Mitochondrial DNA mutations and intercellular mitochondrial transfer in cancer: mechanisms, biological effects, and clinical potential

Abstract

Mitochondrial DNA (mtDNA) mutations are frequently detected in tumor cells and represent a distinctive aspect of cancer genomics. Common types of mtDNA alterations include single nucleotide variants, insertions and deletions, and copy number changes. These mutations often result in a range of biological effects, encompassing both in situ and ectopic mechanisms. In situ, mutant mtDNA may lead to respiratory chain dysfunction, impairing oxidative phosphorylation and shifting energy production toward glycolysis and other metabolic pathways. This metabolic reprogramming, along with altered glutamine and lipid metabolism, is frequently accompanied by reactive oxygen species accumulation, which can activate pro-tumorigenic signaling cascades and contribute to genomic instability. These changes promote cancer cell proliferation, enhance invasive and metastatic potential, and facilitate immune evasion. Moreover, through mitochondrial transfer mechanisms such as tunneling nanotubes, extracellular vesicles, cell fusion, or gap junction channels, mutant mtDNA can be transmitted to other cells, serving as an important mode of intercellular communication within tumors. This process promotes tumor progression and metastasis, regulates apoptotic pathways, facilitates immune evasion, and enhances therapeutic resistance, allowing mutant mtDNA to exert ectopic effects. Clinically, mtDNA mutations hold substantial potential in oncology, with applications spanning tumor diagnosis, disease monitoring, therapeutic resistance prediction, and prognosis assessment. Analysis of tumor tissue or circulating cell-free mtDNA provides a promising non-invasive approach for these purposes. In addition, the involvement of mtDNA mutations in regulating tumor metabolism and mediating intercellular communication underscores their value as potential therapeutic targets, making them a prominent focus of current cancer research.

Introduction

Mitochondrial DNA (mtDNA) is a small, circular DNA genome residing inside mitochondria and is the only source of DNA outside the cell nucleus in human cells. Mitochondria rely on mtDNA-encoded proteins together with nuclear-encoded proteins to form the respiratory chain that produces ATP via oxidative phosphorylation (OXPHOS) and are additionally involved in a wide range of biological processes [1], such as calcium signaling [2], fatty acid synthesis [3], and the regulation of programmed cell death [4]. In 1963, Margit et al. first provided clear evidence of filamentous DNA molecules within the mitochondrial matrix by employing electron microscopy to examine chick embryo cells [5]. Shortly thereafter, Ellen Haslbrunner, Hans Tuppy, and Gottfried Schatz independently isolated and confirmed the presence of DNA in yeast mitochondria through biochemical approaches [6]. Fred Sanger et al. completed the sequencing of the entire 16,569 base pair human mtDNA using the Sanger sequencing method in 1981 [7]. Human mitochondrial DNA is a maternally inherited, circular genome approximately 16.6 kilobases in length that encodes 37 genes, including 13 polypeptides essential for oxidative phosphorylation—comprising subunits of complex I (NADH dehydrogenase), III (bc1 complex), IV (cytochrome c oxidase), and V (ATP synthase)—as well as 22 transfer RNAs and the 12 S and 16 S ribosomal RNAs required for mitochondrial protein synthesis. In addition to its almost entirely coding sequence, mtDNA contains a distinct 1,122-base pair noncoding control region, known as the displacement loop (D-loop region), which serves as the primary regulatory element for replication and transcription [8–11]. All remaining mitochondrial components, including all subunits of complex II (succinate dehydrogenase) and numerous assembly and regulatory factors, are encoded by the nuclear genome. Disparities or functional defects within mitochondrial respiratory complexes encoded by these genes can disrupt oxidative phosphorylation, impair metabolic homeostasis, and increase cellular oxidative stress. Such alterations create a pro-tumorigenic environment by promoting genomic instability, metabolic reprogramming, and enhanced survival signals that facilitate cancer progression [12, 13]. Unlike nuclear DNA, mtDNA lacks introns and protective histones, features that contribute to its unique genetic properties and vulnerability to damage [14–16]. It is present in multiple copies per cell, ranging from hundreds to tens of thousands, depending on the cell’s energy requirements [17]. Due to the absence of histones, limited DNA repair capacity [18], lack of introns, mtDNA is highly susceptible to damage and accumulates mutations at a significantly higher rate than nuclear DNA [19].

In oncology, the understanding of the relationship between mtDNA mutations and tumorigenesis has been continuously evolving (Fig. 1). Interest in mitochondrial function dates back to Otto Warburg’s observations in the 1920s: Cancer cells predominantly rely on glycolysis for energy production even under aerobic conditions, a phenomenon referred to as the “Warburg effect” [20]. Then in 1956, Otto Warburg postulated that impaired mitochondrial function underlies the fermentative metabolism of glucose to lactate and the suppression of oxidative respiration observed in tumor cells under aerobic conditions [21]. In a study conducted in 1998, Vogelstein et al. identified somatic mtDNA mutations in human colorectal cancers, which were absent in the adjacent normal tissues, providing further evidence that mtDNA mutations accumulate within tumor tissues [22]. This landmark finding was subsequently replicated in many cancer types, such as bladder cancer [23], breast cancer [24], lung cancer [25] and pancreatic cancer [26]. These studies independently confirmed that the majority of tumors harbor readily detectable somatic mutations in their mitochondrial DNA. Subsequent studies have progressively revealed that mutations in mtDNA exert profound effects on tumor biology and are no longer regarded merely as passenger events. Single nucleotide variants [27], large-scale deletions [28], and copy number variations [29] can significantly disrupt cellular energy metabolism and redox homeostasis, thereby driving metabolic reprogramming and promoting increased reliance on aerobic glycolysis to meet the high energy demands of rapid proliferation [30–32]. Furthermore, the accumulation of reactive oxygen species (ROS) and metabolic byproducts resulting from mitochondrial dysfunction has been shown to activate multiple pro-tumorigenic signaling pathways, enhancing the invasive and metastatic potential of cancer cells [33]. Emerging evidence also indicates that mtDNA mutations contribute to immune evasion and therapeutic resistance by altering tumor immunogenicity and modulating the tumor microenvironment (TME), including reshaping the functional states of T cells and tumor-associated macrophages [34–36]. Beyond their in-situ consequences within cancer cells, mtDNA mutations have drawn attention for their potential involvement in intercellular mitochondrial transfer, a process that allows mtDNA to be transmitted between cells and exert ectopic effects in recipient cells [37]. This phenomenon was first identified in 2006 by Spees et al., who demonstrated that cells lacking mitochondrial DNA are capable of acquiring mitochondria from adjacent cells in co-culture conditions [38]. In recent years, an increasing body of research has recognized mitochondrial transfer as a crucial mechanism of metabolic cooperation and adaptation in tumors, serving as an important form of intercellular communication in the tumor context [39]. Targeting or harnessing this process is increasingly regarded as a promising therapeutic strategy [40–42]. In this review, we provide a comprehensive overview of the various mtDNA mutations identified in tumors. We then discuss the in-situ effects of mtDNA mutations on cellular metabolism, immune responses, and tumor metastasis. Subsequently, we describe the ectopic effects mediated by mtDNA mutations through mitochondrial transfer in cancer. Finally, we summarize emerging anticancer strategies that target mitochondria and mtDNA.

Fig. 1.

The fundamental functions of mitochondria and the research history of mtDNA mutations in tumors. Mitochondria function as the hub of cellular energy metabolism, redox balance, apoptosis, and signaling. Since the last century, mtDNA mutations and the associated mitochondrial transfer have drawn attention, and recent cancer research has yielded milestone discoveries

Types and characteristics of mtDNA mutations in tumors

Human mtDNA is typically divided into coding and non-coding regions (Fig. 2). The coding region comprises 37 genes, including 13 protein-coding genes involved in OXPHOS, such as ND1–ND6, CO1–CO3, CYB, and ATP6/ATP8, as well as 22 tRNA genes and 2 rRNA genes [43, 44]. The non-coding region primarily consists of the D-loop region, which regulates mtDNA replication and transcription and represents a hotspot for mutations. In recent cancer-related studies, mtDNA mutations have been identified across a wide spectrum of malignancies, with characteristic alterations observed throughout most regions of the mitochondrial genome [45]. Certain mutations are associated with a wide spectrum of cancers, while others have been reported to show specificity to particular cancer types [46–48]. These mutations are predominantly classified into single nucleotide variants, insertions and deletions, and copy number variations (Table 1). Several studies have indicated that a series of factors, including oxidative stress [49], enzyme-induced damage [50], and the absence of effective repair mechanisms, may act as key inducers of mtDNA mutations [51].

Fig. 2.

The structure of mtDNA and its common mutations in tumors. Mitochondrial DNA is a circular double-stranded genome encoding 13 respiratory chain proteins, 22 tRNAs, and 2 rRNAs. In cancer research, mtDNA is frequently subject to multiple types of mutations, including single nucleotide variants (SNVs), insertions/deletions (indels), and copy number variations (CNVs)

Table 1.

mtDNA mutations identified across various cancers

Cancer type	Sample source	Type of mutation	Mutated region	Detection Method	Reference
Bladder cancer	Tumor tissue	Increase in mtDNA copy number	/	qPCR	[52]
Breast cancer	Blood/tumor tissue	Large-scale deletions/decrease in mtDNA copy number	/	Nest PCR/qPCR	[53]
	Tumor tissue/Cancer cell lines	Decrease in mtDNA copy number	/	qPCR	[54]
	Tumor tissue/Cancer cell lines	Decrease in mtDNA copy number	/	qPCR	[55]
	Tumor tissue	Decrease in mtDNA copy number	/	Sanger Sequencing/qPCR	[56]
	Blood/tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	NGS	[57]
	Tumor tissue	Single nucleotide variant	Pervasive mutations	NGS	[58]
	Tumor tissue	Single nucleotide variant/small indel	MT-ND2, MT-ND5, MT-ATP6,MT-CYB	NGS	[59]
	Blood/tumor tissue	Single nucleotide variant/small indel	MT-CO1, MT-CO2, MT-CO3	Sourced from the database	[60]
	Blood/tumor tissue	Single nucleotide variant	Pervasive mutations	NGS	[61]
	Tumor tissue	Decrease in mtDNA copy number	/	qPCR	[62]
Cholangiocarcinoma cell	Cancer cell lines	Single nucleotide variant/small indel	MT-ND4, MT-ND5, MT-CO1, MT-RNR2, D-loop,	NGS	[63]
Colorectal cancer	Tumor tissue	Large-scale deletions	/	PCR	[64]
	Tumor tissue	Single nucleotide variant	MT-ND1, MT-ND5	Sanger Sequencing	[65]
	Cancer cell lines	Increase in mtDNA copy number	/	qPCR	[66]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	Sanger Sequencing/NGS	[67]
	Tumor tissue	Single nucleotide variant/small indel	D-loop	Sanger Sequencing	[68]
	Plasma/tumor tissue	Single nucleotide variant	MT-ND1	Sanger Sequencing	[69]
	Blood	Single nucleotide variant/small indel	Pervasive mutations, especially MT-RNR2	NGS	[70]
	Tumor tissue	Single nucleotide variant	D-loop	NGS	[71]
Endometrial cancer	Tumor tissue	Decrease in mtDNA copy number	/	qPCR	[72]
Epithelial ovarian cancer	Tumor tissue	Single nucleotide variant/increase in mtDNA copy number	MT-CYB	Sanger Sequencing	[73]
Esophageal cancer	Tumor tissue	Single nucleotide variant/small indel	MT-ND4, MT-ND5, MT-CYB	Sanger Sequencing	[74]
	Tumor tissue	Decrease in mtDNA copy number	/	qPCR	[75]
Gastric cancer	Tumor tissue	Increase in mtDNA copy number	/	qPCR	[76]
	Blood	Increase in mtDNA copy number	/	qPCR	[77]
	Cancer cell lines	Single nucleotide variant	MT-ND5	Sanger Sequencing	[78]
	Tumor tissue	Single nucleotide variant/small indel	D-loop	Sanger Sequencing	[78]
	Tumor tissue	Single nucleotide variant/small indel	MT-ND2, MT-ND4, MT-ND5, MT-CO1, MT-RNR1, D-Loop	NGS	[79]
	Tumor tissue	Single nucleotide variant/small indel	D-loop	NGS	[80]
Glioma	Tumor tissue	Single nucleotide variant	Pervasive mutations	Sanger Sequencing/Sourced from the database	[81]
	Tumor tissue	Increase in mtDNA copy number	/	qPCR	[82]
	Tumor tissue	Large-scale deletions	/	PCR/Sanger Sequencing	[83]
	Tumor tissue	Decrease in mtDNA copy number	/	qPCR	[84]
	Tumor tissue	Single nucleotide variant	Pervasive mutations	NGS	[85]
	Tumor tissue	Single nucleotide variant/small indel/increase in mtDNA copy number	MT-RNR1, D-loop	NGS/qPCR	[86]
	tumor tissue	Single nucleotide variant	MT-ND1, MT-ND2, MT-ND3, MT-ND4L, MT-ND5, MT-ND6	NGS	[87]
Hematological malignancies	Blood/bone marrow	Single nucleotide variant/small indel	D-loop	Sanger Sequencing	[88]
	Blood/bone marrow	Single nucleotide variant/small indel	Pervasive mutations	Sanger Sequencing	[89]
	Blood	Single nucleotide variant	Pervasive mutations	Sourced from the database	[90]
	Blood	Single nucleotide variant	MT-ND4	Sourced from the database	[91]
Hepatocellular carcinoma	Blood/tumor tissue	Large-scale deletions/increase in mtDNA copy number	/	Nest PCR/qPCR	[92]
	Tumor tissue	Single nucleotide variant/small indel/decrease in mtDNA copy number	MT-ND1, MT-ND2, MT-ND5, MT-RNR2, MT-CYB, D-loop	NGS	[93]
	Tumor tissue	Single nucleotide variant/small indel/decrease in mtDNA copy number	Pervasive mutations	NGS/qPCR	[94]
	Tumor tissue	Single nucleotide variant/decrease in mtDNA copy number	D-loop	NGS/qPCR	[95]
	Tumor tissue	Single nucleotide variant	MT-RNR1	Sanger Sequencing	[96]
	Serum	Decrease in mtDNA copy number	/	qPCR	[97]
	Tumor tissue	Single nucleotide variant/small indel	MT-ND6	NGS	[98]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations, especially D-loop	sc-CAMS	[99]
Lung cancer	Blood	Decrease in mtDNA copy number	/	qPCR	[100]
	Tumor tissue	Single nucleotide variant	MT-ND1, MT-ND5	Sanger Sequencing	[65]
	Plasma	Decrease in mtDNA copy number	/	qPCR	[101]
	Blood	Decrease in mtDNA copy number	/	qPCR	[102]
Melanoma	Tumor tissue	Single nucleotide variant	MT-CO1, MT-ATP8, MT-ND5, D-loop	NGS	[103]
Meningiomas	Tumor tissue	Large-scale deletions	/	PCR/Sanger Sequencing	[83]
Oral squamous cell carcinoma	Tumor tissue	Decrease in mtDNA copy number	/	qPCR	[104]
	Blood/tumor tissue	Single nucleotide variant/small indel	MT-ND4, MT-ND5, MT-CYB, MT-CO1	NGS	[105]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations, especially D-loop	NGS	[106]
	Sliva	Increase in mtDNA copy number	/	qPCR	[107]
Pancreatic adenocarcinoma	Serum	Single nucleotide variant/small indel/increase in mtDNA copy number	MT-ND1, MT-ND4, MT-ND5, MT-RNR2, D-loop	NGS/qPCR	[108]
Prostate cancer	Serum/tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	NGS	[109]
	Tumor tissue	Single nucleotide variant/increase in mtDNA copy number	Pervasive mutations	NGS/qPCR	[110]
	Tumor tissue	Increase in mtDNA copy number	/	qPCR	[111]
Renal Cell Carcinoma	Tumor tissue	Single nucleotide variant/small indel	MT-ND1, D-Loop	Sanger Sequencing	[112]
Renal oncocytoma	Tumor tissue	Single nucleotide variant/small indel/increase in mtDNA copy number	MT-ND1, MT-ND3, MT-ND4, MT-ND5	NGS/qPCR	[113]
Sinonasal squamous cell carcinoma	Tumor tissue	Single nucleotide variant	MT-ND4, MT-ND5	NGS	[114]
Skin cancer	Blood	Decrease in mtDNA copy number	/	qPCR	[115]
	Tumor tissue	Single nucleotide variant/small indel	D-loop	Sanger Sequencing	[116]
Testicular germ cell tumors	Tumor tissue	Decrease in mtDNA copy number	Pervasive mutations	NGS	[117]
Thyroid cancer	Tumor tissue	Single nucleotide variant/small indel	MT-ND1, MT-ND4, MT-ND5	Sanger Sequencing	[118]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	NGS	[119]
	Tumor tissue	Single nucleotide variant	MT-ND1, MT-ND4, MT-CO1	Sanger Sequencing	[120]
	Tumor tissue	Single nucleotide variant	MT-ND1, MT-ND3, MT-CO3	NGS	[121]
	Tumor tissue	Increase in mtDNA copy number	/	qPCR	[122]
Pan-cancer	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	Sourced from the database	[123]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	Sourced from the database	[124]
	Tumor tissue	Single nucleotide variant/small indel	Pervasive mutations	Sourced from the database	[125]
	Tumor tissue	Single nucleotide variant/small indel	MT-RNR1, MT-RNR2	Sourced from the database	[126]

Single nucleotide variants

Single nucleotide variants (SNVs) represent the most prevalent type of mtDNA alterations identified in malignant tumors, generally referring to single-nucleotide substitutions in the mitochondrial genome, including transitions and transversions, which predominantly occur in protein-coding genes critical for OXPHOS [125]. The formation of single nucleotide variants is primarily associated with base mispairing during DNA replication induced by various types of damage, a process that occurs with particular frequency in mtDNA [45, 127].

Complex I–encoding genomic region

Complex I is the first large protein complex of the respiratory chains. It catalyzes the transfer of electrons from NADH to coenzyme Q and translocates protons across the inner mitochondrial membrane. The complex is encoded by seven protein-coding genes, specifically MT-ND1, MT-ND2, MT-ND3, MT-ND4, MT-ND4L, MT-ND5, and MT-ND6. This region represents one of the most frequently mutated regions identified in tumor research to date, especially MT-ND5 [58, 118, 124, 128]. Mutations in this region have also been widely observed in many types of tumors, including colorectal cancer [22, 69], lung cancer [65], testicular germ cell tumor [117], hepatocellular carcinoma [98] and gastric cancer [78].

Complex III–encoding genomic region

Complex III plays a critical role in electron transfer from reduced coenzyme Q to cytochrome c and contributes to proton translocation across the inner mitochondrial membrane. This complex is specified by a single protein-coding gene, MT-CYB. Although research specifically focusing on MT-CYB mutations in cancer is still scarce, recent studies have identified these alterations in papillary thyroid carcinoma [121], epithelial ovarian cancer [73], and cholangiocarcinoma cell lines [63].

Complex IV–encoding genomic region

Complex IV is the terminal enzyme of the mitochondrial respiratory chain and responsible for the transfer of electrons from reduced cytochrome c to molecular oxygen. The MT-CO1, MT-CO2, and MT-CO3 genomic regions encode this complex. Structural alterations in the associated respiratory complexes have been shown to correlate with immune evasion and poor prognosis in tumors [129, 130]. Recent research on mutations in MT-CO1, MT-CO2, and MT-CO3 in tumors has typically approached these genes as a combined group, rather than examining them in isolation [60, 79, 85].

Complex V–encoding genomic region

Complex V catalyzes the synthesis of ATP from ADP and inorganic phosphate, driven by the proton gradient generated by the electron transport chain. This complex is encoded within two protein-coding genes, MT-ATP6 and MT-ATP8. Several studies have reported MT-ATP gene mutations in tumors [59, 131].

Non-protein-coding regions

There are nearly 25% of mtDNA mutations occurring in non-protein-coding regions including tRNAs and the highly conserved ribosomal RNA [125]. The 22 mitochondrial tRNA genes are located between protein-coding genes, while the 12S and 16S rRNA genes encode core components of the small and large ribosomal subunits essential for mitochondrial protein synthesis. Mutations in these regions are more commonly observed in mitochondrial diseases, but recent studies have also identified them in tumors [126, 132]. In 2021, an article reported the MT-RNR1 mutations in hepatocellular carcinoma [96]. MT-RNR2 mutations have been reported in studies related to rectal cancer [70] and testicular germ cell tumors [117]. SNVs in tRNA genes, such as MT-TL1 are also found in tumors [119].

Non-coding region

Current studies on single nucleotide variants located within the non-coding region of mitochondrial DNA have primarily focused on the D-loop region. The D-loop is the major non-coding region of mtDNA, encompassing the replication origin and transcriptional regulatory elements. As this region does not encode proteins, contains replication origin hotspots, and is subject to limited DNA repair, it represents the one of the most frequent site of SNVs [47, 133]. D-loop region is susceptible to damage by lipid peroxides and ROS, which induce cascades to inflict damage to the mitochondrial OXPHOS system, leading to development of cancer [134–136]. Multiple investigations have reported the occurrence of such single nucleotide variants across various cancer types within this region, including cholangiocarcinoma [63], esophageal cancer [80] and renal cell carcinoma [112].

mtDNA insertions and deletions

Beyond single nucleotide variants, mtDNA insertions and deletions(indels) constitute a major class of structural alterations in tumors, characterized by the gain or loss of nucleotide segments ranging from single bases to multi-kilobase fragments. Replication slippage at repetitive sequences is a common source of insertions and deletions, as polymerase misalignment during replication can introduce extra nucleotides or cause the loss of adjacent bases [137]. Moreover, ROS are capable of inducing double-strand breaks in mtDNA. When these breaks arise within repeat-containing regions, the subsequent repair process may result in large deletions via non-allelic homologous recombination [138, 139].

Large-scale deletions

Large-scale deletions of mtDNA, generally referring to continuous losses of more than 500 base pair, have been increasingly acknowledged as hallmarks of mitochondrial genome instability in diverse human cancers [140]. However, since severe mtDNA deletions often compromise the survival capacity of tumor cells, large-scale deletions are generally reduced in tumor tissues, indicating the presence of negative selection in many cases [124].

Spanning nucleotide positions 8470 to 13,447, the 4977 base pair deletion of mtDNA removes genes encoding essential subunits of respiratory chain complexes, including MT-ND3, MT-ND4, MT- ND4L, MT-ND5, MT-CO3, MT-ATP6 and MT-ATP8, and represents one of the most frequently observed large-scale deletions across various human malignancies [141]. Currently, relevant studies have reported the widely presence of this deletion in hepatocellular carcinoma [92], gliomas and meningiomas [83]. In addition, other large-scale deletions have also been identified in tumor-related studies, such as 4408 base pair deletion in colorectal cancer [64] and 3895 base pair deletion in non-melanoma skin cancer [142].

Small insertions and deletions

Small insertions and deletions are generally defined as insertions or deletions ranging from 1 to 50 base pairs. These events occur at a high frequency in tumor mtDNA and tend to accumulate within specific hypervariable regions, such as the D-loop [143]. Such mutations frequently occur in mononucleotide repeat regions, exemplified by the D310 region, where insertions and deletions in tumor mtDNA have been widely reported and shown to be significantly associated with late-stage disease and poor prognosis [144–146]. Another research in 2018 also reported two special deletions located at positions 3,418–3,432 base pair and 10,941–10,951 base pair in colorectal cancer [67].

mtDNA copy number variations

Mitochondrial DNA copy number variations (CNVs) represent a quantitative genomic shift characterized by the amplification or depletion of entire mitochondrial genomes within cells. Alterations in mtDNA copy number in tumors are frequently reported to be bidirectional, with some studies observing an increase while others reporting a decrease [147]. The causes of the bidirectional alterations in mitochondrial DNA copy number, both among distinct cancer types and within the same tumor type, are not yet fully understood. Nevertheless, considering the central role of mitochondria in cellular metabolism, fluctuations in mtDNA copy number may be linked to metabolic stress and ROS levels.

Increase in mtDNA copy number

An increase in mtDNA copy number is often associated with enhanced OXPHOS activity and metabolic adaptation to sustain high energy demands [66, 148–150]. Moreover, it may also reflect a compensatory response to mitochondrial dysfunction or mutations, aiming to maintain adequate mitochondrial function despite genomic instability [29, 151, 152]. This elevation has been observed across various tumors, ranging from glioblastoma [86], gastric cancer [76], papillary thyroid carcinoma [122], prostate carcinogenesis [111] to bladder cancer [52].

Decrease in mtDNA copy number

Conversely, decrease mtDNA copy number is often linked to a shift toward glycolytic metabolism, enabling cancer cells to thrive under hypoxic conditions [55, 153]. Low mtDNA content may also confer resistance to oxidative stress by decreasing reactive ROS production [151]. In some malignancies, this phenomenon is frequently reported in prior researches, such as breast cancer [62], lung cancer [101], oral squamous cell carcinoma [154] and colorectal cancer [155].

Homoplasmy and heteroplasmy of mtDNA mutations in tumors

Mitochondrial DNA mutations in cancer exhibit two characteristic states, homoplasmy and heteroplasmy, which distinguish them from nuclear genome alterations [156]. Homoplasmy refers to the condition in which nearly all mtDNA copies harbor the same mutation, indicating clonal fixation under selective pressure [128]. Heteroplasmy, by contrast, describes the coexistence of mutant and wild-type mtDNA within the same cell or tissue, with proportions that vary among cells, tissues, and tumor subclones [157]. Moreover, heteroplasmic variants may shift toward homoplasmy during tumor evolution, reflecting clonal selection and adaptive advantage [48, 57].

Homoplasmic mutations often suggest functional relevance in oncogenesis. Only variants that confer proliferative or metabolic benefits are likely to become fixed, making homoplasmy a potential indicator of driver events [128]. Because homoplasmic mutations are highly stable within tumor clones, they are relatively straightforward to detect and can serve as reliable biomarkers for early diagnosis and liquid biopsy applications [158, 159].

Heteroplasmy refers to the coexistence of normal and mutant mitochondrial DNA copies within a single cell [160]. Each cell carries hundreds to thousands of mtDNA copies, and the relative proportion of mutant genomes, the heteroplasmic load, determines functional outcomes [161]. Pathological effects generally arise only when the mutant load exceeds a threshold [162]. Importantly, heteroplasmy is dynamic, with levels shifting during tumor progression or in response to therapy [163]. This makes heteroplasmic variants valuable for monitoring treatment response and recurrence [164]. Low-level heteroplasmy frequently represents passenger events, whereas high-level heteroplasmy, or variants transitioning toward homoplasmy, is more likely to be clinically relevant and associated with tumor progression [151]. Nevertheless, heteroplasmy complicates clinical interpretation, as mutation burdens differ across tissues and tumor subclones, challenging the integration of these data into diagnostic or prognostic models [158].

Despite increasing recognition of their significance, several challenges remain. Detection of low-level heteroplasmy requires ultra-deep or single-cell sequencing, as conventional methods are limited by amplification bias and interference from nuclear mitochondrial sequences (NUMTs) [165–167]. In addition, tumors are composed of heterogeneous components encompassing both malignant and non-malignant cell populations, and such intercellular heterogeneity renders the interpretation of heteroplasmy levels using conventional bulk sequencing approaches challenging [128].

Technological advances in detecting mtDNA mutation

The focus of mtDNA mutation research has tracked methodological advances. Early work detected and analyzed large-scale deletions using Southern blotting and long-range PCR, while PCR–Sanger sequencing targeted SNVs in hotspot regions such as the D-loop [168–170]. With the adoption of high-throughput sequencing (NGS), routine interrogation of low-abundance heteroplasmic SNVs and small indels became feasible [95, 171, 172]. More recently, whole-genome sequencing (WGS) and population-scale cohorts have enabled comprehensive, full-length mitochondrial genome profiling and pan-cancer analyses [102, 123, 124]. WGS is now widely used in large datasets such as TCGA, accelerating a shift from single-variant descriptions to multidimensional integrative analyses.

For mtDNA copy number assessment, qPCR has remained the mainstream approach given its low cost and throughput, though accuracy is affected by amplification efficiency and standard curve variability [102, 173, 174]. Digital PCR, including ddPCR, affords absolute quantification with lower detection limits, supporting precise measurement and method comparison [175]. WGS read-depth methods are increasingly used to infer mtDNA copy number, facilitating joint analyses with mutational profiles, transcriptomics, and other multi-omics layers [176, 177].

In summary, advances in detection technologies have driven mtDNA mutation research in cancer toward higher precision, genome-wide coverage, and pan-cancer integrative analyses supported by bioinformatics resources. Future efforts should focus on improving sensitivity, standardization, and accessibility, while leveraging single-cell and spatial omics to capture intratumoral heterogeneity and accelerating the translation of these methodologies into clinical practice.

The in-situ effects of mtDNA mutations in tumors

In cancer, mutated mtDNA initially exerts a range of biological effects within the cell in which it originates, an influence that is described as its in-situ effect. They have emerged as key modulators of tumor biology beyond their classical role in OXPHOS [178]. Accumulating evidence suggests that mtDNA mutations are not merely passive byproducts of tumorigenesis, but instead actively reshape cellular metabolism [179, 180], redox balance [181], immune interactions [182, 183], chemotherapy resistance [80, 184], metastatic potential [33] and apoptotic sensitivity [185]. These mutations frequently impair components of the electron transport chain, leading to metabolic rewiring that favors aerobic glycolysis and sustains proliferative capacity [186, 187]. These biological consequences underscore the multifaceted role of mtDNA mutations in promoting tumor progression (Fig. 3).

Fig. 3.

The in-suit biological effects of mtDNA mutations in tumors. Mitochondrial DNA mutations in cancer can induce broad biological effects, including reprogramming of energy and material metabolism, accumulation of ROS with consequent genomic instability, dysregulation of apoptotic pathways, altered immune responses, and enhanced migratory and metastatic potential

It is important to note that mtDNA mutations and changes in mitochondrial mass represent related but conceptually distinct layers of mitochondrial regulation. mtDNA mutations primarily affect the genetic integrity and functional competence of individual mitochondria, influencing oxidative phosphorylation efficiency, redox balance, and mitochondrial signaling [45, 188]. In contrast, changes in mitochondrial mass reflect quantitative alterations in total mitochondrial content driven by biogenesis, mitophagy, or mitochondrial turnover, and do not necessarily imply intrinsic defects in mitochondrial function [189, 190]. Although these processes are mechanistically distinct, mtDNA mutations frequently induce secondary changes in mitochondrial mass through compensatory or degenerative responses, thereby jointly shaping cellular metabolic states and stress responses [191].

Metabolic reprogramming

mtDNA mutations have emerged as potent regulators of cancer metabolic reprogramming, influencing not only mitochondrial respiration but also glycolysis, amino acid and lipid metabolism, lactate production [192].

Alterations in energy metabolism

Mutations in mtDNA, particularly in genes encoding electron transport chain components, are frequently detected in human cancers and are closely associated with impaired OXPHOS [193]. These alterations disrupt electron transport efficiency, resulting in reduced ATP production [194] and altered NAD⁺/NADH homeostasis [195], thereby imposing bioenergetic and redox stress that necessitates compensatory metabolic rewiring to sustain cellular viability and proliferation [196].

A common adaptive response is increased reliance on aerobic glycolysis [197, 198], driven by stabilization of HIF-1α [199] and activation of c-MYC [200], which together upregulate key glycolytic enzymes such as hexokinase 2 (HK2), glucose transporter 1 (GLUT1), and lactate dehydrogenase A (LDHA) [201, 202]. As a result, mtDNA-mutant cells exhibit elevated glucose uptake and lactate production even under normoxic conditions, a hallmark of the Warburg effect [203]. Additional mechanisms reinforcing this shift include downregulation of the mitochondrial pyruvate carrier, which limits pyruvate entry into mitochondria [204] and activation of AMP-activated protein kinase (AMPK) in response to reduced ATP/AMP ratios, leading to mTOR suppression and metabolic stress adaptation [205–207].

Despite enhanced glycolysis, mitochondrial oxidative metabolism frequently remains partially functional in tumors harboring mtDNA mutations. Many cancer cells preserve residual Electron Transport Chain (ETC) activity or engage compensatory pathways to sustain mitochondrial respiration [208–211]. Glycolysis and OXPHOS therefore operate in a dynamically coordinated manner rather than as mutually exclusive states, enabling metabolic flexibility under fluctuating oxygen and nutrient availability [212–214]. In addition, mtDNA alterations may contribute to the reverse Warburg effect, whereby cancer-associated fibroblasts (CAF) undergo aerobic glycolysis and secrete lactate or ketone bodies that are imported and oxidized by mtDNA-deficient tumor cells via residual mitochondrial pathways [215, 216]. Beyond ATP generation, increased glycolytic flux also supports anabolic growth and redox balance by supplying biosynthetic intermediates and NADPH [217, 218].

Despite extensive evidence linking mtDNA mutations to impaired OXPHOS and compensatory glycolytic remodeling, whether mitochondrial respiratory dysfunction ultimately promotes or constrains tumor progression remains highly controversial [156]. While severe OXPHOS impairment can impose bioenergetic limitations and reduce tumor fitness, particularly under in vivo selection pressures, partial respiratory defects may instead confer a selective advantage by enhancing metabolic plasticity and stress tolerance [219, 220]. This apparent paradox suggests a non-linear relationship between mitochondrial dysfunction and tumorigenic potential, raising the unresolved question of whether an “optimal window” of OXPHOS impairment exists. Importantly, the functional consequences of mtDNA mutations appear to be strongly context-dependent, influenced by heteroplasmy levels, nuclear genetic background, tumor type, and microenvironmental conditions [221]. Many existing studies rely on in vitro systems or binary classifications of mitochondrial function, which may oversimplify these complex dynamics. Future work integrating heteroplasmy-resolved analyses, isogenic models, and in vivo metabolic tracing will be essential to disentangle causality from adaptation and to define when, and under what conditions, mitochondrial dysfunction serves as a driver rather than a bystander in cancer progression.

Rewiring of glutaminolysis and lipometabolic pathways

Tumor growth critically depends on glutamine, a non-essential amino acid (NEAA) abundant in circulation [222]. As a key anaplerotic substrate for the tricarboxylic acid (TCA) cycle, glutamine provides carbon and nitrogen for the biosynthesis of proteins, lipids, nucleotides, and non-essential amino acids, while also serving as a precursor for glutamate and glutathione (GSH) to maintain redox homeostasis [223, 224]. With impaired electron transport and reduced TCA cycle flux, glutaminolysis becomes the primary anaplerotic route in the tumors [225] Glutamine-derived glutamate is converted into α-ketoglutarate (α-KG), replenishing TCA intermediates and, under mitochondrial dysfunction or hypoxia, fueling reductive carboxylation via IDH1/2 to generate citrate for lipid biosynthesis at the expense of NADPH [226, 227].

In parallel, lipid metabolism is extensively reprogrammed to support membrane biosynthesis, signaling, and energy storage [228, 229], representing a hallmark of cancer metabolic adaptation within the TME [230]. Citrate produced via the TCA cycle or reductive carboxylation is exported to the cytoplasm and converted by ATP-citrate lyase (ACLY) into acetyl-CoA, fueling de novo lipogenesis through fatty acid synthase (FASN) and acetyl-CoA carboxylase (ACC) [231, 232]. Transcription factor SREbase pair1, activated by mitochondrial stress, orchestrates lipid metabolic gene expression [233, 234]. In select tumor types, fatty acid oxidation (FAO) functions as a vital energy pathway, and increased CPT1 activity enables efficient utilization of fatty acids to support tumor growth and survival [235, 236]. In addition, cancer cells often enhance fatty acid uptake through overexpression of fatty acid transport protein (FATP) and CD36, while accumulating evidence highlights the emerging signaling roles of lipids in cancer progression [237–239].

Substantial evidence links mitochondrial dysfunction to enhanced glutaminolysis and lipometabolic rewiring in cancer. however, whether these metabolic programs represent universal compensatory mechanisms remains unresolved. Not all mtDNA mutations or OXPHOS defects uniformly confer glutamine or lipid dependency, indicating strong context dependence shaped by tumor type, nuclear genetic background, and microenvironmental constraints [240]. The relative contribution of glutamine-derived carbon to anaplerosis, redox homeostasis, or lipid biosynthesis varies markedly across experimental models, and reductive carboxylation appears to be selectively engaged rather than obligatorily activated [241–243]. Furthermore, many conclusions regarding glutamine addiction and lipogenic reliance are primarily derived from in vitro systems with supraphysiological nutrient availability, which may overestimate pathway dependence.

Other metabolic alterations

Tumors harboring mtDNA mutations often display distinct changes in lactate handling. Increased glycolytic flux in these tumors results in elevated lactate production, facilitated by upregulated LDHA [244]. Lactate may accumulate within the TME, facilitating immune evasion by acidifying the TME and impairing the cytotoxic activity of T and NK cells [245–247].

Nucleotide metabolism is also perturbed under mitochondrial stress. The activity of dihydroorotate dehydrogenase (DHODH), an enzyme that couples de novo pyrimidine synthesis to the mitochondrial electron transport chain, is often compromised, resulting in altered pyrimidine pools and increased reliance on salvage pathways to sustain DNA and RNA synthesis during proliferation [248–251].

mtDNA mutations induce mitochondrial stress that is communicated back to the nucleus via “retrograde” signals [252, 253]. The key mediators of this retrograde response include transcription factors such as activating transcription factor 4 (ATF4) [254], nuclear factor kappa B (NF-κB) [255], and nuclear factor erythroid 2–related factor 2 (NRF2) [256], which regulate genes involved in amino acid metabolism, antioxidant defense, and mitochondrial biogenesis [13]. Additionally, mitochondrial dysfunction can activate the integrated stress response (ISR) [257], promoting the expression of activating transcription factor 5 (ATF5) and C/EBP homologous protein (CHOP), which further modulate cellular metabolism and apoptosis resistance [258].

ROS modulation

mtDNA mutations disrupt ETC activity in tumor cells, leading to increased mitochondrial electron leakage and elevated ROS production [259–261]. Rather than acting solely as cytotoxic byproducts, ROS function as active signaling mediators in cancer biology [262]. Moderate ROS elevation stabilizes HIF-1α, enhances NF-κB signaling, and activates mitogenic pathways including PI3K/AKT/mTOR and MAPK/ERK, thereby promoting tumor proliferation, survival, and inflammation [263–267]. ROS also influence the tumor stroma and immune landscape, facilitating immune evasion and metastatic potential [268]. Mitochondria-dependent dysfunction in both NK cells and dendritic cells further reinforces this suppressive milieu by diminishing cytotoxic activity and weakening antigen-presentation pathways critical for tumor control [269, 270].

However, excess ROS is inherently cytotoxic [271], necessitating adaptive antioxidant responses in cancer cells to maintain redox balance [272]. This redox balancing act is achieved via upregulation of antioxidant enzymes, including manganese superoxide dismutase (SOD2), glutathione peroxidase (GPx), and activation of the NRF2 transcriptional program [273–275]. Moreover, increasing evidence links ROS to tumor metastasis [276] and angiogenesis [277].

Whether ROS act as oncogenic drivers or merely toxic byproducts of mitochondrial dysfunction remains unresolved. The classical view links ETC impairment to elevated ROS that promote tumorigenesis through redox signaling and genomic instability. However, increasing evidence supports a biphasic model in which moderate ROS levels facilitate adaptive signaling, whereas excessive ROS imposes cytotoxic stress and restricts tumor growth or metastasis [278–280]. Importantly, different mtDNA mutations do not uniformly alter ROS production, and may variably affect ROS magnitude, species, and subcellular localization [278]. Interpretation is further complicated by methodological inconsistencies, as total ROS, mitochondrial ROS, cytosolic ROS, and oxidative damage markers are often conflated, rendering many proposed mechanisms correlative rather than causal [281]. Moreover, it remains unclear which ROS signals confer selective advantage in tumors, including H₂O₂-mediated signaling, lipid peroxidation, or reversible protein thiol oxidation. Resolving these issues will require spatially and chemically resolved approaches that define ROS thresholds and causal signaling pathways, enabling a shift from descriptive redox imbalance to mechanistic understanding.

Immune modulation

Recent studies increasingly recognize mtDNA mutations not only as indicators of metabolic dysregulation but also as active regulators of immune signaling within the TME, exerting context-dependent effects that can enhance antitumor immunity while simultaneously facilitating immune evasion [282].

DAMPs and antitumor immunity

Disruption of mitochondrial integrity often results in the unintended release of mtDNA fragments into the cytosol, particularly under conditions of oxidative stress, mitochondrial membrane permeabilization, or mitophagy failure [283, 284]. These released fragments act as potent immunostimulatory danger-associated molecular patterns (DAMPs), which are sensed by cytosolic pattern recognition receptors (PRRs), most notably the cyclic GMP–AMP synthase (cGAS) [285]. Upon recognizing cytosolic mtDNA, cGAS becomes activated and catalyzes the production of cyclic GMP–AMP (cGAMP), which binds to and activates the adaptor protein STING located on the endoplasmic reticulum membrane [286, 287].

Activation of the cGAS–STING axis initiates TBK1- and IRF3-dependent signaling, inducing IFN-I responses and proinflammatory cytokines such as IL-6 and TNF-α. These signals enhance dendritic cell maturation, antigen cross-presentation, and cytotoxic T lymphocytes (CTLs) activation, thereby promoting antitumor immunity [288], though this pathway has also been implicated in promoting tumor progression in certain cancers [289, 290].

However, accumulating evidence indicates that STING signaling in tumors is highly context dependent and may also promote immune tolerance and tumor adaptation under conditions of chronic or excessive activation [291, 292]. The immunological outcome appears to be influenced by the timing, magnitude, and cellular context of mtDNA release, underscoring the unresolved question of whether mtDNA-driven innate immune activation predominantly restrains or supports tumor progression. Emerging associations between mtDNA alterations and immunotherapy responses remain controversial and require validation through prospective and mechanistic studies.

Metabolic-immunologic reprogramming

Impaired mitochondrial respiration drives a metabolic shift toward aerobic glycolysis, resulting in lactate accumulation [244]. Lactate exerts immunosuppressive effects by inhibiting cytotoxic T lymphocytes and NK cells while promoting the expansion and stability of Tregs [13]. In parallel, elevated succinate levels caused by reduced TCA cycle flux inhibit prolyl hydroxylases and stabilize HIF-1α [293, 294]. Lactate accumulation further suppresses cytokine production, including TNF-α, IL-2, and IFN-γ, through interference with MAPK p38 and JNK/c-Jun signaling pathways [295].

Tumor-intrinsic metabolic imbalance also affects immune cells within the TME. Reduced ATP availability impairs the energy-demanding functions of antigen-presenting cells and weakens costimulatory signaling [296]. Increased ROS resulting from mitochondrial dysfunction further compromises macrophage and dendritic cell function [297]. Together, these immunometabolic alterations favor the dominance of immunosuppressive populations such as M2 macrophages and MDSCs within the TME [298–300].

From a broader perspective, this metabolism-associated immunologic reprogramming highlights immune modulation across three interconnected layers, including antigen presentation and interferon signaling, metabolic competition between tumor and immune cells, and the establishment of immunosuppressive niches [301–303]. Elucidating how mtDNA alterations engage these layers in a context-dependent manner will be critical for defining when metabolic and immunologic rewiring promotes immune evasion versus antitumor immunity.

Regulation of apoptosis and metastatic potential by mtDNA mutatons

In tumor cells, alterations in mitochondrial DNA profoundly reshape regulated cell death through disruptions in energy metabolism, ROS dynamics, and membrane integrity, thus facilitating cancer progression and metastasis [304–306].

In healthy cells, apoptotic stimuli trigger BAX/BAK pores in the mitochondrial outer membrane, releasing cytochrome c and mtDNA into the cytosol [307]. But in cancer cells, mtDNA mutations frequently impair ETC subunits and OXPHOS, compromising mitochondrial outer membrane permeabilization (MOMP) and suppressing cytochrome c release, caspase-9 activation, and apoptosome assembly [308, 309]. In parallel, mtDNA mutations alter ROS homeostasis. Sustained moderate ROS elevation activates survival pathways, including NF-κB and HIF-1α signaling [263, 310, 311], which promote anti-apoptotic programs, suppress pro-apoptotic factors, and may directly inhibit caspase activity through oxidative modification [312]. Consequently, cells evade intrinsic apoptosis even under stress conditions [313, 314]. However, the effect of mtDNA mutation–induced mitochondrial dysfunction on apoptotic susceptibility remains controversial. Respiratory chain impairment and ROS accumulation can promote apoptosis, whereas tumor cells often adapt through anti-apoptotic signaling, metabolic compensation, and mitochondrial biogenesis, thereby bypassing apoptotic checkpoints [45, 156]. Whether specific mtDNA mutation classes preferentially trigger switches between apoptosis, necroptosis, and ferroptosis remains unresolved, and current evidence for such death pathway interconversion is fragmented.

Similarly, mtDNA mutations cause the glycolytic metabolic changes resulting from OXPHOS, enabling tumor cells to survive in hypoxic and malnourished environments [315]. Enhanced glycolysis sustains biosynthetic activity and the energetic demands of invasive growth [316, 317]. Meanwhile, elevated ROS activates pro-metastatic signaling pathways such as NF-κB [263], promoting angiogenesis, extracellular matrix remodeling, inflammatory responses [318, 319], and epithelial–mesenchymal transition (EMT) through upregulation of Snail and Twist [320]. Nevertheless, the role of mtDNA in tumor metastasis remains a matter of debate. Clinically, reduced mtDNA copy number has been associated with osteosarcoma metastasis [321], and increased mitochondrial mutation burden correlates with metastatic progression in non-small cell lung cancer [322], whereas pathogenic mtDNA mutations were reported to impair spontaneous metastasis in melanoma by limiting tumor cell intravasation and circulating tumor cell abundance [220]. These discrepancies likely reflect a non-linear relationship in which moderate mitochondrial dysfunction promotes migratory traits, whereas profound OXPHOS impairment constrains metastatic capacity in a tumor-type- and stage-dependent manner.

The ectopic effects of mtDNA mutations in tumors: mitochondrial transfer

The ectopic effects of mtDNA mutations refer to the biological consequences these mutations exert when transferred outside their original cellular context. This transfer can occur either through the movement of whole mitochondria that carry mutated genomes or through the exchange of mtDNA alone [323–326] (Table 2). Among these possibilities, mitochondrial transfer has been the primary focus of current studies, and several routes including tunneling nanotubes and extracellular vesicles have been identified as active mediators [327]. These pathways enable both malignant and non-malignant cells to donate or acquire mitochondria, thereby altering cellular behavior. Mitochondrial transfer has been demonstrated to restore oxidative phosphorylation, reprogram redox balance, and support tumor proliferation and metastasis in the tumor context [19]. This transfer can also promote immune evasion and inhibit apoptosis, thereby reinforcing therapeutic resistance in tumors [328]. The ectopic functionalization of mtDNA mutations through mitochondrial transfer is thus not merely a rescue mechanism but a mode of intercellular communication in cancer landscape [329].

Table 2.

Mitochondrial transfer in tumors and its biological significance

Cell Type	Mechanism of Mitochondrial Transfer	Direction of transfer	Associated biological effects	Transfer cargo	Reference
Vulvar carcinoma cell, breast cancer cell, pancreatic cancer cell and cancer-associated fibroblast	TNTs	From cancer cell to CAF	Reprogram cellular metabolism, facilitate tumor progression and growth	Mitochondria	[330]
Melanoma cell, breast cancer cell, skin squamous carcinoma cell and Tumor-Infiltrating Lymphocyte	TNTs and EVs	From cancer cell to TIL	Suppress T-cell function and facilitate immune escape	Mitochondria	[103]
Neuron and Breast cancer cell	TNTs	From neuron to cancer cell	Sustain tumor metabolism and facilitate tumor progression and metastatic dissemination	Mitochondria	[331]
Mammary epithelial cell and breast cancer cell	TNTs and EVs	From MEC to cancer cell	Proliferation inhibition and metabolic alterations	Mitochondria	[332]
Macrophage and colorectal cancer cell	EVs	From cancer cell to macrophage	Enhance tumor metastasis and epithelial–mesenchymal transition	mtDNA	[333]
Hepatic stellate cell and hepatocellular carcinoma cell	TNTs and EVs	From HSC to cancer cell	Facilitate tumor growth and suppress apoptosis under stress conditions	Mitochondria	[334]
Astrocyte and glioblastoma	Not specified	From ASC to GBM	Restore mitochondrial respiration and enhance metabolic plasticity	Mitochondria	[335]
Cytotoxic T lymphocyte and breast cancer cell	TNTs	From CTL to cancer cell	Promote tumor cell survival and induce cytotoxic T lymphocyte exhaustion	Mitochondria	[336]
Bone marrow-derived mesenchymal stem cell and colorectal cancer cell	TNTs	From BM-MSC to cancer cell	Promote cancer stemness traits and confer chemoresistance	Mitochondria	[337]
T lymphocyte and glioblastoma cell	TNTs	From T cell to cancer cell	Impair T-cell bioenergetics and compromise T-cell immune function	Mitochondria	[338]
Neural stem cells, brain tumor-initiating cell and astrocyte	TNTs	From NSC/BTIC to astrocyte	Transcriptome reprogramming and proliferation	Mitochondria	[339]
B lymphoma cell	TNTs	Between B cells (bidirectional transfer)	Contribute to tumor cell survival and immune evasion	Mitochondria	[340]
Hepatocellular carcinoma cells	TNTs	From highly invasive cancer cell to low invasive cancer cell	Promote migration and invasion capacity	Mitochondria	[341]
Bone marrow stromal cell and CD8 + T cell	TNTs	From BMSC to CD8 + T cells	Enhance mitochondrial respiration and spare respiratory capacity in CD8 + T cell	Mitochondria	[342]
Adipose stem cell and breast cancer cell	TNTs	From ASC to cancer cell	Acquisition of multi-drug resistance (MDR) to chemotherapeutics	Mitochondria	[343]
Mesenchymal stromal cell and gastric cancer cell	EVs	From MSC to cancer cell	Restore mitochondrial function and enhance chemoresistance	Mitochondria	[344]
Fibroblast and adenoid cystic carcinoma cell	TNTs and cell fusion	From fibroblast to cancer cell	Enhance the migratory and invasive capacities of tumor cells	Mitochondria	[345]
Mesenchymal stromal cell and activated CD8 + T cell	TNTs	From MSC to CD8 + T cells	Facilitate T cell apoptosis and potentiate immunosuppressive activity	Mitochondria	[346]
Colorectal cancer cell and colon epithelial cell	EVs	From cancer cell to normal epithelial cell	Promote tumor proliferation and induce epithelial-to-mesenchymal transition	mtDNA	[324]
Cancer-associated fibroblast and lung cancer cell	EVs	From CAF to cancer cell	Enhance proliferative capacity and promote invasive potential	mtDNA	[347]
Mesenchymal stromal cell and cancer cell	EVs	From MSC to cancer cell	Restore mitochondrial respiratory function and enhance motility and drug resistance	Mitochondria	[348]
Platelet and breast cancer cell	EVs	From platelet to cancer cell	Reprogram metabolism and enhance tumor invasiveness	Mitochondria	[349]
Bone marrow mesenchymal stem cell and KG-1a leukemia stem-like cell	TNTs	From BMSC to cancer cell	Promote cell proliferation and clonal expansion	Mitochondria	[350]
Cancer-associated mesenchymal stem cell and ovarian cancer cell	Cell-to-cell, CICs	From CA-MSC to cancer cell	Enhance chemoresistance, restore proliferative capacity and facilitate metastatic progression	Mitochondria	[351]
Vein endothelial cell and melanoma cell	Endocytosis	From endothelial cell to cancer cell	Enhance proliferative capacity and suppress apoptotic signaling	Mitochondria	[352]
Cancer-associated fibroblast and breast cancer cell	TNTs	From CAF to cancer cell	Improve metabolic capacity and facilitate metastasis	Mitochondria	[353]
T lymphocyte and lung cancer cell	cell-in-cell, CICs	From T cell to cancer cell	Promote metastasis and immune evasion	Mitochondria	[354]
Astrocyte and glioblastoma	TNTs	From astrocyte to cancer cell	enhance oxidative metabolism, promote proliferation, and maintain stemness	Mitochondria	[355]
Microglial cell and neuroblastoma cell	TNTs	Bidirectional transfer	Facilitate toxic protein clearance and mitigate neuronal injury	Mitochondria	[356]
Mesenchymal Stem Cell and glioblastoma Stem Cell	TNTs	From MSC to cancer cell	Boost energy metabolism and chemoresistance	Mitochondria	[357]
lung cancer cell, colon cancer cell and CD8 + T cell	TNTs	From T cell to cancer cell (minor reverse events)	Enhance energy metabolism and promote immune suppression	Mitochondria	[358]
Platelet and osteosarcoma cell	TNTs	From platelet to cancer cell	Facilitate epithelial–mesenchymal transition (EMT) and tumor metastasis	Mitochondria	[359]
Skin-derived mesenchymal progenitor cell and acute myeloid leukemia cell	TNTs	From MPC to cancer cell	Promote chemoresistance and enhance survival capacity	Mitochondria	[360]
Bone marrow mesenchymal stromal cell and multiple myeloma cell	TNTs and cell-to-cell (CICs)	Bidirectional transfer	Confer drug resistance and promote survival advantage	Mitochondria	[361]
Bone marrow mesenchymal stromal cell and acute myeloid leukemia cell	TNTs	From BM-MSC to cancer cell	Confer chemoresistance and attenuate drug-induced apoptosis	Mitochondria	[362]
Macrophage and breast cancer cell	Not specified	From macrophage to cancer cell	Promote cancer cell proliferation	Mitochondria	[363]
Melanoma cell, breast cancer cell, lung cancer cell and T lymphocyte	TNTs	From T cell to cancer cell	Deplete immune cells to facilitate immune evasion	Mitochondria	[364]
Tumor-activated stromal cell and glioblastoma	TNTs, EVs and Endocytic pathway	From TASC to cancer cell	Reprogram metabolism, reduce ROS levels and restore functional mitochondria	Mitochondria	[365]
Mesenchymal stromal cell and melanoma cell	TNTs	From MSC to cancer cell	Promote tumor growth, enhance anti-apoptotic capacity and drug resistance	Mitochondria	[366]
Mesenchymal stromal cell and mtDNA-depleted cancer cell	Not specified	From MSC to cancer cell	Restore respiratory capacity, enable de novo pyrimidine biosynthesis and support tumor growth	Mitochondria	[367]
Cancer-associated fibroblast and lung cancer cell	TNTs	From cancer cell to CAF	Facilitate survival of drug-tolerant cancer cell and promote cancer cell proliferation	Mitochondria	[368]
Cancer-associated fibroblast and non-small cell lung cancer	Not specified	From CAF to cancer cell	Enhance mitochondrial function, promote proliferation and invasion	Mitochondria	[369]

Intercellular mitochondrial transfer and mtDNA mutations

The biological significance of intercellular mitochondrial transfer in cancer has attracted increasing attention in recent years. Meanwhile, mtDNA mutations, which represent a common form of mitochondrial genetic alteration in tumors, may participate in the regulation of this process by altering mitochondrial functional states. Current evidence tends to suggest that mitochondrial transfer is primarily associated with downstream functional consequences of mtDNA mutations, such as mitochondrial dysfunction and metabolic stress, rather than being simply determined by specific mtDNA mutational sites per se [128, 370, 371].

Functional mtDNA mutations that impair oxidative phosphorylation and reduce ATP production markedly weaken the ability of cells to maintain energy homeostasis through endogenous mitochondria, thereby increasing their metabolic dependence on exogenous functional mitochondria [372]. Accumulating evidence indicates that such energy stress can drive tumor cells to acquire exogenous mitochondria and mtDNA via mitochondrial transfer, which are then stably retained under selective pressure to restore respiratory function and support tumor growth [373–375]. For example, under stress conditions, TNFAIP2 and M-sec associated pathways have been shown to promote the formation of tunneling nanotubes and to direct mitochondrial transfer from relatively functional cells toward metabolically compromised cells [376, 377]. In addition, several metabolism-related substances, such as glucose [378], ADP [379] and CD38 [380], have been implicated in mitochondrial transfer. Although these mechanisms have been predominantly characterized in non-tumor systems, they provide important conceptual frameworks for understanding the directionality and execution of mitochondrial transfer under metabolic stress.

Concurrently, respiratory chain related mtDNA mutations are frequently accompanied by increased accumulation of ROS, leading to sustained oxidative stress [41, 108, 268]. Mechanistic studies have demonstrated that ROS can activate stress and growth-related signaling pathways including PI3K-AKT-mTOR, thereby promoting cytoskeletal remodeling and tunneling nanotube formation, which in turn increases the structural capacity for mitochondrial transfer [381]. Moreover, early studies have shown that oxidative stress alone can induce tunneling nanotube formation, depending on stress sensing factors such as p53 [382].

In addition, mtDNA mutation induced instability of mitochondrial membrane potential and accumulation of mitochondrial damage signals lower the functional threshold of mitochondria and compromise their ability to support cellular metabolism and stress adaptation [383, 384]. Beyond primary tumor associated mtDNA mutations, mitochondrial DNA damage induced by chemotherapy or radiotherapy may further exacerbate this process [385–387]. Several mitochondrial damage associated factors, including TFAM [388] and cytochrome C [384], have also been implicated in mitochondrial transfer. However, it remains difficult to distinguish whether loss of mitochondrial membrane potential serves as a direct trigger for mitochondrial transfer or merely reflects a highly stressed cellular state that is intrinsically more permissive to transfer events. Addressing this issue will require the integration of real time imaging approaches with precise functional manipulations to delineate the temporal relationship between changes in mitochondrial functional parameters and mitochondrial transfer events.

Notably, existing evidence has largely focused on global mitochondrial dysfunction or metabolic stress, and has not yet systematically addressed how specific mtDNA mutations or distinct classes of mtDNA alterations influence the frequency, directionality, or efficiency of mitochondrial transfer. Moreover, although multiple stress-related signals are thought to create permissive conditions for mitochondrial transfer, the precise molecular mechanisms that actively promote this process remain insufficiently and incompletely characterized.

While mtDNA mutations can influence intercellular mitochondrial transfer, mitochondrial transfer itself can, under specific conditions, also reshape mtDNA heteroplasmy in vivo [389]. When mitochondria are transferred between cells, recipient cells acquire exogenous mtDNA that coexists with endogenous mtDNA, resulting in mixed mitochondrial genotypes. This has been directly observed in both animal models and patient samples [103, 325, 374]. However, the persistence of heteroplasmy following mitochondrial transfer remains unclear. Whether exogenous mitochondria and mtDNA can be stably maintained within recipient cells is still under investigation [390]. It has been proposed that, in the absence of continuous mitochondrial input, transferred heteroplasmy may be diluted through cell division or eliminated by mitochondrial quality control mechanisms [391, 392]. In parallel, whether mitochondria derived from different donor cell types exhibit distinct fates and functional consequences within recipient cells remains a subject of ongoing debate [328].

From a translational perspective, recent methodological advances, including single-cell sequencing and mtSNV-based analyses, have enabled more refined tracking of mtDNA heteroplasmy changes potentially associated with mitochondrial transfer [167, 393, 394]. However, at the level of whole tumor tissues, it remains difficult to unequivocally attribute heteroplasmy changes to mitochondrial transfer, as clonal expansion [161], therapy-associated selection pressures [80], and stochastic genetic drift [395] can independently remodel mtDNA composition. Circulating mtDNA and extracellular vesicle–associated mtDNA can capture mtDNA mutation spectra and heteroplasmy signals and have been widely explored for cancer diagnosis and monitoring, holding potential to reflect heteroplasmy changes associated with intercellular mitochondrial transfer in patients [48, 61, 108]. However, in the absence of spatial or cellular resolution, analysis of circulating or EV-associated mtDNA alone remains insufficient to directly demonstrate that mitochondrial transfer is the driving force underlying heteroplasmy remodeling in human tumors, and to date no clinical studies have specifically tested mitochondrial transfer using these approaches.

Mechanisms and pathways of mitochondrial transfer

Mitochondrial transfer has emerged as a crucial intercellular communication mechanism that significantly influences tumor progression, therapeutic resistance, and metabolic adaptation [39]. In the tumor microenvironment, cancer and stromal cells can exchange mitochondria via several distinct pathways, including tunneling nanotubes (TNTs), extracellular vesicles (EVs), partial cell fusion (PCF) and Gap junction channels (GJC) [396]. As to the release and capture of free mitochondria, despite being an established transfer route, for example, dependent on heparan sulfate–mediated regulation, it currently lacks experimental verification in tumor (Fig. 4) [397–399].

Fig. 4.

The modes of mitochondrial transfer and the underlying mechanisms. a. Tunneling nanotubes are tubular structures formed by extensions of the plasma membrane, with lengths reaching several hundred micrometers, that directly connect donor and recipient cells. b. A fraction of mitochondria can be encapsulated into extracellular vesicles, released into the extracellular space, and subsequently internalized by recipient cells. c. Partial membrane fusion between donor and recipient cells creates shared cytoplasmic regions, enabling the exchange of mitochondria. d. Gap junctions establish stable intercellular conduits that facilitate the transcellular migration of mitochondria. e. Intact mitochondria can also be released directly into the extracellular milieu and taken up by neighboring cells, achieving transfer without vesicular carriers

Tunneling nanotubes

Tunneling nanotubes (TNTs) are thin, F-actin-based membranous structures that serve as direct cytoplasmic bridges between cells, facilitating long-range intercellular communication over distances of 10 to 200 micrometers. Several studies have indicated that certain TNTs also contain microtubules, particularly those with larger diameters [400, 401]. These dynamic nanotubular connections vary in diameter from 5 to 1000 nanometers and enable the bidirectional transfer of a wide array of intracellular components, including organelles such as mitochondria, proteins, genetic material, and vesicles [327, 402]. TNTs have been observed across various cell types and are functional under both physiological and pathological conditions [356, 403, 404]. In the tumor microenvironment, TNTs have emerged as a critical conduit for mitochondrial transfer between stromal and cancer cells, thereby contributing to tumor progression, metabolic reprogramming, and therapeutic resistance [405].

In cancer systems, TNT formation is predominantly initiated through a protrusion-based mechanism, in which tumor or stromal cells extend actin-rich projections toward neighboring cells to establish contact [396]. A central driver is M-Sec (TNFAIP2), which is significantly upregulated in response to cellular stress and facilitates actin polymerization required for TNT protrusion [377]. Several studies conducted in non-tumor contexts have also demonstrated that the formation process of TNTs is also tightly controlled by Rho GTPases, such as CDC42 and Rac1, which regulate actin dynamics and directional protrusion formation [387, 406]. Furthermore, growth-associated protein 43 (GAP43) has been found to play a role in stabilizing intercellular membrane-associated microtubules in tumor cells [355].

Once TNTs are formed, mitochondrial transport within these structures is mediated by specialized protein complexes. Key among these is the Miro–TRAK1/2 complex, which anchors mitochondria to the actin cytoskeleton and connects them to motor proteins such as Kinesin, Dynein and Myosin-V [407, 408]. These motors enable the unidirectional movement of mitochondria along the TNT axis toward metabolically compromised recipient cancer cells [339, 340, 407]. This unidirectional transport allows metabolically impaired or stressed tumor cells to acquire functional mitochondria from adjacent stromal cells, including mesenchymal stem cells (MSCs) and fibroblasts [353]. The biogenesis of TNTs and subsequent mitochondrial trafficking in cancer are highly responsive to stress-related signaling pathways. Oxidative stress and hypoxia elevate ROS levels, which activate signaling pathways such as Akt and NF-κB, leading to the promoting TNT formation via RalGPS2 or M-Sects [382, 409, 410].

Intercellular mitochondrial transfer via TNTs was first identified in a study by Spees et al. in 2006. Researchers demonstrated the intercellular transfer of functional mitochondria from MSCs to dysfunctional mammalian cells [38]. Recently, multiple studies have also directly confirmed the presence and function of TNT-mediated mitochondrial transfer in various tumors, such as acute lymphoblastic leukemia [411], brain tumor [339], lung cancer [354] and breast cancer [364]. Moreover, although disjunction-based TNT formation has been proposed in other cell types, such as immune or epithelial cells, direct evidence of this mechanism in tumor systems remains limited [412].

Extracellular vesicles

Extracellular vesicles (EVs), including small extracellular vesicles (< 200 nm) and large extracellular vesicles (> 200 nm), are lipid bilayer-enclosed particles secreted by nearly all cell types [413–415]. EVs serve as important mediators of intercellular communication, transporting a wide variety of biological cargos such as proteins, nucleic acids, lipids, and even organelle components like mitochondria, enabling the horizontal transfer of functional biomolecules [416, 417].

In the tumor microenvironment, stromal cells, cancer-associated fibroblasts (CAFs), and immune cells can secrete EVs containing mitochondrial proteins, mtDNA, and intact mitochondria [325, 347, 418]. Large EVs have the capacity to enclose whole mitochondria, in contrast to small EVs, which mainly contain mtDNA and mitochondrial proteins. This encapsulation of mitochondria or mitochondrial fragments into EVs forms what are termed mitoEVs [419]. The process frequently depends on the selective sorting of Snx9 and OPA1 proteins and is regulated by the Parkin/PINK1 signaling pathway, resulting in their extracellular release [323, 420]. Moreover, Rab7-GDP can also prevent the degradation of mitochondria-derived vesicles, thereby indirectly promoting the extracellular release of mitochondria-related cargos via EVs [421].

Once secreted, these EVs are internalized by recipient tumor cells through multiple mechanisms, including clathrin-mediated endocytosis, caveolae-mediated uptake, phagocytosis, and direct membrane fusion [422, 423]. Upon uptake, mitochondrial contents may integrate into the recipient cell’s mitochondrial network, restore or enhance oxidative phosphorylation, modulate intracellular ROS levels, and activate downstream signaling pathways such as the AMPK, NF-κB, or STING axes [333].

Recent studies have provided compelling evidence of this mechanism in various cancers. In colorectal cancer, Guan et al. found tumor-derived EVs enriched with circular mtDNA were found to enhance oxidative phosphorylation and ROS production in adjacent colonic epithelial cells, contributing to tumor progression through paracrine signaling in 2024 [324]. Similarly, in hormone therapy-resistant breast cancer, CAF-derived EVs containing the full mitochondrial genome were shown to restore oxidative metabolism in metabolically dormant cancer stem-like cells [424]. In a study conducted in 2020, Salaud et al. detected tumor-activated stromal cells (TASCs) were demonstrated to transfer mitochondria to glioblastoma cells via both EVs and tunneling nanotubes [365]. However, the mechanisms governing mitochondrial encapsulation, release, and EV uptake remain unelucidated. Notably, mitochondrial dynamics proteins, particularly Drp1-mediated fission and OPA1-regulated membrane remodeling, have been increasingly implicated in facilitating the cytosolic release and subsequent EV-packaging of mtDNA, suggesting a mechanistic crosstalk between mitochondrial dynamics and EV-mediated mitochondrial cargo transfer [425].

Partial cell fusion

Partial cell fusion (PCF) represents a transient, spatially restricted form of intercellular membrane fusion that allows limited cytoplasmic exchange without complete nuclear fusion, which involves localized membrane merging and creating short-lived cytoplasmic bridges [345, 426].

In cancer, partial cell fusion may facilitate the direct transfer of functional mitochondria from stromal or immune donor cells into tumor cells [385, 427]. Functional mitochondria are transferred from donor cells to tumor cells through fused intercellular bridges [374]. This process can be spontaneous or facilitated by specific fusogenic proteins, such as syncytins, which are among the key molecular regulators of cell fusion [428]. Furthermore, upregulation of MFN2 and OPA1 expression ma y facilitate stable cell fusion formation, which constitute a key step in the assimilation of mitochondria into the metabolic circuitry of recipient cells, thereby promoting mitochondrial transfer [345]. Although relatively understudied in tumor biology, partial cell fusion has been detected and confirmed in multiple cancer studies, suggesting it may play a broader role in tumor progression and cell–cell communication [429, 430].

Gap junction channels

Gap Junction Channels (GJCs) are specialized intercellular structures that form direct cytoplasmic connections between adjacent cells, allowing the exchange of ions, metabolites, and small signaling molecules [431]. These channels are composed of connexin proteins, with connexin 43 (Cx43) being a major participant in mitochondrial transfer [432, 433].

Emerging evidence suggests that GJCs play an indirect but essential role in the transfer of larger organelles, such as mitochondria, particularly within the tumor microenvironment [434–437]. While GJCs cannot physically conduct whole mitochondria due to their limited pore size, they act as key regulatory platforms that initiate or support this intercellular organelle transfer [438, 439]. Some studies have also demonstrated that GJCs can facilitate the intercellular transfer of mitochondria by promoting the formation of annular gap junctions [434]. Actually, GJCs enable the bidirectional exchange of bioactive molecules like ROS, which may trigger donor cells to engage in mitochondrial donation [440]. They also modulate the formation and stabilization of TNTs [441]. Moreover, connexin-mediated signaling may regulate the directionality and frequency of mitochondrial movement, influencing which cells become donors or recipients [436].

Mitochondrial transfer as a form of intercellular communication in tumors

Mitochondria are not merely bioenergetic organelles but central hubs of cellular communication [329]. Mitochondrial communication occurs both intracellularly and intercellularly, by releasing mitochondrial-derived signals or physically transferring mitochondria or their components to neighboring cells [442]. With mitochondria acting as both signal producers and responders through release of ROS [443], mtDNA [324], and metabolites such as succinate [444], it plays a pivotal role in intercellular communication. Besides, according to emerging findings, one of the most striking modes of intercellular mitochondrial communication is the horizontal transfer of mitochondria between cells [103, 331, 373].

Within the TME, mitochondrial transfer has emerged as a sophisticated form of cell-to-cell communication, enabling cancer and non-malignant cells to exchange not just molecules, but entire organelles with bioenergetic, genetic, and signaling capacities [332, 445]. Recent studies have revealed that mitochondrial transfer is a widespread phenomenon within the tumor microenvironment, occurring not only between cancer and non-malignant cells but also among cancer cells themselves and between different populations of non-malignant cells (Fig. 5). Notably, some studies have documented bidirectional mitochondrial transfer, underscoring the intricate nature of this phenomenon within the tumor microenvironment [356, 430].

Fig. 5.

The mitochondrial transfer–mediated intercellular communication in tumors. In the tumor microenvironment, mitochondrial transfer facilitates communication between cancer cells to enhance adaptation and resistance, from cancer cells to stromal or immune cells to reshape their functions, from non-malignant cells to cancer cells to restore mitochondrial activity, and among non-malignant cells to sustain homeostasis and influence tumor progression

Directionality and selectivity of mitochondrial transfer

From a broader physiological and pathological perspective, mitochondrial transfer is not unique to cancer but represents a form of intercellular interaction observed in tissue homeostasis, injury repair, and stress adaptation [370, 373]. Within this framework, the directionality and selectivity of mitochondrial transfer are not fixed but are shaped by the functional states of donor and recipient cells as well as by microenvironmental conditions.

Mitochondrial transfer more consistently reflects selection based on mitochondrial functional status rather than mtDNA genotype alone. When recipient tumor cells exhibit mtDNA depletion or severe oxidative phosphorylation defects, they tend to acquire mitochondria containing relatively intact mtDNA from non-tumor cells to restore respiratory function and tumorigenic capacity [357, 358, 372]. Consequently, the most commonly observed outcome is the influx of donor mtDNA capable of rescuing oxidative phosphorylation, which functionally approximates wild-type or otherwise competent mtDNA [367]. However, this does not preclude the transfer of mutant mtDNA. In the TME, tumor cells can also export mitochondria carrying tumor-associated mtDNA mutations to non-tumor cells, consistent with a redistribution of mutational burden and highlighting the context dependence of transfer outcomes [103].

The directionality of mitochondrial transfer is largely governed by the metabolic gap and respiratory stress between donor and recipient cells [364]. Cells experiencing impaired oxidative phosphorylation or sustained metabolic stress are more likely to act as recipients, whereas cells with relatively preserved mitochondrial function preferentially serve as donors. Transfer routes and spatial organization further modulate this process [327, 347]. In addition, the intracellular fate of transferred mitochondria represents a critical selective checkpoint, as only those that evade immune recognition or mitophagic clearance and successfully integrate into the recipient mitochondrial network are likely to be retained [446].

In cancer, these factors often act in concert, resulting in mitochondrial transfer that is bidirectional yet asymmetric [396]. Tumor cells harboring mutant mtDNA and concomitant mitochondrial dysfunction are more likely to acquire functionally competent mitochondria from normal cells to alleviate metabolic stress, while simultaneously exporting their own mitochondria carrying mutant mtDNA and impaired function to surrounding cells [368]. As a result, mitochondrial transfer appears relatively nonspecific in direction at the level of individual events but ultimately trends toward a dynamic redistribution of functionally normal and dysfunctional mitochondria among different cellular populations. This process likely reflects an adaptive equilibrium that emerges at the cellular community level under persistent metabolic pressure, although its precise mechanisms, frequency, and long-term biological consequences remain to be systematically elucidated.

From non-malignant cells to cancer cells

Among the various directions of mitochondrial exchange, the transfer from non-malignant to cancer cells has been most extensively reported in tumor-related studies. This mode of communication becomes particularly evident when cancer cells are exposed to metabolic or oxidative stress, during which they actively acquire functional mitochondria from surrounding non-malignant cells, including CAFs [40, 369], platelets [359], T cells [358], neuron [331], macrophage [363, 380] and MSCs [351, 357] and other cells.

The functional outcomes of this transfer are multifaceted. This process enhances the OXPHOS capacity of cancer cells, improves redox homeostasis, and promotes survival under conditions of metabolic strain or chemotherapy [42]. In this context, mitochondrial uptake acts not merely as energy supplementation, but as a signal of support and adaptation, contributing to tumor cell resilience and progression [447]. Concomitantly, tumor stress elicits augmented mitochondrial biogenesis and supportive shifts in the metabolic profile of donor non-malignant cells [366]. Moreover, the sequestration of mitochondria from immune cells by tumor cells disrupts normal immune cell physiology, leading to functional impairment and unfavorable clinical outcomes, and thereby constitutes a critical mechanism of immune evasion [448]. Collectively, these findings underscore mitochondrial import from non-malignant cells as a pivotal process shaping cancer cell plasticity and resilience, thereby driving adaptation across diverse tumor contexts.

Accumulating evidence suggests that mitochondrial donation to cancer cells now encompasses a wider variety of donor cell types and functional consequences than previously appreciated. This growing body of work underscores the dual potential of mitochondrial transfer as both a mechanistic target for therapy and a prognostic or biomarker tool in clinical oncology.

From cancer to non-malignant cells

In addition to receiving organelles from surrounding stromal cells, cancer cells can also transfer mitochondria outward to non-malignant cells within the tumor microenvironment. Distinct from the supportive exchanges observed in the opposite direction, these events typically involve the extrusion of damaged or dysfunctional mitochondria [449]. The primary recipients of such dysfunctional mitochondria are immune and stromal cells. T lymphocytes that take up mitochondria released from cancer cells often display impaired effector functions, contributing to weakened antitumor immunity [103]. Similarly, stromal cells can acquire defective mitochondria, leading to metabolic disturbances that may indirectly favor tumor progression [380, 430].

The consequences of this unidirectional transfer are therefore twofold. Cancer cells benefit from unloading damaged organelles, while recipient non-malignant cells experience functional compromise [450]. In the case of immune cells, this results in attenuated surveillance and immune evasion, whereas in stromal cells it may lead to maladaptive metabolic reprogramming [370, 451]. These findings highlight mitochondrial export as a strategy by which cancer cells manipulate their microenvironment to reinforce tumor survival.

Research in this area remains limited. At present, only a few non-malignant cell types have been shown to acquire mitochondria from tumor cells, whereas many other potential recipient populations, such as dendritic cells and NK cells, remain to be clarified. Moreover, it remains to be clarified how tumor cells selectively dispose of damaged mitochondria, which recipient cells are preferentially targeted, and whether specific regulatory pathways orchestrate this transfer. These aspects demand deeper exploration.

Within cancer cell populations

Beyond their interactions with stromal and immune cells, cancer cells themselves are capable of exchanging mitochondria, establishing a form of cooperation within heterogeneous tumor populations [452, 453]. This mode of transfer is less frequently described than stromal-to-cancer exchange, yet it provides important insight into how malignant cells collectively adapt to hostile conditions. Typically, highly invasive or metabolically competent cancer subclones serve as donors, transferring mitochondria to less aggressive counterparts with limited bioenergetic capacity [341].

The functional outcomes of this intra-cancer cell exchange are significant. Recipient cells acquire enhanced metabolic flexibility, enabling them to balance glycolysis and oxidative phosphorylation according to environmental needs. At the same time, they gain increased motility, resistance to apoptosis, and in some cases greater tolerance to therapeutic stress. Through these processes, invasive and drug-resistant phenotypes can be horizontally transmitted across tumor cell populations, amplifying functional plasticity and reinforcing tumor progression [454, 455].

However, reports of mitochondrial exchange among cancer cells are currently limited to selected cancer models, leaving unresolved the question of whether this is a universal mechanism or merely a particular occurrence in specific tumors or subclonal subsets.

Among non-malignant cells

Mitochondrial transfer also occurs among non-malignant cells within the tumor microenvironment, though this phenomenon has only recently gained recognition. They may enhance antitumor immunity in some settings while contributing to immunosuppression in others. Stromal cells have been shown to donate functional mitochondria to exhausted tumor-infiltrating T lymphocytes, restoring their respiratory reserve and improving effector activity [342, 346]. This metabolic reinforcement can delay exhaustion and strengthen immune surveillance against tumor cells. In contrast, stromal cells can also transfer mitochondria to regulatory T cells, boosting their suppressive capacity and thereby promoting an immunosuppressive environment that favors tumor progression [456]. Moreover, mitochondrial exchange has also been observed among B lymphocytes [340].

The observation of mitochondrial transfer among non-malignant cell populations underscores that this process is a pervasive feature of the tumor microenvironment rather than a function exclusive to malignant cells. Importantly, its outcomes are context-dependent, encompassing both tumor-promoting and tumor-restraining effects, thereby suggesting potential avenues for clinical intervention.

Biological effects of mitochondrial transfer in tumors

Mitochondrial transfer has been increasingly recognized as a pivotal form of intercellular communication within the tumor microenvironment, profoundly impacting cancer cell fate and disease progression. Through the acquisition of exogenous mitochondria from stromal, immune, or neighboring tumor cells, metabolically impaired cancer cells can restore oxidative phosphorylation, stabilize redox homeostasis, and enhance bioenergetic fitness [396]. Beyond metabolic support, mitochondrial transfer exerts broad biological effects, promoting tumor proliferation, suppressing apoptosis, driving metastasis, modulating immune responses, and conferring resistance to chemotherapy [457]. These biological effects are mechanistically related to the in-situ effects exerted by mutated mtDNA in primary cells; however, comparative studies in this regard are currently lacking.

Metabolic reprogramming and tumor proliferation

Mitochondrial transfer between non-malignant cells and tumor cells plays a central role in promoting tumor cell proliferation by reprogramming cancer metabolism and restoring bioenergetic competence [352, 458]. One of the primary metabolic consequences of mitochondrial acquisition is the restoration of mitochondrial respiration. Many tumor cells exhibit reduced OXPHOS activity due to the Warburg effect. Within the tumor microenvironment, several types of stromal or supportive cells can donate intact, functional mitochondria to neighboring cancer cells. This process enables tumor cells to overcome intrinsic mitochondrial deficiencies, including OXPHOS dysfunction and mitochondrial DNA damage [409, 459]. The earliest studies, conducted in 2015 in the context of breast cancer and melanoma, demonstrated that ρ0 tumor cells lacking mitochondrial DNA exhibited reduced proliferation rates but could restore respiratory function by acquiring mitochondria from neighboring cells [372].

This phenomenon has been documented in a variety of tumors. For example, in 2022, Kumar et al. demonstrated in melanoma that stromal cell-derived mitochondria promote tumor proliferation by restoring OXPHOS capacity [366]. In another study on breast cancer, Bajzikova et al. found that mitochondrial DNA-deficient cells restored tumorigenic potential by acquiring mitochondria from host stromal cells, thereby re-establishing respiration and promoting tumor proliferation [455]. Several other studies in multiple myeloma [380], prostate cancer [40], and glioblastoma [365] have similarly validated this observation.

The restoration of mitochondrial respiration leads to increased ATP production and improved metabolic efficiency, providing the energetic resources necessary for sustained proliferation. In addition, mitochondrial transfer contributes to redox stabilization. Functional mitochondria help reduce intracellular ROS accumulation, alleviating oxidative stress and protecting cancer cells from apoptosis [39]. These redox benefits further support cell cycle progression and prevent growth arrest, which was reported in hematological malignancies [385, 460]. However, a 2023 study in the context of breast cancer revealed that mitochondria derived from macrophages can accumulate ROS within cancer cells and activate the ERK pathway, thereby stimulating cell proliferation [363].

Immune modulation and immune evasion

Intercellular mitochondrial transfer plays a significant role in reshaping immune responses within the TME and is increasingly recognized as a contributor to tumor immune evasion [338]. When mitochondria are transferred from immune cells to cancer cells, immune cells experience marked metabolic disruption [358, 461]. Wang et al. in 2023 reported that mitochondrial transfer occurred between lung cancer cells and infiltrating lymphocytes. Mitochondria were transferred from lymphocytes to tumor cells by cell-in-cell structures (CICs), which subsequently promoted immune evasion by upregulating the expression of PD-L1 [354]. In another article published in 2021, Saha et al. demonstrated that breast cancer cells acquire functional mitochondria from CD8⁺ T cells and NKT cells via intercellular nanotubes, leading to metabolic exhaustion of immune cells, thereby contributing to immune suppression and tumor immune evasion [364].

Mitochondrial transfer also occurs in the reverse direction, from cancer cells to immune cells. Human tumor cells have been shown to deliver damaged or metabolically altered mitochondria to infiltrating lymphocytes [462]. These transferred mitochondria may contain oxidized mtDNA and elevated ROS levels. These transferred mitochondria may lead to the exhaustion of immune cells, and finally cause the immunity evasion [448]. A study published in 2024 demonstrated that cancer cells can transfer mitochondria harboring mutated mtDNA to tumor-infiltrating lymphocytes (TILs). These transferred mitochondria carry the mitophagy-inhibitory factor USP30, enabling them to evade the intrinsic mitochondrial quality control mechanisms of T cells. This results in homoplasmic mitochondrial replacement within TILs, thereby compromising antitumor immunity [103].

Mitochondrial transfer between non-malignant cells within the tumor microenvironment has also been implicated in modulating immune responses. In 2024, Baldwin et al. demonstrated that mitochondrial transfer from bone marrow stromal cells to CD8⁺ T cells via intercellular nanotubes enhances T cell metabolic fitness and significantly improves antitumor immunity in both solid and hematologic malignancies [342]. Moreover, the differentiation of various Th cell subsets can be modulated by MSCs via mitochondrial transfer. It has been demonstrated that the transfer of mitochondria from MSCs to regulatory T (Treg) cells enhances the suppressive capacity of the latter, thereby facilitating immunosuppression [346, 456].

Metastasis and apoptosis of tumors

Mitochondrial transfer plays a pivotal role in promoting tumor metastasis by regulating signaling cascades, cellular phenotype, and intercellular communication, beyond its classical metabolic functions [316]. Transferred mitochondria significantly elevate ROS, which may activate pro-metastatic pathways such as TGF-β, NF-κB, and AKT [454, 463, 464]. These pathways contribute to cytoskeletal remodeling, extracellular matrix degradation, and the induction of epithelial-mesenchymal transition (EMT), a critical program that endows cancer cells with enhanced motility and invasive capacity [359, 465]. In 2025, Hoover et al. demonstrated that neurons promote tumor metastasis by transferring mitochondria to cancer cells, which enhances their metabolic plasticity, redox balance, and survival under metastatic stress, facilitating successful colonization at distant sites [331]. Comparable phenomena promoting tumor progression via mitochondrial transfer have been reported in ovarian cancer [466], osteosarcoma [359] and breast cancer [467], underscoring its relevance across multiple cancer types. However, a study by Zhou et al. in 2024 revealed that mitochondrial transfer occurring in bone-metastatic tumors may suppress tumor metastasis, potentially through the activation of anti-tumor responses mediated by the cGAS/STING pathway [468].

Importantly, mitochondria transferred from highly metastatic to less aggressive cancer cells can confer increased migratory and invasive potential to the recipient population. This lateral transfer of metastatic capability suggests that mitochondrial content is a functional determinant of tumor cell aggressiveness, which has been revealed in researches across various cancers [341, 454, 469].

Mitochondrial transfer has also been implicated in modulating apoptotic resistance in cancer cells, thereby facilitating tumor progression. Functional mitochondria transferred from non-tumor cells sources restore membrane potential and suppress intrinsic apoptotic pathways [348, 355]. Regulation of ROS following transfer can also Inhibit apoptosis [359]. Nevertheless, some studies have shown that mitochondrial transfer can also promote apoptosis under certain conditions [384]. When transferred mitochondria increase intracellular ROS beyond a tolerable threshold, they may trigger mitochondrial membrane permeabilization and activate intrinsic apoptotic pathways [470, 471]. This pro-apoptotic effect appears to depend on the redox state of the recipient cell and the metabolic properties of the donor mitochondria [457].

Therapeutic resistance

Tumor cells frequently develop chemoresistance, a common phenotype acquired through mitochondrial transfer [472]. This phenomenon has been observed in various cancer types where mitochondria transferred from other tumor cells or non-tumor cells in TME enable recipient tumor cells to resist chemotherapy-induced stress [357, 365, 466, 473]. This phenomenon was first observed in a 2013 study by Pasquier et al., which demonstrated that endothelial cells preferentially transfer mitochondria to cancer cells via tunneling nanotubes, leading to enhanced chemoresistance by promoting mitochondrial acquisition and survival advantages in recipient tumor cells [466]. By restoring mitochondrial respiration, reducing reactive oxygen species and stabilizing mitochondrial membrane potential, transferred mitochondria prevent apoptosis and support survival under cytotoxic conditions [359].

For example, in a 2024 study on gastric cancer, He et al. demonstrated that MSCs transferred mitochondria to tumor cells via microvesicles in response to oxaliplatin treatment, restoring mitochondrial function and suppressing mitophagy, thereby enhancing chemotherapy resistance [344]. Del Vecchio et al. demonstrated that mitochondrial transfer from adipose stem cells to breast cancer cells promotes multi-drug resistance by enhancing ATP production, as reported in another study at the same year [343]. This pro-resistance role of mitochondrial transfer has also been observed in other cancer types, such as multiple myeloma [430, 474], acute lymphoblastic leukemia [411] and glioblastoma [475].

Mitochondrial transfer has also been shown to induce stem-like features in cancer cells, further enhancing drug resistance and promoting relapse [476].

However, mitochondrial delivery from healthy non-transformed cells has been shown to sensitize tumor cells to therapy [477]. Mitochondrial introduction from exogenous sources may reverse the Warburg effect and promote cell death under therapeutic stress [478]. Although these studies were conducted in the context of artificially induced mitochondrial transplantation.

Clinical applications of mtDNA mutations in tumors

Based on the in-situ and ectopic effects of mtDNA, a range of clinical applications targeting common mtDNA mutations in tumors has been increasingly recognized. In clinic, mtDNA mutations have emerged as versatile biomarkers with significant clinical implications in oncology [479]. Owing to their high copy number, maternal inheritance, and elevated mutation rate, mtDNA alterations can be detected in both tumor tissues and various body fluids, enabling applications that span early diagnosis, prognostic evaluation, and therapeutic guidance [158]. Meanwhile, recent studies have reported significant advances in multiple emerging therapeutic strategies targeting mtDNA mutations, thereby opening a novel avenue for cancer therapy [480]. These applications have been extended to various aspects of oncology, including diagnosis, evaluation, treatment, and prognosis (Fig. 6).

Fig. 6.

The clinical prospects of mtDNA mutations. Mitochondrial DNA mutations and mitochondrial transfer hold significant clinical implications in cancer. They enable noninvasive diagnosis through liquid biopsy, support risk assessment for metastasis and drug resistance, and provide therapeutic opportunities such as metabolic targeting, mitochondrial transplantation, or transfer inhibition. Moreover, dynamic monitoring of mtDNA mutations aids prognosis and relapse prediction, highlighting their potential in precision oncology

mtDNA mutations as emerging biomarkers for cancer diagnosis

The unique biological characteristics of mitochondrial DNA mutations, including their high mutation rate and multicopy presence in cells, make them promising candidates for clinical biomarker development in oncology [479].

Profiling mtDNA mutations directly from tumor tissues has been explored as a complementary approach to traditional histopathological diagnosis by formalin-fixed paraffin-embedded (FFPE) and quantitative PCR (qPCR). Several studies have reported that certain mtDNA mutational signatures may be specific to tumor types [73, 124, 481]. These profiles can support tumor classification and may also aid in distinguishing malignant from benign lesions [62, 482, 483]. Moreover, due to the relative genomic stability of mtDNA within a tumor, its mutational landscape can serve as a molecular fingerprint. This has enabled its use in identifying the origin of metastatic lesions when the primary site is unknown [484, 485].

Despite these possibilities, tissue-based mtDNA mutation analysis remains limited in clinical use due to the invasive nature of biopsy procedures and tumor heterogeneity [486]. Consequently, the most clinically actionable direction for mtDNA mutations lies in their detection in body fluids, especially as circulating cell-free mitochondrial DNA (cf-mtDNA) [487, 488]. Moreover, mitochondrial DNA enclosed within circulating extracellular vesicles holds potential as a tumor biomarker, representing a form of cell-free mtDNA in a broader sense [61].

Cf-mtDNA refers to mitochondrial DNA fragments released into the bloodstream and other body fluids from apoptotic, necrotic, or actively secreting tumor cells [489]. Circulating cf-mtDNA exhibits several characteristics that make it particularly suitable for cancer diagnostics. It is typically more abundant than nuclear cfDNA due to the high copy number of mtDNA per cell [490]. Moreover, tumor-derived cf-mtDNA tends to carry cancer-specific mutations, especially in regulatory regions like the D-loop, and displays higher levels of fragmentation and oxidative damage [479]. These features contribute to its potential as a disease-specific biomarker. Detection of cf-mtDNA is generally achieved through high-sensitivity techniques such as, quantitative real-time PCR (qPCR) [491], droplet digital PCR (ddPCR) [492] and next-generation sequencing (NGS) [493, 494]. These methods enable both qualitative and quantitative analysis of cf-mtDNA mutations or copy number alterations [495].

In the context of cancer screening and diagnosis, cf-mtDNA analysis from blood samples has shown promising performance. In 2025, Liu et al. developed a high-performing diagnostic model for hepatocellular carcinoma using cf-mtDNA fragmentomic features, particularly in early-stage and AFP-negative cases [496]. In 2022, Vikramdeo et al. demonstrated that mtDNA mutations enriched in circulating can distinguish PDAC patients from healthy individuals, highlighting cf-mtDNA as a promising non-invasive diagnostic biomarker [108]. A 2023 study also reported that cf-mtDNA in colorectal cancer patients exhibited increased coding-region variants and low-level heteroplasmy compared to tumor tissue, indicating its potential as a non-invasive biomarker for CRC detection [497]. Meanwhile, the diagnostic potential of cf-mtDNA is also being explored in other tumors, such as ovarian cancer [498], colorectal cancer [499], gastric cancer and esophageal cancer [489].

While blood remains the most widely used source, cf-mtDNA in other biofluids has shown diagnostic utility as well. A 2022 study by Zhou et al. demonstrated that urinary cf-mtDNA in renal and colorectal cancer patients exhibits unique fragmentation and mutation signatures that enable sensitive, noninvasive cancer detection using NGS [500]. Also, in 2023, Sayal et al. demonstrated that elevated salivary cf-mtDNA levels serve as an independent predictor of poor prognosis in head and neck squamous cell carcinoma [107]. These alternative sample sources offer opportunities for completely non-invasive diagnostics and patient self-collection strategies, particularly in large-scale population screening.

Despite these advances, several limitations remain. The heterogeneous nature of cf-mtDNA release, influenced by tumor type, size, metabolic status, and treatment history, complicates standardization [501]. Furthermore, contamination from leukocyte-derived mitochondrial DNA and pre-analytical variables such as storage and isolation protocols can impact assay reproducibility [502]. Finally, the interpretation of cf-mtDNA mutations must be cautiously approached, as somatic mutations may also arise during aging or in benign conditions, leading to false positives if not properly controlled [45].

Evaluated value of mtDNA mutations in therapeutic sensitivity and metastatic potential

mtDNA mutations have emerged as clinically relevant indicators for predicting therapeutic sensitivity in cancer patients [117, 503]. According to several studies, specific mutations in mitochondrial genes, which impair the function of respiratory chain, have been associated with reduced susceptibility to platinum-based chemotherapeutic agents [164, 504]. Serial monitoring of circulating cell-free mtDNA during treatment enables the detection of dynamic changes in mutation burden, providing real-time insights into evolving drug resistance and guiding timely adjustments to therapy regimens [505]. In targeted therapy, mtDNA mutations may play an important role in determining the efficacy of mitochondrial metabolism-directed agents [506]. By identifying patients with specific metabolic vulnerabilities, mtDNA-based stratification can enhance the precision of therapeutic allocation and potentially improve treatment outcomes [182]. Moreover, mtDNA mutational status has been correlated with response variability to immune checkpoint blockade [186]. High levels of mtDNA instability may alter tumor immunogenicity through changes in reactive oxygen species production and antigen presentation, contributing to poor therapeutic responses in some patients [507, 508]. Related studies suggested that incorporating mtDNA mutation assessment into baseline evaluations could aid in selecting candidates more likely to benefit from immunotherapy [282].

Accumulating evidence also indicates that mitochondrial transfer–associated phenomena may provide clinically relevant biomarkers for evaluating resistance [371]. For instance, upregulation of the mitochondrial transport regulator Miro1 is frequently observed in tumor contexts and reflects enhanced intercellular trafficking capacity, highlighting its potential as a prognostic marker of invasion and metastasis [509, 510]. Similarly, increased expression of the gap junction protein Cx43 at the tumor–stroma interface facilitates mitochondrial transfer via tunneling nanotubes, and its readily detectable protein level suggests possible utility in tissue-based diagnostics [350, 511]. Beyond protein markers, the detection of mtDNA heterogeneity or hybrid mitochondria within T cells may emerge as a liquid biopsy–accessible indicator of bioenergetic impairment and early immune exhaustion, providing a promising avenue for predicting resistance to immunotherapy [512]. Therefore, the assessment of mitochondrial transfer signatures holds potential value for predicting therapeutic resistance and guiding clinical decision-making in immuno-oncology.

mtDNA mutations also hold value in assessing metastatic potential [513]. High mtDNA mutation load has been clinically associated with an increased likelihood of distant metastasis in malignancies such as breast cancer [514], colorectal cancer [65], and head and neck squamous cell carcinoma [515]. Integrating mtDNA mutation profiles with established clinical parameters such as TNM staging and histological grading can further refine metastasis risk stratification. Such combined models may improve the accuracy of predicting which patients require more intensive monitoring and early intervention to prevent disease dissemination [159].

Targeting mtDNA mutations: implications for cancer therapy

Recent advances in mitochondrial biology have paved the way for tumor therapeutic strategies that exploit the mtDNA mutations, ranging from the inhibition of mutant mtDNA–dependent mitochondrial dysfunction [516] to the direct target specific mutations [517]. Artificial mitochondrial transplantation has also emerged as an innovative research direction in tumor therapy [518]. Moreover, the recognition of intercellular mitochondrial transfer as a mechanism of metabolic adaptation and therapy resistance has introduced new targets for treatment [405].

Targeting of mtDNA mutation-induced mitochondrial dysfunction

Defects in OXPHOS caused by mutations in mitochondrial-encoded respiratory chain subunits can alter the efficiency of ATP production, creating metabolic dependencies that are exploitable in cancer therapy [519]. This dependency makes them vulnerable to pharmacological inhibition of mitochondrial respiration, and multiple related agents have already entered clinical use or advanced clinical trials [516, 520]. IACS-010759, a potent small-molecule inhibitor of mitochondrial complex I, has demonstrated significant antitumor activity in high OXPHOS cancer models, including acute myeloid leukemia (AML) and glioblastoma [521]. In addition, several agents, including CPI-613 [522], SR18292 [523] and HP661 [524], have also demonstrated the ability to target metabolic dependencies arising from mutations in mitochondrial respiratory chain subunits in studies.

ROS imbalance represents another vulnerability associated with mtDNA mutation-induced dysfunction. Elevated ROS can promote tumor progression but also sensitize cancer cells to ROS-inducing therapies [196]. These therapies have been developed in various research efforts exploring cancer treatment, such as VB12-sericin-PBLG-IR780 [525] or combination of metformin/efavirenz/fluoxetine [526].

Moreover, therapeutic strategies targeting glutamine metabolism in cancer have also been the focus of considerable research [527, 528]. In addition, some studies have focused on targeting lipid metabolism, such as inhibiting CD36, thereby exerting anti-tumor effects [529]. Inhibiting the replication and repair of damaged mtDNA in tumor cells may also hold therapeutic significance in cancer treatment [530, 531].

Among the available strategies, interventions aimed at mtDNA mutation–induced mitochondrial dysfunction are the most thoroughly investigated and show the greatest progress toward clinical translation. Nevertheless, the inherent complexity and redundancy of metabolic pathways compromise their specificity, leading to considerable toxicity and limiting the prospects for precise tumor-cell eradication.

Direct therapeutic strategies against mtDNA mutations

Targeting pathogenic mtDNA mutations directly offers the potential for highly specific cancer interventions. Drug delivery systems that direct therapeutic agents specifically to mitochondria enhance the effectiveness of treatment pathogenic mtDNA mutations cancer [532]. Lipophilic cations such as triphenylphosphonium (TPP) and mitochondria-penetrating peptides (MPP) can accumulate within the mitochondrial matrix, enabling targeted delivery of chemotherapeutics or OXPHOS inhibitors [533, 534]. Such strategies have been shown to increase drug concentration at the site of action, bypass resistance mechanisms, and minimize off-target toxicity [535–537].

In recent studies, therapeutic approaches have been identified that directly recognize tumor cell–specific mtDNA mutations and subsequently target and eliminate the affected cells with high specificity. In 2023, Li et al. developed a mitochondrial DNA mutation–induced drug release system (MIDRS) that detects tumor-specific mtDNA SNVs and triggers Cas12a-mediated release of photosensitizers or chemotherapeutics, enabling precise tumor cell killing with minimal off-target effects [517]. In another research in 2022, Tsuji et al. developed a novel pyrrole–imidazole polyamide–triphenylphosphonium conjugate (CCC-021-TPP) that selectively recognizes the homoplasmic MT-ND6 A14582G mutation in non-small-cell lung cancer A549 cells [322]. Koshikawa et al. also developed a hairpin-type pyrrole–imidazole polyamide conjugated with triphenylphosphonium (CCC-018-TPP) that specifically recognizes the mtDNA A3243G mutation in 2021 [538]. Similar therapeutic approaches have also been proposed in several other studies [539, 540]. Unlike conventional approaches targeting mitochondrial function, this strategy focuses on single-nucleotide specificity to enable precision cancer therapy, though evidence remains confined to preclinical models without adequate pharmacokinetic or toxicological evaluation.

Genome editing technologies have provided an alternative strategy for directly correcting or eliminating pathogenic mtDNA mutations [541]. Traditionally, the CRISPR–Cas system has been considered unsuitable for mtDNA manipulation because mitochondria lack canonical homologous recombination repair pathways and exogenous RNA cannot efficiently access the mitochondrial matrix [542, 543]. However, advances in protein engineering have opened new avenues for directly targeting and repairing pathogenic mtDNA mutations. Early platforms such as mitochondria-targeted transcription activator-like effector nucleases (mitoTALENs) and zinc-finger nucleases (mtZFNs) selectively recognize mutant sequences and induce their targeted cleavage, allowing the mutant genomes to be replaced by wild-type mtDNA through the cell’s intrinsic replication advantage [544–546]. More recently, DddA-derived cytosine base editors (DdCBEs) have further enhanced the precision of mtDNA editing by enabling C-to-T conversions without generating double-strand breaks, thereby providing a new toolkit for the direct correction of mutant mitochondrial genomes [547]. These tools have progressed further in the experiment of certain mitochondrial genetic disorders, but their application in oncology remains untested.

However, the transition of these technologies from basic research to clinical application still faces substantial obstacles, the most prominent of which is the unresolved issue of off-target effects. Although tools such as DdCBE and TALED achieve mitochondrial localization through mitochondria-targeting sequences, accumulating evidence indicates that they can nonetheless induce detectable off-target edits in nuclear DNA [548, 549]. This reflects several underlying mechanistic gaps, including partial mislocalization of the editors to the nucleus, incomplete mitochondrial targeting efficiency, and the intrinsic lack of absolute sequence specificity in the deaminase domains [550, 551]. These tools can also introduce proximal off-target edits within the mitochondrial genome itself, and certain deaminase variants retain residual RNA-editing activity, further exacerbating unintended molecular perturbations [546, 550, 552]. Such off-target effects cannot yet be fully mitigated through sequence engineering or expression control, thereby limiting the predictability of their in vivo specificity and long-term safety. Future studies may consider enhancing mitochondrial selectivity by engineering improved mitochondrial targeting peptides or implementing controllable expression systems, as well as developing deaminase variants with higher sequence specificity to minimize unwanted base conversions [553, 554].

In addition to off-target effects, other clinical limitations further constrain the translational potential of these mitochondrial DNA editing tools. The spectrum of editable mutations remains extremely narrow. Current platforms can only mediate single-base C-to-T or A-to-G conversions, leaving the majority of pathogenic variants unaddressed and offering no capacity for insertions, deletions, or larger-scale genomic modifications [555, 556]. As a result, their applicability to diseases driven by mtDNA mutations is highly restricted. Meanwhile, delivery challenges continue to be a major barrier to in vivo use. The large size of these editors and their poor ability to traverse the mitochondrial double membrane make efficient and tissue-specific targeting exceedingly difficult [557–559]. Despite these barriers, the development of next-generation mitochondrial gene-editing tools remains a promising direction for future clinical innovation.

Mitochondrial replacement approaches

Mitochondrial transplantation has emerged as a potential strategy to restore normal bioenergetics in cells with defective mitochondria due to pathogenic mtDNA mutations. In cancer, where specific mtDNA alterations can drive metabolic reprogramming, therapy resistance, and aggressive behavior, replacing damaged mitochondria with healthy counterparts offers a means to counteract these changes [560, 561]. Several studies have explored transplanting normal mitochondria into cancer cells to restore their normal metabolism, chemotherapy sensitivity and apoptosis mechanisms, thereby achieving tumor suppression. For example, in 2023, Celik et al. showed that transplanting healthy mitochondria into prostate and ovarian cancer cells enhanced their sensitivity to cisplatin, docetaxel, and paclitaxel, achieving tumor suppression and apoptosis comparable to high-dose chemotherapy while not promoting proliferation [562]. In 2022, Zhou W. et al. also reported that transplantation of healthy mitochondria, especially from female mice, suppressed hepatocellular carcinoma growth by reducing aerobic glycolysis and ROS, arresting the cell cycle, and inducing apoptosis [563]. Comparable anticancer effects have likewise been reported in studies involving various malignancies, such as gastric cancer [564], melanoma [565] and breast cancer [477], highlighting the considerable therapeutic promise of this mechanism across cancer types. It has also been reported that mitochondrial membrane–derived nanovesicles, by mimicking the natural process of mitochondrial transfer, can deliver drugs precisely to tumor mitochondria and synergistically induce cell death, thereby effectively inhibiting glioma [566].

Compared with the widely reported spontaneous mitochondrial transfer within the tumor microenvironment, artificial mitochondrial transplantation exhibits a distinct tumor-suppressive effect [567]. The critical mechanism underlying this difference may lie in the origin and functional state of donor mitochondria [353, 560]. Spontaneous transfer generally occurs from cancer-associated fibroblasts or immune cells, whose mitochondria are often exposed to hypoxia and oxidative stress, leading to functional alterations or metabolic reprogramming. Once acquired by tumor cells, these mitochondria enhance metabolic flexibility and survival capacity, thereby promoting tumor progression [450]. In contrast, artificial transplantation typically introduces healthy mitochondria derived from normal tissues or stem cells, characterized by intact respiratory chains, low mutational burden, and preserved metabolic activity [568]. When incorporated into tumor cells, such mitochondria can correct metabolic deficiencies, restore oxidative phosphorylation, reduce dependence on glycolysis, and potentially reactivate apoptotic pathways, ultimately conferring anti-tumor effects [477, 478].

Several techniques have been developed for delivering exogenous mitochondria into target cells, which were broadly classified into in vivo and in vitro approaches [480]. In vivo methods are primarily designed for therapeutic applications in animal models or clinical settings including direct injection [569], vascular perfusion [570] and intranasal administration [571]. In contrast, in vitro mitochondrial transfer methods are mainly used in mechanistic studies and cell line engineering. MitoCeption involves co-incubating isolated mitochondria with recipient cells followed by centrifugation, promoting mitochondrial internalization through endocytic-like processes [572]. MitoPunch uses a mechanical plunger to force mitochondria through transient pores in the plasma membrane, enabling efficient transfer into large numbers of cells simultaneously [573].

Despite the remarkable therapeutic potential demonstrated by mitochondrial transplantation across various disease models, its clinical translation still faces a series of critical challenges that remain unresolved [480, 518, 574]. The appropriate source of donor mitochondria has yet to be standardized, as mitochondria from different tissues or individuals vary markedly in metabolic properties, mtDNA background, and stress tolerance [575]. Donor-derived low-frequency mutations, age-related damage, or nuclear–mitochondrial incompatibility may further compromise their long-term performance in recipient cells, making the identification of functionally robust and safe mitochondrial sources an ongoing bottleneck [576, 577].

Preserving mitochondrial bioactivity during isolation and storage also remains technically difficult. Ex vivo mitochondria rapidly lose membrane potential and respiratory competence, and the lack of standardized preparation and quality-control frameworks limits batch consistency, reproducibility, and regulatory feasibility [576, 578]. Immunogenicity represents an additional concern that unmethylated CpG-rich mtDNA can activate TLR9 and cGAS–STING pathways, while differences in mitochondrial membrane proteins may trigger phagocytic clearance or adaptive immune responses, potentially amplifying inflammation or tissue injury, especially in immune-dysregulated or tumor-associated environments [579, 580].

Delivery and cellular uptake efficiencies remain low in vivo [581]. Only a small fraction of transplanted mitochondria reaches target tissues and integrates into the host mitochondrial network, with many internalized mitochondria diverted to lysosomal degradation [582, 583]. These limitations collectively lead to transient and unpredictable therapeutic effects, underscoring the need for mechanistic clarification and technological innovation before mitochondrial transplantation can advance toward clinical application.

As an emerging therapeutic approach, mitochondrial transplantation remains in the stage of proof-of-concept validation and mechanistic exploration, and thus is still far from widespread clinical application. Its future development will depend heavily on advances in cell engineering, particularly in delivery systems, mitochondrial engineering, and targeting strategies, while also requiring continued progress in materials science to improve stabilization, controlled release, and in vivo protection of transplanted mitochondria. It is important to note that mitochondrial composition in tumors is highly heterogeneous, and both metabolic demands and genetic backgrounds differ markedly from those observed in inherited mitochondrial disorders. Consequently, strategies that appear promising in the context of mitochondrial genetic diseases may not be directly applicable to cancer therapy. Some studies have focused on the artificial transplantation of specific mitochondrial components. Recent research on mitochondrial diseases suggests that, given the central role of mtDNA mutations in metabolic reprogramming of cancer cells, the targeted transfer of mtDNA may represent a more feasible clinical strategy than whole-mitochondrion transplantation [584].

Inhibition of intercellular mitochondrial transfer

Intercellular mitochondrial transfer has been widely demonstrated to play a crucial role in cancer cell survival, metabolic reprogramming, and therapy resistance. Blocking this process can disrupt mitochondria-mediated intercellular communication within the tumor microenvironment, thereby enhancing the sensitivity of cancer cells to chemotherapy and radiotherapy while improving responses to immunotherapy [370, 409, 585]. Several promising targets for suppressing mitochondrial transfer in cancer treatment have already been elucidated by current studies (Table 3).

Table 3.

Potential targets for inhibiting mitochondrial transfer

Therapeutic target	Prospective therapeutic agent	Detailed mechanism	References
Targeting actin filament polymerization	Cytochalasin D	Disrupt F-actin, inhibit TNT formation, block mitochondrial transfer	[411]
	Cytochalasin B	Suppress actin polymerization, decrease TNT formation and mitochondrial transfer	[586]
	Latrunculin B	Bind G-actin, block polymerization, reduce intercellular mitochondrial transfer	[454]
	Tolytoxin	Disrupt F-actin, lower TNT number and attenuate mitochondrial transport	[587]
	Everolimus	Inhibit mTOR, restrain F-actin polymerization and generation of cellular protrusions, thus diminish TNT formation	[588]
Targeting microtubule polymerization	Vincristine	Inhibit microtubule polymerization, disrupt TNT pathway, block mitochondrial transfer	[460]
	Nocodazole	Microtubule depolymerizing agents, block mitochondrial transport	[460]
	Colcemid	Microtubule depolymerizing agents, disrupt microtubule-dependent TNTs, reduce the potential of intercellular mitochondrial transport	[589]
Targeting small GTPase signaling	NSC23766	Rac1 inhibitors, reduce TNT formation, possess potential capacity to inhibit mitochondrial transfer	[590]
	ML141	Cdc42 inhibitors, decrease TNT number, potentially restrain mitochondrial transport	[591]
	L-778,123	Ras/Rho prenylation inhibitors, hinder TNT formation, decrease intercellular mitochondrial exchange	[364]
	EHT 1864	Rac1 inhibitors, displace GDP/GTP binding site to inactivate Rac1, thereby inhibit Rac1-dependent TNT formation	[592]
	IPA-3	PAK1 inhibitors, diminish α-Syn-induced TNT biogenesis, indicate possible weakening of TNT-mediated mitochondrial transport	[593]
Targeting the Arp2/3 complex	CK-666	Restrain Arp2/3, hinder actin nucleation, diminish TNT	[406]
	CK-869	Arp2/3 inhibitors, prevent branched F-actin nucleation, applied in TNT inhibition experiments	[594]
	Wiskostatin	N-WASP inhibitors, indirectly prevent Arp2/3 activation, decrease TNT formation	[595]
Targeting cell adhesion	Anti-ICAM-1 antibody	Disrupt intercellular adhesion, diminish TNT formation and mitochondrial transfer	[411]
	Anti-β1-integrin antibody	Function-blocking β1-integrin, markedly suppress HGF-induced TNT formation and mitochondrial transport	[596]
Targeting motor proteins	Blebbistatin	Inhibit myosin II ATPase, impair stress fiber contraction, substantially lower TNT occurrence frequency, imply possible restraint of TNT-mediated mitochondrial transport	[597]
	Ciliobrevin D	Dynein inhibitors, block microtubule-dependent bidirectional transport, possibly inhibit intercellular mitochondrial movement	[598]
	Dynapyrazoles	Dynein inhibitors, prevent organelle transport, mitochondrial intercellular transfer relies on motor proteins, show potential suppressive effect	[599]
Targeting CD38	Daratumumab	Anti-CD38 monoclonal antibodies, inhibit mitochondrial transfer under tumor conditions, decrease metabolic capacity of tumor cells	[600]
	Isatuximab	Anti-CD38 monoclonal antibodies, suggest potential role through inhibiting CD38-mediated mitochondrial transfer	[601]
Targeting Cx43	Carbenoxolone	Gap junction blockers, decrease TNT levels, disrupt mitochondrial transport	[338]
	Tonabersat	Cx43-GJ modulators, prevent Cx43-mediated intercellular pathways	[602]
	Meclofenamate	Cx43-GJ blockers, suppress Cx43-dependent intercellular transport	[603]
	αCT1	Cx43 mimetic peptides, selectively suppress channel function, decrease TNT formation	[604]
	Gap26	Cx43 mimetic peptides, suppress Cx43 hemichannels and GJ channels, decrease intercellular communication	[605]
	Gap27	Cx43 mimetic peptides, inhibit Cx43 hemichannels and GJs, regulate intercellular signaling and migration	[606]
Targeting M-sec	NPD3064	Directly interact with and suppress TNFAIP2/M-sec, decrease M-sec-mediated TNT formation	[607]
Targeting EV release	GW4869	Inhibit nSMase2, reduce EV release and enclosed mitochondria	[473]
	Imipramine	Suppress aSMase, decrease EV secretion from tumor cells	[608]
	Manumycin A	Block Ras/nSMase pathway, diminish EV biogenesis and secretion from tumor cells	[609]
	Calpeptin	Calpain inhibitors, decrease EV release from tumor cells	[610]
	Nexinhib20	Rab27a inhibitors, substantially diminish EV secretion in the tumor microenvironment	[611]
	YQ456	Myoferlin inhibitors, reduce EV secretion from tumor cells, significantly suppress tumor progression	[612]
	Ketotifen	Mast cell stabilizers, markedly suppress EV release from tumor cells, increase drug sensitivity	[613]
Targeting EV uptake or endocytosis	Dynasore	Dynamin inhibitors, block EV uptake, inhibit MSC-to-neuron mitochondrial transfer	[614]
	Pitstop-2	Block clathrin terminal domain, prevent CME, diminish EV uptake	[615]
	Chlorpromazine	CME inhibitors, decrease EV uptake by tumor cells	[616]
	Heparin	Compete against HSPG binding, hinder uptake of exogenous mitochondria or EVs	[617]
	PPS	Sulfated polysaccharides, block uptake of exogenous mitochondria or EVs	[617]
	EIPA	Na⁺/H⁺ exchange inhibitors, prevent macropinocytosis, decrease uptake of exogenous mitochondria	[618]
Targeting mitochondrial quality control	CMPD-39	USP30 inhibitors, enhance mitophagy, decrease functional mitochondria available for transfer	[103]
	Urolithin A	Natural metabolites, activate PINK1/Parkin pathway, enhance mitophagy, clear damaged mitochondria, possess potential to inhibit mitochondrial transfer	[619]
	Mdivi-1	Drp1 inhibitors, block excessive mitochondrial fission, disturb dynamic balance, decrease functional mitochondria available for intercellular transfer	[343]
Targeting intracellular ROS levels	DPI	NOX2 inhibitors, hinder mitochondrial transfer under tumor conditions	[41]
	Metformin	Suppress complex I and NOX2/ROS–ICAM-1 signaling, decrease mitochondrial transfer	[620]
	Dexamethasone	Reduce ROS and mitochondrial transfer, enhance Venetoclax efficacy	[362]
	Apocynin	NOX inhibitors, reduce ROS, decrease mitochondrial transfer potential	[621]
	NAC	ROS scavenger, clear ROS generated by injured cells, block mitochondrial transfer	[622]

ROS are recognized as key signals driving mitochondrial transfer in the tumor microenvironment [623]. Excessive ROS can activate relevant transfer pathways, thereby supporting tumor metabolism and chemoresistance. This area has been studied most extensively, and several agents have been reported to inhibit mitochondrial transfer. For example, DPI, a NOX2 inhibitor, has been shown in AML to suppress mitochondrial transfer from stromal to cancer cells by lowering intracellular ROS levels [41]. Metformin, through inhibition of complex I and subsequent attenuation of ROS production as well as disruption of the NOX2/ICAM-1 signaling pathway, similarly reduces mitochondrial transfer and enhances chemosensitivity in AML [620]. Dexamethasone has also been reported to downregulate ROS, thereby reducing mitochondrial transfer and augmenting the activity of Venetoclax [362]. Other agents, including apocynin, may also attenuate ROS levels and thus limit mitochondrial transfer, although direct evidence in tumor contexts remains limited [621].

Current research on inhibiting mitochondrial transfer in cancer largely focuses on preventing TNT formation, with multiple targets showing significant therapeutic potential [405]. Actin remodeling is essential for TNT initiation and elongation, and inhibition of actin polymerization markedly reduces intercellular mitochondrial transport. Cytochalasin D has been shown to block mitochondrial transfer from mesenchymal stromal cells to T-ALL cells and enhance chemosensitivity [411], while Cytochalasin B [586] and Latrunculin B [454] have demonstrated similar effects in other researches.

Microtubules provide the internal scaffold and drive the long-range trafficking of mitochondria within TNTs. Vincristine and Nocodazole have been proven to block TNT-dependent mitochondrial transfer in leukemia models and reduce resistance [460], while colcemid, a microtubule depolymerizing agent, shows comparable potential [589].

Small GTPases such as Rac1 and Cdc42 are direct regulators of TNT formation. Inhibition of these proteins is believed to interfere with the acquisition of mitochondria by cancer cells. NSC23766, a Rac1 inhibitor, reduces TNT formation and may inhibit mitochondrial transfer in cancer [341, 590]. And ML141, a Cdc42 inhibitor, has similar effects [591]. L-778,123, by blocking Ras/Rho protein modifications, significantly reduces mitochondrial exchange between immune and cancer cells [336, 364].

The Arp2/3 complex is a critical nucleator of branched actin filaments. Compounds such as CK-666 [406] and CK-869 [594] have been reported to reduce TNT formation and thereby potentially inhibit mitochondrial transfer, though direct evidence in tumor contexts remains to be established. However, in certain neuronal systems, inhibition of Arp2/3 has paradoxically been reported to promote TNT formation [624].

Cell adhesion molecules also play essential roles in mitochondrial transfer by providing stable interfaces for TNT-mediated exchange. In T-ALL models, anti-ICAM-1 antibodies significantly decreased mitochondrial transfer from MSCs to cancer cells [411]. And anti-β1 integrin antibodies have shown similar potential [596].

Motor proteins represent another class of potential targets, as they directly regulate mitochondrial trafficking within TNTs. Agents such as Ciliobrevin D [598] and Dynapyrazoles [599] inhibit dynein-dependent bidirectional transport, suggesting possible applications in limiting mitochondrial transfer, though direct validation in tumor models is still lacking.

CD38, a transmembrane glycoprotein that promotes TNT formation, has been targeted with monoclonal antibodies in hematological malignancies. Daratumumab has been shown in AML models to block mitochondrial transfer from stromal to cancer cells and reduce metabolic activity [600]. Isatuximab, another CD38 antibody, has not been directly tested in this context, but its clinical use in multiple myeloma suggests similar potential [601].

Cx43–mediated gap junctions and hemichannels contribute to TNT establishment and mitochondrial exchange. Inhibitors such as Carbenoxolone [338, 625], Tonabersat [602], and Meclofenamate [603] have been shown to suppress intercellular communication in models of breast cancer and brain metastasis. In addition, Cx43 mimetic peptides such as αCT1 have been reported to potentially reduce mitochondrial transfer in the tumor microenvironment [604].

M-sec is a membrane-associated protein essential for TNT formation, directly driving the establishment of intercellular conduits and thereby playing a central role in mitochondrial transfer [626]. Recent reports have identified NPD3064 through a chemical array screening approach as a compound that directly binds to and inhibits TNFAIP2/M-sec, thereby reducing M-sec–mediated TNT formation [607]. However, dedicated investigations into its role in tumor-associated mitochondrial transfer are still lacking, and further research is required to establish its therapeutic relevance in cancer.

Beyond TNTs, there is growing evidence that inhibiting EV-mediated mitochondrial transfer is also feasible. EVs represent another important route of mitochondrial trafficking, and blocking their release or uptake reduces the acquisition of functional mitochondria by cancer cells. GW4869, a neutral sphingomyelinase inhibitor, has been shown to reduce EV release and mitochondrial content [416, 473, 627]. Agents such as Imipramine [608] and Manumycin A [609] similarly suppress EV secretion and tumor progression, although direct evidence of mitochondrial cargo remains limited. Another strategy involves blocking EV uptake by recipient cells. For example, the dynamin-related endocytosis inhibitor Dynasore can prevent EV internalization, underscoring its potential relevance in limiting EV-mediated mitochondrial transfer [614, 628].

The quality control of donor mitochondria also influences their ability to be transferred. Enhancing mitophagy can reduce intercellular transfer of functional mitochondria. Recent studies have demonstrated that CMPD-39, a USP30 inhibitor, promotes mitophagy and decreases mitochondrial transfer [103]. Other agents that stimulate mitophagy may likewise contribute to this process [343, 629]. Studies have also found that blocking certain regulatory proteins, such as CD9, CD63 and CD81 can significantly reduce EV release and affect processes related to mitochondrial quality control, thereby indirectly inhibiting mitochondrial transport [630, 631].

Finally, additional proteins have emerged as particularly promising targets. Miro1/2 are small GTPases located on the mitochondrial outer membrane that mediate attachment to microtubule-based motor proteins, thereby driving mitochondrial transport within TNTs [407]. Importantly, studies in tumor settings have shown that knockdown of Miro1/2 significantly suppresses mitochondrial transfer from stromal to cancer cells, highlighting their therapeutic relevance [330, 510, 632]. Gap43, a regulator of cytoskeletal remodeling and synapse-like structures, is also indispensable for TNT generation and stability [404]. Despite their critical roles, no pharmacological inhibitors targeting these proteins are currently available, with most studies relying on genetic manipulation or early tool-based approaches [341, 355]. Collectively, Miro1/2 and Gap43 are recognized as highly promising but pharmaceutically untapped targets for future therapeutic intervention.

Although current approaches remain underdeveloped, given the reported pivotal role of mitochondrial transfer in tumor progression, targeting mitochondrial transfer within tumors represents a highly promising avenue for future research [37, 409]. Unfortunately, there are still no widely recognized and highly specific molecular targets that allow for the precise inhibition of mitochondrial transfer in the tumor microenvironment. The limitation that makes existing interventions prone to unintended disruption of normal physiological processes. Moreover, the molecular nodes that are currently operable are deeply embedded in fundamental cellular functions, and systemic interference with these pathways may introduce substantial off-target effects and toxicity [373]. For example, suppressing Miro1 to block mitochondrial transfer in tumors may induce notable neurotoxicity and disrupt mitochondrial homeostasis [633]. Likewise, the small GTPases required for TNT formation are central regulators of cell morphology, adhesion dynamics, migration, and phagocytosis, and excessive inhibition could compromise tissue homeostasis or impair immune responses [583].

Meanwhile, extensive studies indicate that mitochondrial transfer plays an essential role in tissue repair and organ protection. Transfers from mesenchymal stem cells to injured alveolar epithelial cells or from astrocytes to neurons have been shown to enhance cellular energy metabolism, mitigate acute damage, and improve survival outcomes [634, 635]. Consequently, nonspecific inhibition of mitochondrial transfer may compromise these endogenous reparative mechanisms and introduce additional risks of tissue injury. Evidence from tumor models further demonstrates that mitochondrial transfer is also involved in maintaining the metabolic fitness and antitumor functions of immune cells [342]. Broad suppression of this process may inadvertently impair the immune system’s capacity to mount effective antitumor responses and diminish the overall efficacy of immunotherapy. In addition, most supporting data are limited to cellular and animal studies without robust long-term or clinically relevant validation. Overall, research in this area is still at an early stage, and no therapeutic agents designed primarily to block mitochondrial transfer have yet advanced into clinical trials.

Prognostic and relapse-associated roles of mtDNA mutations in cancer

Dynamic tracking of mtDNA mutation burden through liquid biopsy offers a promising approach for prognostic assessment and early relapse detection in cancer patients [158].

Some mtDNA mutations are regarded as indicators of poor prognosis in cancer [636]. Meanwhile, changes in cf-mtDNA mutation burden can provide early indications of therapeutic outcomes [637]. A marked and sustained reduction in mutation load after therapy is often associated with longer recurrence-free survival (RFS) and overall survival (OS). Conversely, persistent elevation or rapid rebound in mutation levels may indicate the presence of minimal residual disease or resistant tumor cell clones, both of which carry a higher risk of future relapse.

For example, in 2018, Kalavska et al. reported that in ovarian cancer, higher post-treatment cf-mtDNA levels were associated with shorter progression-free survival, suggesting cf-mtDNA could serve as a prognostic biomarker [638]. In 2021, Bulgakova et al. found in a study on non-small cell lung cancer (NSCLC) that increased circulating cf-mtDNA levels prior to treatment were linked to higher risk of metastasis and shorter survival, indicating its potential as a prognostic marker for NSCLC [639]. Moreover, cf-mtDNA has also demonstrated potential as a prognostic biomarker in certain malignancies, including hepatocellular carcinoma [496] and squamous cell carcinoma [640].

Moreover, previous studies have demonstrated that cancer cells can acquire mitochondria from immune cells to enhance their metabolic activity and proliferative capacity, a process closely associated with disease progression and poor prognosis [457, 458]. MERCI is a computational approach that integrates single-cell transcriptomic data with mitochondrial DNA variants to quantify intercellular mitochondrial transfer [461]. Building upon this method, the TMT Score was developed to capture transfer activity through a defined gene expression signature. Notably, high TMT Scores have been strongly correlated with unfavorable survival outcomes, thereby providing a novel molecular tool for prognostic assessment in cancer patients [358].

Conclusions and prospects

Mitochondria are central organelles responsible for energy metabolism, ROS generation, and the regulation of apoptosis, and their functional abnormalities can drive metabolic reprogramming, promote malignant phenotypes, and shape the immune microenvironment. Because mtDNA lacks protective histone structures and has limited repair capacity, it is prone to mutations in the ROS-rich microenvironment, and in tumor tissues it is frequently altered with considerable diversity, including single nucleotide variants, insertions and deletions, and copy number alterations. Mutated mtDNA can exert in situ effects by acting on the mitochondrial function of tumor cells themselves, thereby causing metabolic dysregulation, disruption of ROS homeostasis, altered immune regulation, and aberrant apoptotic signaling, which together enhance malignant phenotypes. It can also be transferred between cells to surrounding or distant recipient cells, where it exerts ectopic effects that induce metabolic abnormalities, increase therapeutic resistance, facilitate immune evasion, and promote metastatic dissemination. At the same time, mutated mtDNA holds potential as a biomarker, showing significant value in early cancer diagnosis, therapeutic resistance and metastasis risk stratification, and prognostic monitoring. In terms of therapy, strategies such as artificial mitochondrial transplantation and inhibition of mitochondrial transfer, although still at the exploratory stage, provide novel approaches to restore mitochondrial function or block oncogenic signaling.

Although the ubiquity of mtDNA mutations in tumors has been established, a major challenge in current research lies in elucidating the specific roles of individual mtDNA mutations in tumorigenesis. Recent studies have attempted to infer the roles of mtDNA mutations in tumorigenesis and cancer progression by estimating the ratio of nonsynonymous to synonymous substitutions (dN/dS) at protein-coding sites [67, 163, 641]. In general, a dN/dS ratio close to 1 indicates neutral selection, suggesting that the biological effect of the mutation on tumor development is essentially neutral. A dN/dS ratio greater than 1 implies that the mutation may promote tumor progression, whereas a ratio less than 1 suggests negative selection, indicating that the mutation could impair tumor survival. Nevertheless, whether a specific mutation in a given gene exerts a particular biological effect, rather than functioning simply as tumor-promoting or tumor-suppressing, remains insufficiently explored. This gap may be attributable to the relatively limited functional repertoire of mtDNA and the frequent occurrence of mutations in tumors. The cytoplasmic hybrid (Cybrid) model [642] and mitochondrial mouse models [643] carrying mtDNA mutations are currently the most commonly used experimental systems for investigating the biological effects of specific mitochondrial DNA mutations. However, both approaches face limitations such as low efficiency and insufficient clinical translatability. Future studies could benefit from integrating mtDNA databases established through WGS to comprehensively analyze the precise nature and biological impact of mtDNA mutations in cancer [125]. In addition, the application of mitochondrial genome engineering technologies to introduce specific mutations in experimental systems may provide a valuable approach for elucidating the biological consequences of individual mtDNA variants in tumorigenesis [547].

Current evidence has consistently shown that mitochondrial transfer in tumors allows mutated mtDNA to exert ectopic effects beyond the primary cells. However, the precise molecular mechanisms underlying these effects remain poorly understood. Whether such ectopic influences operate through the same signaling pathways as those observed within the primary cells, or involve distinct and as yet uncharacterized mechanisms, is still uncertain. This unresolved question has, to some extent, limited a systematic understanding of the functional roles of mtDNA mutations. Future efforts should focus on addressing this limitation by integrating multi-omics approaches, applying real-time imaging of mitochondrial function, and utilizing gene-editing technologies to trace the trajectories of ectopic mtDNA, thereby clarifying the critical nodes of the associated signaling pathways. Meanwhile, earlier studies often relied on membrane potential dyes such as MitoTracker [330] and TMRM [363] to trace mitochondria, but these dyes can themselves transfer between cells or disrupt mitochondrial function, leading to false positives resembling mitochondrial transfer. Recent work employing genetic markers has partially mitigated these limitations and improved the reliability of tracing such events [103, 331]. Although various donor and recipient cell types have been identified in tumor-associated mitochondrial transfer, important gaps remain. In particular, the potential involvement of dendritic cells and NK cells has not yet been explored. Addressing these questions will be essential for constructing a more complete map of mitochondrial transfer in tumors and for deepening insights into tumor metabolism and immune regulation. The drivers of mitochondrial transfer and the characteristics of the transferred mitochondria remain insufficiently studied. Beyond documenting the phenomenon itself, systematic insights into why transfer occurs and which mitochondria are preferentially involved are still lacking, highlighting an important direction for future research.

In clinical translation, therapeutic approaches targeting mutated mtDNA in tumors have progressed from broadly addressing mtDNA-associated metabolic pathways to focusing on specific mutations, consistent with the paradigm of precision oncology. Future advances are likely to build on refined drug delivery systems [644] and robust mtDNA gene-editing platforms [544], aiming to achieve greater efficacy with reduced toxicity. The pervasive presence of mtDNA mutations across cancer types underscores the potential of this strategy to inform pan-cancer therapeutic innovation. Blocking mitochondrial transfer to mitigate the pathogenic effects of mutated mtDNA represents a promising avenue in cancer therapy. However, research is still at an early stage, with insufficient mechanistic insight and limited in vivo validation, making near-term clinical application unlikely. Future efforts should prioritize the identification of efficient, low-toxicity targets and the development of specific inhibitors. Given the ubiquity of mitochondrial transfer in tumors, repurposing existing anticancer agents with latent anti-transfer activity may offer a practical shortcut [600, 620]. Emerging technologies such as artificial mitochondrial transplantation have shown therapeutic potential in certain mitochondrial disorders. Although their application in cancer treatment remains underexplored, this strategy possesses translational potential and may represent a novel direction for future oncological therapies [104].

Overall, mtDNA mutations hold particular relevance in cancer, as they are not restricted to the intracellular environment. Through mitochondrial transfer, they can disseminate across cells, exerting biological effects beyond their site of origin and shaping tumor progression at a collective level. Because of this unique biological property, mtDNA mutations also hold considerable promise for diagnostic, prognostic, and therapeutic applications. Future efforts may prioritize leveraging large-scale mutation datasets and AI-driven analytics to uncover clinically relevant patterns, integrating mtDNA profiling with multi-omics approaches for a systems-level view of tumor biology, and establishing standardized frameworks for detection and interpretation to accelerate their translation into oncology practice.

Abbreviations

ACC

Acetyl-CoA Carboxylase

ACLY

ATP-Citrate Lyase

ADP

Adenosine Diphosphate

AKT

Protein Kinase B

AMPK

AMP-Activated Protein Kinase

ATP

Adenosine Triphosphate

ATF4 / ATF5

Activating Transcription Factor 4 / 5

CAF

Cancer-Associated Fibroblast

cGAS

Cyclic GMP–AMP Synthase

cGAMP

Cyclic GMP–AMP

CHOP

C/EBP Homologous Protein

CNV

Copy Number Variation

CO1–CO3

Cytochrome c Oxidase Subunit I–III

CPT1

Carnitine Palmitoyltransferase 1

CTC

Circulating Tumor Cell

CYB

Cytochrome b

DAMPs

Damage-Associated Molecular Patterns

ddPCR

Droplet Digital PCR

DHODH

Dihydroorotate Dehydrogenase

D-loop

Displacement Loop

ETC

Electron Transport Chain

EMT

Epithelial–Mesenchymal Transition

EV

Extracellular Vesicle

FAO

Fatty Acid Oxidation

FASN

Fatty Acid Synthase

GAP43

Growth-Associated Protein 43

GJC

Gap Junction Channel

GLUT1

Glucose Transporter 1

GPx

Glutathione Peroxidase

GSH

Glutathione

HIF-1α

Hypoxia-Inducible Factor 1-alpha

IL-6 / IL-2

Interleukin 6 / 2

ISR

Integrated Stress Response

JNK

c-Jun N-terminal Kinase

LDHA

Lactate Dehydrogenase A

MAPK

Mitogen-Activated Protein Kinase

MFN2

Mitofusin 2

MPC

Mitochondrial Pyruvate Carrier

MSC

Mesenchymal Stem Cell

mtDNA

Mitochondrial DNA

M-Sec

Tumor Necrosis Factor Alpha-Induced Protein 2

mTOR

Mechanistic Target of Rapamycin

ND1–ND6

NADH Dehydrogenase Subunits 1–6

NF-κB

Nuclear Factor Kappa-Light-Chain-Enhancer of Activated B Cells

NRF2

Nuclear Factor Erythroid 2–Related Factor 2

NUMT

Nuclear Mitochondrial DNA Segment

OPA1

Optic Atrophy 1

OXPHOS

Oxidative Phosphorylation

PD-L1

Programmed Death-Ligand 1

PI3K

Phosphoinositide 3-Kinase

ROS

Reactive Oxygen Species

rRNA

Ribosomal RNA

SOD2

Superoxide Dismutase 2

SNV

Single Nucleotide Variant

STING

Stimulator of Interferon Genes

SREBP1

Sterol Regulatory Element–Binding Protein 1

TBK1

TANK-Binding Kinase 1

TCA

Tricarboxylic Acid Cycle

TCGA

The Cancer Genome Atlas

TIL

Tumor-Infiltrating Lymphocyte

TNT

Tunneling Nanotube

TPP

Triphenylphosphonium

tRNA

Transfer RNA

USP30

Ubiquitin-Specific Protease 30

VEGF

Vascular Endothelial Growth Factor

WGS

Whole Genome Sequencing

Author contributions

YJC, HZS and YCZ collected the related studies and drafted the manuscript. MMX, HQP, XNY, WW and JX participated in the design of the review. XJY and SS initiated the study and revised the manuscript. JY provided constructive comments during the revision of the manuscript. All authors have read and agreed on the published version of the manuscript.

Funding

This project was financially supported by the National Natural Science Foundation of China (92374102, 82473391, 82503859), the Science and Technology Commission of Shanghai Municipality (YDZX20243100002003, 23ZR1479300, 25ZR1402085), the Shuguang Program of Shanghai Municipal Education Commission and Shanghai Education Development Foundation (25SG10) and Noncommunicable Chronic Diseases-National Science and Technology Major Project (2024ZD0525500/2024ZD0525501/2024ZD0525505).

Data availability

No datasets were generated or analysed during the current study.

Declarations

Ethics approval and consent to participate

Not applicable.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.