, mutant IDH catalyzes the induction of oncometabolite (R)−2-hydroxyglutarate ((R)−2-HG) which acts as a competitive inhibitor of α-KG-dependent dioxygenases. This competition leads to widespread DNA and histone hypermethylation, resulting in epigenetic dysregulation. Consequently, hematopoietic progenitor cell differentiation is blocked, a key event in the promotion of leukemogenesis. Mutant IDH inhibition by AGI-5198 reduces (R)−2-HG formation, causing epigenetic dysregulation and a block in cellular differentiation [274]; enasidenib and vorasidenib are FDA-approved for the treatment of acute myelogenous leukemia and glioma with IDH2 and IDH1/2 mutations, respectively [257, 275]. CPI-613, which targets α-ketoglutarate and pyruvate dehydrogenases, is undergoing phase I/II/III clinical trials for leukemias, lymphomas, small cell lung cancer (SCLC), and metastatic pancreatic adenocarcinoma [258, 259, 276]. However, CPI-613 administration is associated with a graded spectrum of adverse effects, varying in severity from mild manifestations to life-threatening complications. Mild reactions frequently include gastrointestinal disturbances, such as nausea and vomiting, whereas hematological and metabolic abnormalities, including anemia, lymphopenia, hypocalcemia, hyponatremia, and dysgeusia, are more pronounced. Notably, the most severe toxicity observed is dose-limiting nephrotoxicity, which may present as acute renal injury requiring clinical intervention [258, 259, 276].
The TCA cycle produces NADH and FADH2, which supply electrons to ETC complexes I–V and convert adenosine diphosphate (ADP) + Pi to ATP. Dysregulation in complexes I–V has been verified in various cancer types, including melanoma and myeloid leukemia [277–280], making ETC-targeting drugs attractive options for cancer therapy. Here Complex I inhibitors are dominantly introduced which mainly include metformin [281], IACS-010759 [260] and mitochondria-targeted tamoxifen (MitoTam) [282, 283]. Metformin, a mainstay in diabetes treatment, exhibits significant promise as an anticancer therapeutic [281]. It acts by directly targeting mitochondrial complex I, thereby reducing ATP output and increasing the AMP/ATP ratio. This energy deficit is recognized by AMPK kinase, initiating a signaling cascade that inhibits mTOR. Consequently, this pathway suppresses protein synthesis and disrupts cell cycle progression, effectively checking the aberrant proliferation of cancer cells. Among these pharmaceutical agents, a subset has progressed to clinical investigation, where a diverse profile of efficacy and adverse effects has been demonstrated. For example, IACS-010759, a synthetically engineered complex I inhibitor, demonstrated potent activity in OXPHOS-dependent AML models. Targeting complex I induced energy depletion and reduced aspartate synthesis, thereby compromising nucleotide biosynthesis, promoting apoptosis, and suppressing tumor growth, while maintaining a favorable tolerability profile [260]. However, clinical evaluations revealed a narrow therapeutic window, with dose-limiting toxicities including hyperlactatemia (occurring in 83% of patients with solid tumors) and neurotoxicity (manifesting as peripheral neuropathy in 35% of patients with solid tumor) [284]. Furthermore, the inhibitory effects were transient and accompanied by resistance mechanisms [284]. In a phase I/Ib clinical trial evaluating MitoTam, a mitochondrially targeted tamoxifen derivative, notable efficacy was observed, particularly in patients with RCC [283]. Among the cohort, five out of six patients with RCC achieved a clinical benefit rate, characterized by disease stabilization or partial response, underscoring the potential of the agent in this malignancy. Concurrently, the safety profile revealed frequently reported adverse events, primarily hematological toxicities such as anemia and neutropenia, pyrexia or hyperthermia, and thromboembolic complications. These findings support further investigation of MitoTam in targeted cancer therapies, while ensuring careful monitoring for associated toxicities. Additionally, Complex II inhibitors include trans-4,5-dihydroxy-2-cyclopentene-l-one [285] and 1,2,3,4-tetrahydroazuleno[1,2-b] [286], while enhancers include mesothelin protein [363] and berberine [364]. Complex III inhibitors include atovaquone [287] and compound 6c [288]. Complex IV is inhibited by doxorubicin [289].
ATP synthase catalyzes the synthesis of ATP by harnessing the proton concentration gradient which is established by the mitochondrial ETC across the inner membrane, to drive ADP phosphorylation. Bedaquiline, an inhibitor of mitochondrial ATP synthase, depletes cellular energy reserves, which significantly impairs the functions of critical cancer cells, including anchorage-independent growth and migratory capacity [290]. The drug repurposing profile of bedaquiline expands the therapeutic horizon for cancer treatment. However, as an FDA-approved drug with known clinical adverse effects, notably QT interval prolongation and hepatotoxicity, its translation into oncology necessitates further rigorous testing and safety evaluation.
Paradoxically, coenzyme Q10 (CoQ10), an endogenous antioxidant and integral component of the mitochondrial ETC, is not typically targeted directly in therapeutic strategies. Instead, it is more commonly investigated for its role as a chemosensitizer or supplement [291]. This application preference is likely attributable to its inherently low systemic bioavailability when administered exogenously.
Targeting OXPHOS inhibits PD-L1 expression, reducing tumor immune evasion. AMPK binds and phosphorylates PD-L1 for ER degradation [292] when activated by mitochondrial metabolism disruption [172]. Liu et al. developed IR-LND, a compound linking the tumor-targeting IR-68 with the anti-tumor drug lonidamine (LND) [293]. This compound self-assembles with albumin, forming a tumor-targeted IR-LND@Alb nanosystem that selectively accumulates in tumor mitochondria, disrupts OXPHOS, activates AMPK, and promotes PD-L1 degradation. This reduces the required LND dose by over 100-fold for PD-L1 inhibition and tumor hypoxia reversal, simultaneously achieving mitochondrial metabolism targeting and immune checkpoint inhibition for cancer therapy.

Other metabolic processes
FAO, also called β-oxidation, involves a series of cyclic catabolic reactions performed by a trifunctional enzyme complex in the mitochondrial matrix. This pathway produces high ATP levels and is activated to sustain high proliferation and cope with stress in many tumors [294]. Carnitine palmitoyltransferase 1 (CPT1) is a rate-limiting FAO enzyme and a major cancer treatment target. Etomoxir, the earliest CPT1 inhibitor, hinders tumor progression and function in several cancers, including glioma [295], breast cancer [296], and prostate cancer [297]. However, clinical application is difficult to achieve with etomoxir owing to liver toxicity [298, 299]. In contrast, as a dual CBT1 and CBT2 inhibitor, ST1326 alleviates AML with higher selectivity and lower toxicity and may exhibit better potential in future trials [261].
Glutamine, the most abundant amino acid in human plasma, is the primary transporter of nitrogen between organs. Although it is classified as non-essential, it is vital for synthesizing amino acids, proteins, nucleotides, and glutathione and activating mTORC1 [300]. Many cancer cells exhibit “glutamine addiction,” relying on glutamine as an anaplerotic input, replenishing TCA cycle intermediates; therefore, targeting glutamine metabolism represents an innovative cancer treatment approach [301–303]. Glutamine metabolism targets include glutaminase, glutamate dehydrogenase, aspartate aminotransferase, and glutamate pyruvate transaminase 2 (GPT2). Glutaminase inhibitors mainly include BPTES [304] and CB-839, with CB-839 having undergone clinical trials for several hematological tumor treatments, including CRC, melanoma, non-SCLC, and TNBC [262]. When administered in combination with nivolumab, CB-839 demonstrated a well-managed adverse event profile, characterized by favorable tolerability [305]. The safety signature closely aligned with the established data for CB-839 monotherapy and the known safety parameters of nivolumab. Commonly reported adverse events were predominantly mild to moderate in severity, including fatigue, nausea, and photophobia. The incidence of grade 3 or higher adverse events was low, underscoring the combination’s controlled risk profile. Ongoing clinical investigations are evaluating the long-term safety and efficacy of this therapeutic approach. However, research focusing on GDH, GOT, and GPT2 inhibitors remain limited to the laboratory because of their low specificity and uncertain side effects.

Metabolic regulators
Mitochondrial metabolism is regulated by oncogenes and tumor suppressor genes. The MYC oncogene and TP53 tumor suppressor represent pivotal regulators of mitochondrial metabolism. The MYC oncoprotein acts as a master regulator of mitochondrial biogenesis, orchestrating the expression of nuclear-encoded mitochondrial genes to enhance ETC function and meet the anabolic demands of proliferating cancer cells [306]. Conversely, the TP53 tumor suppressor promotes oxidative phosphorylation and mitochondrial integrity by transactivating genes such as SCO2, AIF, and GLS2, while simultaneously curbing glycolysis [307]. The importance of this reciprocal regulation lies in the profound metabolic rewiring it imposes, where MYC-driven tumors become critically dependent on mitochondrial function, and loss of TP53 abrogates a key metabolic checkpoint, facilitating the Warburg effect and supporting biosynthetic processes. This dependency creates a therapeutic vulnerability: targeting mitochondrial metabolism, such as by inhibiting mitochondrial transcription or translation in MYC-overexpressing cancers, can induce synthetic lethality, while restoring or mimicking the metabolic functions of p53 offers a strategy to suppress tumorigenic metabolism. Nevertheless, the pleiotropic nature of MYC makes it challenging to delineate and specifically target its mitochondrial functions for therapeutic intervention. Nevertheless, the direct pharmacological targeting of MYC has progressed into clinical trials [308]. Experimental approaches that activate TP53 either by pharmacological inhibition of its negative regulators or genetic overexpression converge on a common mechanistic endpoint when combined with ULK1 impairment [309]. This concerted action disrupts mitochondrial QC by inhibiting mitophagy, resulting in the accumulation of damaged organelles and elevated ROS. This cellular stress triggers a cascade involving NLRP3 and caspases 8 and 3, which cleave gasdermin E (GSDME) to execute pyroptosis.
Targeting mitochondrial OXPHOS for cancer therapy has two primary limitations: significant toxicity and inherent resistance. The narrow therapeutic window of potent OXPHOS inhibitors, exemplified by the hyperlactatemia and neurotoxicity observed with IACS-010759, stems from the essential role of mitochondrial function in healthy cells. Furthermore, cancer cells exhibit metabolic plasticity, often developing resistance by switching to alternative energy sources, such as glycolysis or glutaminolysis.

Targeting MQC
MQC in cancer therapy focuses on modulating mitochondrial autophagy, promoting mitochondrial disruption, mitochondrial fission and fusion intervention, and mitochondrial hijacking inhibition by tumor cells instead of normal cells.
Targeting tumor cell mitochondria exerts cytotoxic effects through mtDNA release; this leads to the activation of various tumor cell death induction pathways, and the release of mtDNA, resulting in activation of the cGAS-STING pathway to enhance immune-mediated tumor cell killing. Fan et al. designed an oxaliplatin and acetaminophen complex named OPA2, which induces mitochondrial membrane remodeling and mtDNA leakage, leading to double-stranded DNA accumulation and synergistically activating the intracellular cGAS-STING pathway [263]. Additional approaches applied to induce mitochondrial dysfunction in tumor cells include ion homeostasis imbalance, which will be discussed later. Interestingly, a reduction in mitochondria does not always harm tumor cells. Indeed, Zhao et al. discovered that after radiotherapy, glioma cells initiated a self-protective mechanism through suppression of PGC1α expression, thereby inducing mitochondrial biogenesis inhibition and ultimately improving radiotherapy resistance. Glioma cell resistance to radiotherapy has been shown to be alleviated following PGC1α upregulation [126]. These findings imply that the amount of mitochondria and their effects exhibit heterogeneity across tumor types, necessitating extensive evaluation for drug development and individualized precision medicine.
Artificial interventions in tumor cell mitochondrial fission and fusion have potential for cancer therapy. A previous study showed that enhanced expression of Yin Yang 2 (YY2), a novel inhibitor of CSC preservation, hinders asymmetric division of liver CSCs, resulting in a reduction in the CSC reservoir and diminished tumor-initiating potential [113]. This effect is mediated by inhibition of mitochondrial fission caused by suppression of dynamin-related protein 1 transcription [113]. Similarly, promotion of mitochondrial infusion may hinder tumor development. Indeed, Wang et al. discovered that the natural compound naringenin inhibits CRC proliferation and promotes apoptosis by activating AMPK phosphorylation and mitochondrial fusion [310]. According to the literature, mitochondrial fission inhibition or mitochondrial fusion promotion suppresses tumor progression, whereas mitochondrial fission promotion facilitates tumor development.
Tumor cells have been shown to hijack mitochondria from normal cells, with Saha reporting that this can be achieved via physical nanotubes [118]. Therefore, the inhibition of nanotube formation represents a potential therapeutic target. In the same study, the researchers combined a farnesyltransferase and geranyltransferase 1 inhibitor, L-778123, to partially block nanotube formation and mitochondrial transfer, revealing improved anti-tumor outcomes in an aggressive immunocompetent breast cancer model [118]. However, the mechanisms underlying nanotube formation remain unclear.
Mitophagy regulation is an important strategy for tumor MQC. Although mtDNA release activates the cGAS-STING pathway and induces apoptosis, tumor cells initiate the innate process of clearance or mitophagy of leaked mtDNA [311]. Yu et al. designed a complex of nano-inducers (PCM) with a phenolic group of PEG-polyphenol-chlorin e6 (Ce6), which achieved mitochondrial respiratory inhibition by inducing Cu2+ metabolic chaos [312]. This system, combined with mitophagy inhibitor (Mdivi-1), which protects proteotoxic products from mitophagy-elimination, allows more mtDNA release into the cytosol and stimulates cGAS-STING signaling [312]. Moreover, Lv et al. developed KCKT, a mitochondria-targeting nanotherapeutic agent that undergoes acid-responsive self-assembly into nanofibrillar structures within the lysosomal microenvironment of malignant cells [313]. Ultrasound irradiation revealed localized ROS overproduction, lysosomal membrane destabilization, and nanofiber release in vivo. Liberated nanostructures exhibit selective mitochondrial tropism, where persistent ROS generation drives membrane depolarization and bioenergetic collapse through sustained oxidative assault. This design concurrently resolves two longstanding barriers in mitochondria-targeted oncology: 1) pH-triggered structural reorganization circumvents lysosomal sequestration and optimizes payload delivery efficiency, and 2) lysosomal impairment disrupts mitophagic flux via impaired autolysosomal maturation, establishing a self-reinforcing pathological cycle of unmitigated oxidative damage and compromised organellar QC. By synergizing mitochondria-directed cytotoxicity with autophagic checkpoint inhibition, this self-amplifying dual mechanism establishes a paradigm shift in the therapeutic strategy for precise tumor eradication. However, mitophagy inhibition has been shown to enhance tumor immune checkpoint activity. Indeed, Xie et al. demonstrated that ATAD3A suppressed PINK1 expression, blocked PD-L1 mitochondrial localization, and mitigated mitophagy-induced PD-L1 degradation, thereby strengthening its immune-suppressive effects [147]. The inhibition of ATAD3A expression significantly increased the proportion of cytotoxic T lymphocytes and reduced exhausted T cell ratios in vivo [147]. The dual roles of mitophagy in tumor biology are evident, and effects vary according to the TME. Mitophagy inhibition can activate immune responses by accumulating mtDNA, but may also worsen mitochondrial dysfunction and promote malignant transformation. Conversely, enhancing mitophagy suppresses oncogenic signaling by removing damaged mitochondria but may weaken immune activation. Thus, therapeutic strategies should prioritize tumor type, immune status, and treatment stage––inhibiting mitophagy in immunogenic tumors to enhance immune effects or targeting autophagy nodes (e.g., PINK1/ATAD3A) in mitochondria-dependent tumors to disrupt metabolic homeostasis, combined with immune checkpoint blockade for synergistic antitumor effects. The assessment of dynamic tumor adaptability is key to the success of this process.
Of particular note, targeting mitochondrial membrane integrity may represent a pivotal strategy for tumor killing and drug resistance control in tumor cells. Disruption of the mitochondrial outer membrane facilitates the leakage of cytochrome c into the cytosol, where it assembles with dATP, APAF1, and caspase-9 to form the apoptosome, subsequently initiating the apoptotic cascade [314, 315]. Neoplastic cells frequently evade this form of programmed cell death by overexpressing anti-apoptotic members of the BCL-2 protein family [316]. This elevated expression inhibits the pro-apoptotic factors from facilitating the oligomerization of BAX and BAK, thereby preserving the mitochondrial outer membrane integrity and promoting cell survival [317]. Consequently, the pharmacological inhibition of BCL-2 has emerged as a viable therapeutic strategy to induce apoptosis in malignant cells. Venetoclax is a prototypical agent in this class, inhibiting BCL-2 via a selective BH3 mimetic [316]. The study demonstrated that Venetoclax effectively eliminates not only the bulk population of AML blasts but also induces apoptosis in phenotypically defined CD34+CD38−CD123+ leukemic stem and progenitor cells. This activity is critical for targeting the disease reservoir to prevent relapse. Furthermore, samples from patients with chemoresistant disease retain sensitivity to venetoclax, offering a novel therapeutic avenue for those with refractory or relapsed AML. However, adaptive resistance to Venetoclax, which is often mediated by mechanisms such as the upregulation of alternative survival pathways or specific mutations, poses a significant clinical challenge. Bielcikova et al. observed that upon acquiring resistance to Venetoclax, AML cells undergo adaptive alterations in mitochondrial ultrastructure [318]. Resistant cells exhibit mitochondria with narrower and more densely packed cristae, orchestrated by the mitochondrial dynamin protein OPA1 upregulation. This remodeling serves to sequester cytochrome c, thereby counteracting drug-induced apoptosis. A genome-wide CRISPR/Cas9 loss-of-function screen identified the mitochondrial chaperonin caseinolytic peptidase B protein (CLPB) as a critical genetic determinant conferring sensitivity to Venetoclax in AML. Mechanistically, CLPB directly interacts with the cristae-shaping proteins OPA1 and HAX1, stabilizing the cristae architecture. Ablation of CLPB perturbs this homeostasis, leading to aberrant proteolytic cleavage of OPA1, disorganization of cristae, loss of mitochondrial membrane potential, and consequently, cytochrome c release, which amplifies the apoptotic signal. Therefore, therapeutic targeting of CLPB or the broader regulatory pathways governing mitochondrial architecture has emerged as a promising novel approach to overcome Venetoclax resistance. This strategy demonstrates enhanced efficacy, particularly in combination with hypomethylating agents such as azacitidine, offering a novel therapeutic avenue for patients with AML. Arsenic trioxide (As₂O₃), a compound historically recognized for its high toxicity, exhibits a synergistic effect with Venetoclax for treating AML [365]. The underlying mechanism involves arsenic trioxide-induced ROS generation, which is partially counteracted by activation of the Nrf2-mediated antioxidant pathway. Venetoclax, however, inhibits this Nrf2 activation, and the combination markedly increases oxidative stress, disrupts mitochondrial metabolic function, and selectively eradicates leukemia stem cells. This approach remains effective even against Venetoclax-resistant cell populations.
Targeting MQC in cancer therapy, though promising, faces significant limitations. A major challenge is the dual, context-dependent role​ of processes such as mitophagy, which can either suppress or promote tumors depending on the environment [319]. This complexity makes predicting therapeutic outcomes difficult. Additionally, tumor metabolic plasticity​ allows cancer cells to bypass mitochondrial disruption by switching energy sources, contributing to treatment resistance. Off-target toxicity is another concern, as MQC is essential for normal cell function, and inhibitors like Mdivi-1 can adversely affect healthy tissues.

Targeting mtDNA
Drugs and molecules targeting tumor mtDNA can be grouped into three categories: 1) those that damage the mtDNA structure, 2) those that target mtDNA replication and transfer, and 3) those that target mtDNA repair.
Drugs damage tumor mtDNA primarily through three approaches. First, by the incorporation of nucleoside analogs into the mitochondrial mtDNA chain; this type mainly includes zidovudine, which competes with the natural substrate dGTP, incorporates itself into viral DNA, and has been extensively used in the treatment of HIV [320]. Zidovudine also impedes human head and neck cancer progression through cytotoxicity and oxidative stress enhancement via mtDNA interference [321]. Second, via damage to the mtDNA double-stranded structure. Doxorubicin can bind to DNA topoisomerase II beta (TOPIIβ) in the mitochondria, forming a ternary complex that leads to DNA double-strand breaks and enhances anticancer activity [264, 322]. However, owing to its high affinity for cardiomyocytes, Doxorubicin exhibits severe cardiotoxicity [322]. Finally, by binding to the bases of mtDNA bases. Cisplatin binds to the N7 position of purines in DNA, forming DNA adducts, which induce the DNA damage response and apoptosis of cancer cells; however, its combination with nuclear DNA is limited [265, 323]. Drug resistance is a prominent obstacle to the use of cisplatin, which many researchers are attempting to optimize [324].
Beyond directly damaging the mtDNA, another strategy is to hinder its replication and transcription. Structural alterations induced by certain drugs can indirectly impede these processes, whereas more direct approaches target key enzymes such as mtDNA polymerase γ (POL γ), mitochondrial transcription factor A (TFAM) and dihydroorotate dehydrogenase (DHODH). POL γ is critical for mtDNA replication, although other polymerases also exists [325]. The compound 3,3'-[(1,10-Biphenyl)−40,40-(diyl)bis(azo)]bis[4-amino-1-naphthalenesulfonic acid] (CR) exhibits a strong affinity for POL γ protein. By inhibiting POL γ activity and impairing oxidative mtDNA damage repair, it induces mitochondrial dysfunction and triggers apoptosis in MLH1-deficient human colon cancer cells [266]. TFAM is a nuclear-encoded mitochondrial protein that, once imported into the mitochondria, activates mtDNA transcription and packages mtDNA into DNA–protein complexes known as mitochondrial nucleoids [326]. It binds to mtDNA using its high-mobility group domain and induces DNA bending, facilitating assembly of the transcription initiation complex [326]. These traits render mtDNA irreplaceable. Wang et al. highlighted the critical role of TFAM, which is highly expressed in PC, in maintaining mitochondrial functional stability and supporting mitochondrial biogenesis [327]. Consequently, TFAM inhibition has emerged as a therapeutic strategy. For example, melatonin and a combination of epoxomicin and cisplatin can inhibit mitochondrial function and ultimately promote cellular apoptosis by inhibiting TFAM; this was also shown to be achieved by mitochondrially-targeted vitamin E [328, 366, 367]. Dihydroorotate dehydrogenase (DHODH) is a flavin-dependent iron-sulfur enzyme located on the inner mitochondrial membrane, serving as the fourth and rate-limiting enzyme in the de novo pyrimidine biosynthesis pathway [329]. DHODH inhibition represents a promising therapeutic strategy for impairing tumor growth. Leflunomide is a prototypical inhibitor of this enzymatic target. Liu et al. demonstrated that treatment with leflunomide or shRNA-mediated knockdown of DHODH markedly suppressed the proliferation of human melanoma cell lines [330]. Notably, this anti-proliferative effect and the accompanying cell cycle arrest were reversed by the addition of exogenous uridine, thereby establishing that the observed cytostatic effect is a direct consequence of impaired pyrimidine synthesis due to DHODH inhibition.
Targets involved in mtDNA repair include 8-oxoguanine DNA glycosylase (OGG1), apurinic/apyrimidinic endonuclease 1 (Ape-1), and poly(ADP-ribose) polymerase (PARP). OGG1 excises 7,8-dihydro-8-oxoguanine (8-oxoG) from the double-stranded DNA to initiate base excision repair (BER). TH5487, an inhibitor of OGG1, suppresses the activity of OGG1, preventing it from repairing 8-oxoG and leading to the accumulation of 8-oxoG lesions in the genome [368]. This effect has been demonstrated to radiosensitize head and neck cancer cells to high-LET protons [331]. APE-1 (also called Ref-1) is a multifunctional enzyme involved in the BER pathway, which is crucial for oxidative and alkylated DNA damage [332]. APE-1/Ref-1 inhibition by E3330 blocks malignant pleural mesothelioma cell EMT, proliferation, and migration [362]. Upon DNA damage, such as single-strand breaks, PARP is activated, binds to the lesion site, and undergoes auto-ribosylation to facilitate repair complex assembly [333]. Therefore, PARP inhibitors have been shown to hinder mtDNA repair in various types of cancers [334].
Despite its promise, targeting tumor mtDNA faces distinct challenges that guide future research directions. First, tumor heterogeneity​ leads to variable treatment responses, necessitating the development of biomarkers for patient stratification to ensure the application of therapies to the most susceptible cancer subtypes. Second, the mitochondrial membrane barrier​ compromises drug delivery; this may be overcome by designing advanced delivery systems such as mitochondrially targeted nanocarriers or metal complexes. Third, potential toxicities​ (e.g., nephrotoxicity) and resistance emergence remain significant concerns. These issues underscore the need for comprehensive long-term safety evaluations and rational combination strategies with immunotherapy or radiotherapy to enhance efficacy and overcome resistance. Finally, the current lack of robust clinical validation​ must be addressed through targeted trials to translate these promising experimental approaches into clinical practice.

Targeting mitochondrial ion homeostasis
Calcium ions are integral to several intracellular biological processes; therefore, targeting mitochondrial calcium homeostasis may represent a strategy for cancer treatment. The MCU complex, located in the mitochondrial membrane, includes MCU and accessory proteins, including the mitochondrial calcium uniporter (MICU) family (MICU1, MICU2, and MICU3); MCUb; EMRE; MCUR1; and SLC25A23, a main regulator of cellular mitochondrial Ca2+ transport and balance [369]. Therefore, the MCU complex is a key regulator of Ca2⁺ influx and a compelling theoretical target. However, translating this potential into effective cancer therapeutics has proven challenging. Although other drugs and molecules have been validated for MCU regulation in non-cancer diseases [369], few are effective in cancer treatment. For example, histidine triad nucleotide-binding protein triggers mitochondrial Ca2+ influx by regulating the MCU complex and promoting apoptosis in PC [335]. Therefore, alternative strategies that bypass MCU regulation have been explored, such as directly overloading mitochondria with Ca2⁺. Zheng et al. designed biodegradable Ca2+ nanomodulators that decompose at low pH to release Ca2+ and curcumin, leading to a sudden mitochondrial Ca2+ ion surge, pyroptosis, and remarkable suppression of tumor proliferation and metastasis [267].
Ferroptosis in tumor cells is becoming a research hotspot, with mitochondria playing a pivotal role in regulation of iron homeostasis. Drugs targeting iron homeostasis, maintained by mitochondria in tumor cells, have achieved substantial progress. Hsieh et al. designed zero-valent iron nanoparticles (ZVI-NPs) that could be abstracted by tumor cells through endocytosis. ZVI-NPs disrupt mitochondrial membrane potential, causing ROS overproduction and ferroptosis. The underlying mechanism involves AMPK activation, which induces GSK3β/β-TrCP-dependent degradation of NRF2. Because NRF2 is a master regulator of antioxidant responses, its degradation inhibits the cell’s defense against lipid peroxidation, thereby directly sensitizing tumor cells to ferroptosis.
Copper ion homeostasis is pivotal in regulating tumor cell survival and progression. Furthermore, modulation of artificial copper homeostasis in neoplastic cells is another potential therapeutic strategy. Yu et al. demonstrated a copper-centric nanotherapeutic strategy targeting mitochondrial proteotoxicity for anti-tumor immunity. A phenolic-chlorin-modulated nanoinducer exploits copper valence transition (Cu2⁺ → Cu⁺) through metal-phenolic coordination, enabling cuproptosis-specific proteotoxic stress via two copper-dependent mechanisms: (1) Cu⁺-mediated dihydrolipoamide S-acetyltransferase (DLAT) oligomerization to disrupt mitochondrial respiration and (2) ROS amplification through copper-redox-cycling-enhanced photodynamic therapy. Crucially, cytosolic mtDNA leakage is sustained through synergistic copper-driven mitochondrial destabilization and mitophagy inhibition, thereby potentiating cGAS-STING activation and subsequent NK/T cell-mediated anti-tumor immunity (with a 24% increase in tumor-infiltrating NK cells) [312]. This study establishes coordination chemistry-guided copper reprogramming as a paradigm for leveraging metal biology in cancer immunotherapy. Furthermore, IDH, a copper ionophore, has been extensively studied for its anti-tumor effect for 1) its ability to transport copper into the mitochondria, causing mtDNA damage and triggering ROS generation; this leads to apoptosis and inhibits the proliferation of GNAQ/11 (G protein alpha subunits q/11)-mutant uveal melanoma cells [268]. 2) Elesclomol enhances copper transporter ATP7A degradation, resulting in copper retention and ROS build-up within the mitochondria; this further accelerates SLC7A11 degradation and triggers ferroptosis onset [269].
Targeting tumor mitochondrial ion homeostasis involves regulation of key ions (calcium, iron, and copper) to induce synergistic antitumor effects via pyroptosis, ferroptosis, and cuproptosis. This offers precise targeting of metabolic weaknesses, immune activation, and improved delivery through nanocarriers. However, challenges remain, including intricate ion regulatory networks that can cause off-target effects, potential nanomaterial toxicity, and variable patient responses. The latter is illustrated by the transition from the theoretical promise of MCU modulation to its currently inconsistent efficacy, highlighting the need for improved patient stratification.

Targeting mitochondrial coupling among other organelles
Organelles establish close associations through membrane-contact sites, thereby coordinating homeostasis and regulation of cellular function. The interaction between mitochondria and other organelles, particularly ER and lysosomes, has been recognized for decades and plays a significant role in tumor progression and development. Given its crucial function, disrupting​ these interactions has emerged as a promising therapeutic strategy. Strategies targeting the coupling between mitochondria and other organelles for therapeutic intervention in cancer have been explored to achieve superior outcomes.
The ER dynamically interacts with mitochondria via MAM, regulating tumor progression. MAM is a central hub for lipid, calcium ion (Ca2⁺), and reactive oxygen species (ROS) signaling, promoting mitochondrial fission, autophagy, and metabolic reprogramming, thereby supporting tumor cell survival and proliferation [336]. Consequently, targets associated with these interactions have been used to explore novel therapeutic approaches for cancer treatment. For example, research indicates that HK2, a glycolytic enzyme, is located between the ER and mitochondria [337], thereby displacing HK2 from MAM using a specific peptide induces mitochondrial calcium overload, resulting in calcium-dependent calpain activation, mitochondrial depolarization, and cancer cell apoptosis [337]. Furthermore, HK2-targeting peptides exhibited anticancer activity in both patient-derived chronic lymphocytic leukemia B cells and in breast and colon cancer cells transplanted into mice. Engineered nanoparticles can also target the ER and mitochondria in cancer therapy. Guo et al. developed redox-responsive nanoparticles (GCT@CM NPs) that integrate 1G3-Cu and toyocamycin, synergistically enhancing chemotherapy regimens and immunotherapies [338]. In TME, toyocamycin inhibits ER stress adaptation, whereas 1G3-Cu induces mitochondrial dysfunction, collectively driving immunogenic cell death and activating immune responses. When combined with anti-PD-L1 antibodies, this approach reverses immunosuppression, inhibits tumor growth, and prevents recurrence and metastasis. Nanoparticles also enable tumor targeting, magnetic resonance imaging, and safe biocompatibility, thus offering a versatile platform for combined therapies.
Interaction between lysosomes and mitochondria influences tumor progression by dynamically regulating metabolic stress and cell death [339]. Liu et al. developed PVP-modified Cu-gallic acid nanoparticles to enhance lysosomal-mitochondrial cascade damage, thereby inducing cytoproptosis and pyroptosis in breast tumor cells and improving anti-tumor immunotherapy [340]. CuGA nanoparticles hijack lysosomal iron to form a bimetallic catalyst (Cu(Fe)GA) via ion exchange, leading to metal ion dysregulation (Fe2 +/3⁺, Cu⁺/2⁺, Ca2⁺) and Cu⁺-mediated mitochondrial TCA cycle disturbance. This ultimately causes pyroptosis dependent on caspase-3/GSDME. This strategy induces tumor cell death and activates immune responses in vivo, effectively suppressing metastasis.
In summary, targeting interorganelle coupling, as exemplified by disrupting MAM-associated proteins, such as HK2, to cause calcium overload​ or using nanoparticles for coordinated ER-mitochondrial attack, disrupts tumor survival through integrated metabolic and signaling interference. This system-level approach overcomes the limitations of single-target therapies, induces immunogenic cell death, and enhances therapeutic precision. Further research on molecular networks, targeted drug delivery systems, and multi-omics/AI integration for synergistic therapy design is needed. The combination of ICIs is crucial for developing an efficient, low-toxicity tumor treatment framework.

Targeting mitochondria in the TME
The TME, the complex and dynamic environment surrounding a tumor, includes cellular and non-cellular components and is crucial in tumor development, progression, and therapy response. Mitochondrial dysfunction or hyperfunction of cellular components in the TME induces dynamic influences on cancer cell survival; hence, it has potential as a cancer treatment target. In this section, we discuss targeting mitochondrial alterations in the TME cells for tumor therapy.
Targeted therapies focusing on mitochondria in the TME are crucial for enhancing the effectiveness of cancer immunotherapy, primarily by enhancing T cell cytotoxic efficacy and slowing T cell exhaustion. Tanaka et al. verified that bezafibrate could augment the mitochondrial metabolic process in T cells, thereby resisting the immune suppression of T cells through PD-1/PD-L1 and improving T cell cytotoxicity in mice [270]. To overcome T cell exhaustion induced by mitochondrial loss and dysfunction, Baldwin et al. co-cultured CD8+ T and bone marrow stromal cells, revealing mitochondrial transfer from stromal cells to T cells as intercellular nanotubes, and the obtained T cells exhibited strengthened metabolic fitness and cytotoxicity to solid tumors [162]. This study identifies intercellular mitochondrial transfer as a pioneering model for organelle-based therapeutics, facilitating the development of advanced cellular treatment strategies. In addition to increasing the number of mitochondria, blockade of mitochondrial loss functions in similar ways. Saha clarified that tumor cells can remove mitochondria from immune cells via physical nanotubes [118]. Therefore, nanotube formation inhibition is a potential target, and the combination of farnesyltransferase and geranyltransferase 1 inhibitor, L-778123, partially blocks nanotube formation and mitochondrial transfer, thus improving anti-tumor outcomes in an aggressive immunocompetent breast cancer model [118]. Chimeric antigen receptor T (CAR-T) cell therapy, a specific cancer immunotherapy, is subject to CAR-T cell exhaustion [341]. Therapies targeting mitochondria have also shown potential in alleviating this exhaustion, thereby improving the efficacy of CAR-T cell-based treatment. Exhausted CAR-T cells exhibit notable mitochondrial dysfunction characterized by morphological and metabolic derangements [342], which was addressed by designing CAR molecules with 4-1BB costimulatory domains and employing drugs targeting mitochondrial metabolism, including PI3K inhibitors (LY294002, Idelalisib), Akt inhibitors (Akti-1/2), mTORC1 inhibitors (rapamycin); metformin; and PPAR-α agonists (bezafibrate and fenofibrate). These interventions were found to reprogram mitochondrial metabolism, with significantly improved mitochondrial morphology and function. The enhanced FAO and OXPHOS pathways reduce oxygen reliance, facilitating CAR-T cell survival in hypoxic tumor cores. Additionally, the proportion of memory-like CAR-T cells increases, enabling the cells’ persistence within the body for several months. This reduction in cancer recurrence underscores the superiority of mitochondria-targeted therapies in enhancing CAR-T cell functionality. Furthermore, the FDA-approved IDH2 inhibitor, enasidenib, and CAR-T cell culturing method involving a galactose-enriched medium have been shown to modulate mitochondrial metabolism and function, thereby delaying CAR-T cell exhaustion [343, 344]. Similarly, inhibition of mitochondrial IDH2 in TAMs has been reported to remodel their phenotype to promote anti-tumor immunoreactions. Lu et al. showed that tumor-derived succinate-loaded microparticles (SMPs) can remodel the metabolic state of TAMs [345]. Mechanistically, succinate is delivered into the mitochondria and nucleus by SMPs, leading to IDH2 and histone H3K122 succinylation within the lactate dehydrogenase A promoter region, promoting classical M1-like macrophage polarization and, therefore, anti-tumor immunoreactions, indicating that the immune cells in the TME possess the potential to attack tumor cells after remodeling.
In the TME, nonimmune cells play a significant role, and targeting these cells’ mitochondria demonstrates immense potential for tumor therapy. Qi et al. developed a lipid nanoparticle delivery system targeting fibroblast activation protein that simultaneously inhibits HK2 and mitochondrial cytochrome c oxidase I in CAFs [346]. This intervention resulted in a marked reduction in glycolysis and CAF mitochondrial activity, effectively transforming them into “fuel stations” for antitumor immune responses. The glucose stored within these CAFs promoted the activation and proliferation of T cells to overcome TME energy constraints. These findings highlight the critical role of CAF metabolic reprogramming in enhancing the efficacy of immunotherapies and offer a promising strategy for addressing the limitations of current immune checkpoint therapies for solid tumors. These approaches may facilitate the development of next-generation combination therapies.
In summary, mitochondria-targeted therapies in the TME enhance antitumor effects by regulating metabolic reprogramming in immune and nonimmune cells. This approach is advantageous because of its precise modulation of immune-metabolic vulnerabilities. However, the high plasticity of mitochondrial dynamics (e.g., compensatory autophagy or metabolic adaptation) and spatiotemporal specificity of interventions remain challenges. Future advances may require the development of novel intelligent delivery systems combined with immune checkpoint inhibition or epigenetic modulators to establish synergistic metabolic-immune therapies for cancer treatment.

Conclusions
In this review, we explore the mechanisms that maintain mitochondrial homeostasis, how these processes become disrupted in cancer, and the therapeutic potential of targeting mitochondria. The relationship between mitochondrial imbalance and cancer is complex, and although many strategies aim to restore homeostasis, most remain preclinical due to limited understanding of their interactions. Regulatory mechanisms such as mitophagy can exert both pro- and anticancer effects, further complicating therapeutic development.
Several challenges restrict the clinical translation of mitochondria-targeted therapies. First, cancer heterogeneity requires approaches tailored to specific cancer cell subsets and development stages. For example, metabolic reprogramming in is dynamic [347]. Tumor cells were long thought to rely primarily on enhanced glycolysis to support growth, which is the basis for tumor-imaging by FDG-PET [348]. However, increasing studies revealed that OXPHOS increases in specific subset such as cancer cell stem cells or during metastasis [121]. Thus, metabolic states are not uniform but depend on cellular stage and environment. Cancer cells may exhibit both high glycolysis and high OXPHOS, using glycolysis for rapid biosynthesis when nutrients are abundant, while activating OXPHOS under stress. Future therapies will move beyond simply inhibiting glycolysis or oxidative phosphorylation and instead target specific cancer subtypes or metabolic vulnerabilities, or exploit metabolic plasticity to render cancer cells more susceptible to attack. Besides, although ROS were historically viewed as tumor-promoting, recent studies show that antioxidants such as vitamin E can increase cancer risk in some individuals (e.g., smokers) and may reduce the effects of chemotherapy and radiotherapy [349, 350]. The effects of ROS depend on their level, duration, and cellular background [351]. Therefore, ROS-targeted cancer therapy should not focus solely on antioxidants but may aim to elevate ROS selectively in cancer cells to induce death while minimizing harm to surrounding healthy cells. Improving tumor selectivity is challenging but can exploit differences between tumor and normal cells, such as variations in mitochondrial membrane potential, which enables the use of potential-sensitive materials [352, 370]. Additionally, chemical features of the TME, including low pH and hypoxia, can be leveraged to improve selectivity. Third, drug resistance, an inevitable challenge, is driven by intricate mechanisms such as adaptive modifications in mitochondrial functions, enhanced drug efflux mechanisms, upregulation of survival signaling pathways (e.g., anti-apoptotic proteins, e.g., the Bcl-2 family), tumor heterogeneity, and mtDNA mutations, which limits efficacy, conferring selective advantage to neoplastic cells. The probability of resistance can be substantially minimized by utilizing targets related to multiple mitochondrial functions, including electron transport chain components, apoptosis modulators, or ROS signaling. Finally, the rapid evolution of AI and precision medicine has catalyzed groundbreaking advancements in tumor mitochondria-targeted therapies. Predictive computational models enable simulations of drug design and efficacy within virtual tumor environments. Advanced machine learning algorithms can facilitate data analysis with extensive genomic, proteomic, and metabolic datasets to classify patients with cancer into a molecular group receiving distinct mitochondrial-targeted therapy [353]. By combining these cutting-edge technologies and individualized treatment approaches, it is expected that oncology and standards of cancer care can be transformed through improved therapeutic outcomes.