Mitochondria-Derived Vesicles and Mitochondrial Extracellular Vesicles in Health and Cardiovascular Disease

Abstract

Mitochondria-derived vesicles (MDVs) and mitochondrial extracellular vesicles (mitoEVs) represent two related extensions of mitochondrial dynamics that link organelle maintenance to communication within and between cells. MDVs are small vesicles that bud directly from mitochondria, selectively packaging components of the outer membrane, inner membrane, or matrix. They serve as a localized quality-control mechanism that removes oxidized or damaged material without engaging the entire mitophagic machinery. After budding, MDVs typically enter the endolysosomal pathway, where they can fuse with late endosomes or lysosomes for cargo degradation. A subset of MDVs also targets other organelles, particularly peroxisomes, contributing to organelle crosstalk, lipid metabolism, and redox balance. By contrast, mitoEVs released into extracellular space contain intact functional mitochondria, mitochondrial contents (proteins, DNAs/RNAs, lipids, etc.), and non-mitochondrial cargo (i.e., mRNAs, non-coding RNAs, etc.), which can be transferred to recipient cells and subsequently, inducing either pathogenic or beneficial outcomes. Therefore, mitoEVs have been implicated in metabolic cooperation, immune regulation, tissue remodeling, and ageing. Accordingly, this review summarizes recent progress on the diverse mechanisms for the biogenesis of MDVs and mitoEVs as well as available protocols for their isolation. The roles of MDVs and mitoEVs in mediating mitochondrial quality/quantity control and multiple layers of crosstalk between intracellular organelles and different cell types in health and diseases are highlighted. Lastly, mitoEV-mediated pathogenic effects and therapeutic potential in cardiovascular disease are also discussed.

1. Introduction

Mitochondria are double-membrane organelles consisting of an outer mitochondrial membrane (OMM), a folded inner mitochondrial membrane (IMM), an intermembrane space, and an enclosed matrix. The OMM contains proteins such as TOM20, VDAC1, and MFN2, which mediate metabolite exchange, membrane fusion, and communication with the cytosol^1,2. The IMM, organized into cristae, is rich in OPA1, TIM23, and the complexes of the electron transport chain that drive oxidative phosphorylation. The matrix contains the enzymes of the tricarboxylic acid (TCA) cycle, including pyruvate dehydrogenase (PDH), as well as mitochondrial DNA³. Together, these features define key structural and functional markers of mitochondrial integrity. This architecture enables mitochondria to integrate metabolic activity with signaling and quality-control processes that support cellular homeostasis.

The most recognized role of mitochondria is the generation of ATP through oxidative phosphorylation, a function that is especially critical in energy-demanding tissues such as cardiomyocytes and skeletal muscle. Beyond bioenergetics, mitochondria are central hubs for cellular homeostasis. For example, they regulate redox balance by controlling reactive oxygen species (ROS) levels, participate in calcium buffering, coordinate lipid metabolism, and govern cell survival pathways through the initiation of apoptosis^4–6. Mitochondria also influence cell growth, proliferation, and detoxification processes, highlighting their multifaceted contributions to cellular physiology⁷. Because of these diverse functions, mitochondria are particularly important in cardiovascular health. Indeed, the heart relies on a continuous supply of ATP to sustain contractile activity, and even subtle impairments in mitochondrial efficiency can compromise myocardial performance. To preserve their functionality, mitochondria are tightly integrated with cellular signaling networks. Their plasticity, achieved through continuous fusion and fission as well as contacts with other organelles, allows the exchange of metabolites and the coordination of stress responses^{8, 9}. Furthermore, multiple mitochondrial quality control (mitoQC) pathways have evolved to ensure proper mitochondrial signaling and function, linking organelle integrity to overall cellular health. Selectively activation of a specific mitoQC pathway is determined by the degree of injury and by the particular mitochondrial protein or lipid cargo that is damaged. As for whole mitochondria damaged irreversibly, it would be cleared by the ubiquitin-dependent mitophagy that involves the serine/threonine kinase PINK1 and the E3 ubiquitin ligase Parkin. Recently, mitochondria-derived vesicles (MDVs) have been identified as a novel mechanism in mitoQC¹⁰. Under physiological conditions, MDVs could be produced for trafficking of cargos important for cellular health such as ion, protein and metabolites and at the same time recycling of damaged mitochondrial materials from lysosome¹¹. In addition, MDVs can be dramatically generated in response to various stressors such as ROS, hypoxia, heat and lipopolysaccharide (LPS), and thereby regulating cellular homeostasis¹⁰. Accordingly, MDVs have been implicated in diverse physiological processes and various human diseases such as metabolic disorders and cardiovascular diseases.

Extracellular vesicles (EVs) are nanoscale, cargo-bearing structures actively secreted by cells into the extracellular space and mediate intercellular or inter-organ communications through transferring proteins, lipids, and nucleic acids between cells. The concept of EVs dates back to the 1940s, when “minute breakdown products” of blood cells were first described and later termed “platelet dust”^{12, 13}. By the 1970s–1980s, vesicle-like structures were identified in diverse tissues, establishing the foundations of EV biology^14–17. Today, EVs are firmly recognized as central regulators of autocrine and paracrine signaling, with a broad range of roles spanning immunity, tissue repair, cancer progression, and cardiovascular physiology^18–20. In recent years, the functional repertoire of EVs has expanded beyond molecular cargo to include the transfer of organelles. Among these, mitochondria have gained particular attention due to their pivotal roles in energy production, redox balance, calcium buffering, and apoptosis^{21, 22}. Indeed, accumulating evidence has shown that mitochondrial components (e.g, DNA, RNA, protein, lipids), MDVs, damaged or intact mitochondria can be selectively loaded into EVs, termed mitochondrial EVs (mitoEVs), which provide a new layer of mechanisms for metabolic support and stress adaptation in recipient cells^23–25. Therefore, mitoEVs have been represented as emerging dimension of intercellular and inter-organ communications.

In this review, we highlight recent knowledge of MDVs and mitoEVs on interconnected outputs of mitochondrial remodeling and quality control, rather than on subsets of extracellular vesicles in general. In particular, we summarize cargo-selective biogenesis pathways and key regulatory nodes that govern mitochondrial vesicle production under physiological and stress conditions, map MDV trafficking routes that mediate inter-organelle communication, and link these intracellular processes to extracellular mitochondrial export through distinct mitoEV subtypes. We further outline accessible protocols for the isolation and characterization of mitoEVs, including specialized subcategories such as exophers that are applicable across tissues. Finally, we discuss emerging evidence that implicates MDVs and mitoEVs in cardiovascular disease and aging with integrating findings from both animal models and human studies, and that explore critical knowledge gaps and opportunities for targeted therapeutic strategies leveraging mitoEV biology.

2. Biogenesis of mitochondria-derived vesicles (MDVs)

MDVs were first observed by Neuspiel et al.²⁶ in HeLa cells where outer-membrane MAPL (mitochondria-anchored protein ligase) gene was overexpressed. They identified that such MDVs selectively incorporate their cargo, evidenced by that vesicles containing MAPL exclude another outer-membrane marker, TOM20, and vesicles containing TOM20 exclude MAPL. Interestingly, MDVs with different cargo have distinctive intracellular fate, that is MAPL-positive MDVs, but not TOM20-containing MDVs, fusing with a subset of peroxisomes. Later-on studies further indicate that MDVs can be either single-membrane vesicles or double-membrane vesicles, with single-membrane MDVs via budding off OMM or IMM, and double-membrane MDVs forming from both OMM and IMM. The type of MDVs formed depends on the specific mitochondrial components being targeted for removal or transport. These initial studies defined MDVs as having three criteria including: 1) cargo-selective; 2) small single or double-membraned structures with an average diameter of 60–150 nm; and 3) generation is independent of the autophagy and mitochondrial division machinery^{26, 27}. Currently, it is well recognized that the formation of MDVs, originated from OMM, IMM, or both mitochondrial membranes, acts as a basal mitochondrial housekeeping mechanism and a first-line defense against stress conditions. The following section highlights different mechanisms and mediators that regulate MDV biogenesis (Fig. 1A–D).

Figure 1: Mechanisms of Mitochondria-derived vesicle (MDV) biogenesis under physiological and stress conditions.

(A) Under basal conditions, TOM20+ MDVs bud from the outer mitochondrial membrane via MIRO1/2- and Drp1-dependent scission, selectively packaging OMM, IMM, and IMS proteins such as TOM20, TIMM23, CYCS, and COX1 to maintain mitochondrial homeostasis. (B) During oxidative stress, PINK1/Parkin promote TOM20−/PDH+ MDV formation through SNX9 and OPA1, whereas under inflammation, PINK1/Parkin accumulation suppresses MDV generation by inhibiting RAB9 and SNX9 recruitment. (C) Inner membrane–derived vesicles VDIMs form via VDAC1 pores and ESCRT-dependent scission in a TRPML1/Ca²+-dependent manner, enabling lysosomal degradation of damaged IMM and mtDNA components. (D) β-hydroxybutyrate promotes the generation of TOM20−/PDH+ MDVs through Snx9 β-hydroxybutyrate, whereas fumarate induces SNX9-dependent, mitoDNA-containing MDVs, highlighting a link between cellular metabolic state and mitochondrial quality control. CYCS, cytochrome c; COX, cytochrome c oxidase subunit; TIMM23, translocase of inner mitochondrial membrane 23; TOM20, translocase of the outer mitochondrial membrane 20; Miro1/2, mitochondrial Rho 1/2; MID, mitochondrial dynamics proteins; MFF, mitochondrial fission factor; DRP1, dynamin related protein 1; PINK1, PTEN-induced kinase 1; RAB9, member RAS oncogene family 9; SNX9, sorting nexin 9; OPA1, optic atrophy 1; ROS, reactive oxygen species; ESCRT, endosomal sorting complex required for transport; TRPML1, transient receptor potential mucolipin 1; VDAC, voltage-dependent anion-selective channel; PDH, pyruvate dehydrogenase; and STOML2, stomatin-like protein 2. Modified from König et al.²⁸, McLelland et al.³³, Prashar et al.³⁸ and Tang et al.²⁹

2.1. MIROs/Drp1-dependent biogenesis of outer membrane-originated MDVs

Regarding the biogenesis of MDVs in mammalian cells, König et al.²⁸ performed a complete proteomic and lipidomic analysis of steady-state TOM20⁺-MDVs and established a commonly mechanistic model. It reveals that small vesicles budding from the OMM is initiated by the microtubule-associated motor proteins MIRO1 and MIRO2 (MIRO1/2), which drive microtubule pulling of thin OMM protrusions (Fig. 1A). At these sites, the mitochondrial dynamin related protein 1 (Drp1) is subsequently recruited by its redundant receptors (MID49, MID51, MFF), where they assemble into small foci close to the tip of thin OMM protrusions and then catalyze the scission of thin membrane tubules to pinch off TOM20⁺-MDVs with a diameter of approximately 160 nm, measured by super-resolution microscopy. More than 50% of the resulting TOM20⁺-MDVs are enriched with the OMM proteins (i.e., TOM20/40/70), the intermembrane space proteins (i.e., CYCS), the IMM proteins (i.e., TIMM23), and mitochondrial DNA (mtDNA)-encoded COX1, suggesting that these MDVs have a double-membrane structure. Nonetheless, some OMM proteins (i.e., MAPL) and matrix enzymes (i.e., PDH) are not encased in these MDVs, underscoring their selective packaging²⁹. Additional analysis of PDH⁺-MDVs also reveals that the biogenesis of MDVs relies on preceding MIRO1/2-dependent mitochondrial-membrane protrusions pulled along microtubule filaments, followed by a Drp1-dependent scission event to form MDVs, suggesting a global role for such GTPase-driven outer membrane-originated MDV biogenesis²⁹.

It is important to note here that the generation of TOM20⁺-MDVs at steady state appears PINK1/Parkin-independent, as neither PINK1 nor Parkin are identified in the TOM20⁺-MDV proteome^{30, 31}. This indicates that PINK1/Parkin are not essential for steady-state MDV formation. However, a couple of studies have shown that loss of PINK1 or Parkin results in an accumulation of TOM20⁺-positive MDVs, due to increased MDV flux in PINK1-deficient cells or impaired trafficking to late endosomes/lysosomes in Parkin-null cells; albeit both are dispensable for the steady-state maintenance of the mitochondrial network^{30, 31}. Nonetheless, upon stress conditions, both PINK1 and Parkin act as positive regulators for oxidative stress-induced MDVs and negative regulators for inflammation-induced MDVs, as discussed below.

2.2. PINK1/Parkin-regulated biogenesis of MDVs upon stress and inflammatory conditions

Both PINK1 and Parkin are well characterized to regulate the degradation of depolarized mitochondria via autophagy (termed mitophagy)³². However, McLelland et al.³³ observed that, under conditions of mitochondrial oxidative stress, PINK1/Parkin are required for the generation of double-membraned TOM20-negative MDVs in which Snx9 and OPA1 regulate selectively uploading oxidized/damaged inner membrane and matrix proteins (Fig. 1B). Snx9, a sorting nexin involved in endosomal trafficking, is recruited to mitochondria to initiate matrix-containing vesicle budding, meanwhile OPA1, a dynamin-like GTPase located in the inner membrane, shapes cristae and supports vesicle formation. Silencing either Snx9 or OPA1 selectively reduces TOM20⁻PDH+ MDVs, without affecting TOM20+-vesicle production²⁷. More interestingly, PINK1/Parkin-mediated MDV generation and turnover occur at an early stage (over a period of 1–4 h after stress), contrasting with PINK1/Parkin-mediated mitochondrial turnover via mitophagy which occurs over a much larger scale of time (over a period of days post-stress)^{33, 34}. Currently, it remains unclear how PINK1/Parkin regulate such TOM20-negative MDV biogenesis upon oxidative stress at the early stage. It is plausible that ROS-oxidative stress may cause PINK1 accumulation at the OMM and promote the subsequent Parkin recruitment and ubiquitin ligase activity, leading to OMM curvature and pinch-off.

Unlike the above ROS-induced MDV biogenesis that is positively regulated by PINK1/Parkin, in the context of heat stress (HS) or inflammatory conditions, mitochondrial antigen presentation (MitoAP) that relies on the generation and trafficking of TOM20-negative MDVs in macrophages and dendritic cells by a vacuolar pathway distinct from mitophagy. Surprisingly, such HS/inflammation-triggered MDV biogenesis is inhibited by PINK1 and Parkin³⁵. Specifically, Matheoud et al.³⁵ demonstrated that in the absence of either PINK1 or Parkin, heat stress and inflammatory conditions could stimulate macrophages/dendritic cells to generate small double membrane-bound TOM20-negative MDVs (80–120 nm in diameter) for presenting mitoAP. They further identified that formation of such mitochondrial antigen-containing MDVs requires Sorting nexin 9 (Snx9) and Rab9, both of which recruitment to mitochondria are actively blocked by PINK1 and Parkin. Meanwhile, Rab7 is required for the fusion of MDVs with late endosomes to load MitoAP on the surface of cells. The possible mechanism underlying PINK1/Parkin-mediated inhibition of MDV biogenesis could be ascribed to proteasome-dependent degradation of Snx9. In the presence of PINK1/Parkin, heat stress and inflammation activate PINK1/Parkin-associated ubiquitination of Snx9 for degradation, preventing its recruitment to mitochondria and thereby, suppressing TOM20-negative MDV formation. Future studies would be warranted to define the exact role played by PINK1 and Parkin in these MDV biogenesis and cargo selection.

2.3. Biogenesis of MDVs originated from inner mitochondrial membrane (IMM)

Recently, emerging evidence has indicated that each individual crista, formed by the invagination of the IMM, is an independent and highly dynamic anatomic unit with its own membrane potential independently of adjacent cristae^{36, 37}. Nonetheless, an individual crista could also undergo damage under steady state and stress conditions. While IMM is sub-compartmentalized into the inner boundary membrane (IBM) that prevents the spread of damage from one crista to the whole mitochondria, such a damaged crista mixed with healthy one would still cause mitochondrial dysfunction. Hence, this local crista injury has to be removed to maintain mitochondrial homeostasis. Along this line, Prashar et al.³⁸ recently identified a new mechanism to generate MDVs directly from the IMM, termed VDIMs (vesicles derived from the IMM). Utilizing super-resolution microscopy and different mitochondrial compartment-selective dyes in immortalized mouse embryonic fibroblasts (MEFs) at steady state, they observed that the biogenesis of VDIMs is initiated by herniation of the IMM through VDAC1-generated pores in the outer mitochondrial membrane (Fig. 1C). Subsequently, lysosomes in proximity capture these protrusions, with the final scission step mediated by the ESCRT complex in a microautophagy-like manner and generates much larger VDIMs with an average diameter of around 500 nm (Fig. 1C). VDIM formation is strongly stimulated under oxidative stress and critically depends on TRPML1-driven calcium release from lysosomes, but does not require either Drp1 or MIRO1 that are crucial for OMM-originated MDV formation³⁸. Importantly, these VDIMs can be generated at resting state in all cell types tested and selectively incorporate IMM components and damaged mtDNA excluding outer membrane and matrix cargo, thereby representing a unique and highly specialized mitoQC mechanism.

2.4. Metabolites/Snx9 promote formation of IMM/matrix-MDVs

Metabolites are necessary energy sources for maintaining cellular homeostasis and can function as crucial modulators of protein post-translational modification (PTM). Recently, Tang et al.²⁹ reported that β-hydroxybutyrate (BHB), one of the ketone bodies predominantly produced through fatty acid oxidation in hepatic mitochondria, could induce modification of Snx9 protein by the covalent attachment of lysine β-hydroxybutyrylation (Kbhb) in hepatocellular carcinoma cells (Fig. 1D). Given that Snx9 is responsible for the biogenesis of inflammation-induced MDVs (discussed above); therefore, Tang et al.²⁹ tested the effects of Kbhb-Snx9 on MDV formation and interestingly found that Kbhb-modified Snx9 significantly augments interaction with at least 29 IMM/Matrix-resident proteins (i.e., OPA1, STOML2, TIMM44, MRPL17/20/24/37/47) and thereby, enhancing the formation of IMM/matrix-MDVs (TOM20⁻/PDH⁺) in hepatic cells upon treatment with mitochondrial toxins. However, the study by Tang et al.²⁹ raises several points that warrant further investigation.. For example, they observed that Kbhb-Snx9 interacts with VDAC2, an OMM-resident protein that forms a β-barrel pore, allowing for the transport of ions and metabolites across the membrane. Hence, it would be interesting to explore whether such VDAC2-pores contribute to the biogenesis of BHB-induced IMM/matrix-MDVs, similar to that VDAC1-pores are required for the generation of VDIMs (discussed above).

As for whether other metabolites also affect MDV biogenesis, a recent study by Zecchini et al.³⁹ showed that treatment of mouse kidney epithelial cells with monomethyl fumarate (MMF), one of the TCA cycle metabolites, promotes formation of TOM20⁻/PDH⁺-MDVs with double-membrane bound (average diameter of 450 nm). Such fumarate-induced MDVs contain mtDNAs and also requires the presence of Snx9 but not Rab9 for their release from mitochondria (Fig. 1D). However, the mechanism underlying fumarate-mediated MDV formation remains unclear. Furthermore, along with BHB and fumarate discussed above, many metabolites are known to modulate mitoQC processes to ensure efficient energy production and prevent the accumulation of harmful metabolites; therefore, future studies would be needed to clarify whether and how these metabolites regulate mitoQC in an MDV-dependent manner.

2.5. Other factors/genes that regulate MDV biogenesis

Additionally, there are other factors/genes that have been identified to affect MDV formation so far. For example, a higher abundance of MDVs is produced in cardiac fibroblast H9c2 cells upon glucose deprivation, compared with cells under normal growth conditions⁴⁰. Guo et al.⁴¹ recently showed that short exposure of oligodendrocyte precursor cells to a low concentration of carbon monoxide (CO) yields massive generation of MDVs without interfering with mitochondrial morphology and membrane potential. This could be interpreted by the fact that CO preconditioning could induce mild oxidative stress and thereby activating MDV biogenesis. What’s more, Zhao et al.⁴² recently demonstrated that the shedding of MDVs from mitochondria in pulmonary ECs requires LKB1-mediated mitochondrial recruitment of Rab9 GTPase. They observed that genetic ablation of LKB1 in ECs impaired MDV formation and accelerated mitochondrial fragmentation both in vitro and in vivo, which could be restored by overexpression of LKB1. Importantly, emerging evidence indicates that MDV biogenesis is also influenced by biological sex. In the heart, for example, MDV abundance exhibits sex-dependent differences across the daily cycle, with higher MDV numbers observed in males during the active (dark) phase and reduced MDV formation in females during the same period, suggesting intrinsic sex-specific regulation of mitochondrial quality-control pathways⁴³. However, another study by Liu et al.⁴⁴ quantifying circulating cell-specific MDVs in neurodegenerative disease cohorts did not detect significant sex-associated differences, suggesting that sex-dependent effects on MDV biogenesis and downstream vesicle release may be context dependent and influenced by tissue origin, disease state, and the specific MDV-derived vesicle population examined⁴⁴. Considering their small size, heterogenous and highly dynamic nature, as well as current technology limitations (super-resolution microscopy, appropriate labeling of MDVs and quantification challenge), we believe that some other mechanisms/mediators that regulate MDV biogenesis and its related sex difference may be still undiscovered.

3. Transport pathways of MDVs and their inter-organelle communication

MDVs transport their selective cargo (i.e., proteins, lipids, DNAs and RNAs) from mitochondria to other organelles, primarily late endosomes/lysosomes, but also peroxisomes, phagosomes, multivesicular bodies (MVBs), and EVs. These different transport pathways comprise multiple layers of MDV-mediated mitoQC to remove damaged components and facilitate inter-organelle communication for maintaining cellular homeostasis.

3.1. MDVs targeting lysosomes

A major traffic pathway for MDVs is directed toward lysosomes, where these vesicles mediate selective degradation of oxidized components which enables localized turnover. However, the process of MDV entrance into lysosomes is complicated and involves different mediators or protein complexes. For example, one prior study McLelland et al.³⁴ revealed that MDV-lysosome fusion is heterotypic and requires soluble N-ethylmaleimide-sensitive factor attachment protein receptor (SNARE). They showed that Syntaxin-17 (Stx17), a subset of the SNARE complex, detects mitochondrial outer membrane curvature via a unique clamping structure and facilitates its fusion to lysosomes. Recently, Hao et al.⁴⁵ reported that Stx17, SNAP29, and VAMP7 form a ternary SNARE complex crucial for PINK1/Parkin-dependent MDVs targeting lysosomes (Fig. 2A). In addition, recent study by Hayashi et al.⁴⁶ has demonstrated that Tollip plays a critical role in MDV transport to lysosomes and acts as an adaptor molecule to promote lysosomal degradation of damaged ER membrane proteins. Ryan et al.³⁰ also showed that Tollip captures MDVs and recruits cytoplasmic Parkin to form a sophisticated complex, with Rab7 adhering to mature MDVs, thereby directing them towards lysosomal degradation. What’s more, Zhou et al.⁴⁷ reported that VPS35, dissociates from the retromer complex, recognizes TOM20⁺-MDVs and facilitates them to bind with DRP1 oligomers, which drives MDVs transporting from the mitochondrial surface to the lysosome for breakdown. Collectively, these studies suggest that different MDVs carrying with selective cargo may employ distinctive mediators to fuse them with lysosomes.

Figure 2: Inter-organelle communication of MDVs.

(A) PINK1/Parkin-dependent MDVs are directed to lysosomes for degradation via the SNARE complex (STX17–SNAP29–VAMP7), while Tollip captures MDVs, recruits Parkin, and interacts with Rab7 to promote lysosomal fusion. (B) MAPL+ MDVs mediate communication with peroxisomes. The retromer complex (VPS35, VPS26) drives the formation of MAPL+ MDVs while excluding TOM20+ vesicles, enabling targeted delivery to peroxisomes. (C) In de novo peroxisome biogenesis, MAPL+ MDVs transfer PEX3 and PEX14 from mitochondria to ER-derived PEX16+ vesicles, forming pre-peroxisomes that mature into functional peroxisomes. This budding process is orchestrated by the mitochondrial E3 ligase MARCH5, independently of DRP1. PINK1, PTEN-induced kinase 1; STX17, syntaxin 17; VAMP7, vesicle-associated membrane protein 7; SNAP29, synaptosome-associated protein 29; RAB7, member RAS oncogene family 7; TOLLIP, toll-interacting protein; MAPL, mitochondria-associated protein ligase; VPS, vacuolar protein sorting-associated protein; PEX, peroxisomal biogenesis factor; and MARCH5, membrane associated ring-CH-type finger 5. Modified from Hao et al.⁴⁵, Braschi et al.⁴⁸ and Zheng et al.⁵¹

3.2. MDVs communicate with peroxisomes

One of the first insights into MDV traffic to peroxisomes came from work by Neuspiel et al.²⁶, who observed that MAPL⁺-MDVs were targeted to peroxisomes, whereas TOM20⁺-MDVs did not share this fate. Later-on study by Braschi et al.⁴⁸ showed that components of the retromer complex, such as Vps35 and Vps26, are recruited to budding sites on mitochondria, and helps sculpt MDVs with selectively carrying MAPL (a mitochondrial ubiquitin ligase) and excluding others like TOM20 (Fig. 2B). Indeed, the retromer complex of VPS35 and VPS26 is essential for targeting MAPL⁺ MDVs to the peroxisome, as evidenced by the fact that silencing either VPS35 or VPS26 reduces the co-localization of MAPL⁺ MDVs with peroxisomes⁴⁹. Nonetheless, further studies are warranted to fully understand the specific mechanisms involved in the targeting of MDVs to peroxisomes for maintain cellular homeostasis in steady state and stress/disease conditions.

Along this line, recent studies have indicated that MAPL⁺ MDVs can promote the biogenesis of peroxisomes, as MAPL is an E3 ubiquitin ligase that has a dual role in the regulation of mitochondrial morphology and peroxisome elongation⁴⁸. Furthermore, PEX3 and PEX14, peroxisomal membrane proteins that are essential for peroxisome formation⁵⁰, have been found to target mitochondria and subsequently are packed into MDVs, which then fuse with the vesicles derived from the ER containing PEX16, resulting in peroxisomal precursor structures and de novo generation of peroxisomes⁵⁰. In addition, Zheng et al.⁵¹ identified MARCH5, a RING-type ubiquitin ligase anchored in the mitochondrial outer membrane, as a central regulator of de novo peroxisome biogenesis (Fig. 2C). In peroxisome-deficient human cell lines, MARCH5 drives the budding of vesicles containing PEX3 from mitochondria which mature into pre-peroxisomes. Interestingly, this budding process does not rely on classical mitochondrial fission machinery such as DRP1 or other dynamin-related GTPases, which usually sever mitochondrial membranes. Instead, MARCH5 uses its ubiquitin ligase activity to orchestrate cargo selection and MDV scission. Loss of MARCH5 abolishes this trafficking pathway, highlighting its essential role in maintaining peroxisome homeostasis.

3.3. Other MDV trafficking pathways

In a different context, such as bacterial infection, mitochondria can produce MDVs packed with antimicrobial components and converge to the bacteria-containing phagosomes⁹. For instance, Abuaita et al.⁵² recently observed that the infection of macrophages with Staphylococcus aureus stimulated the generation of MDVs which selectively encased SOD2. These MDVs travel to and fuse with bacteria-containing phagosomes where SOD2 converts superoxide anions into hydrogen peroxide (H₂O₂), which kills the invading bacteria. This process mediates the intracellular communication between phagosomes and mitochondria, allowing cells to make use of MDVs for antimicrobial defense.

When the lysosomal degradation is exceeded upon stress/disease conditions, MDVs are fused with MVBs to generate exosomes or packed into microvesicles for EV discharge, as discussed below. Altogether, these findings posit MDVs as distinctive mediators of intra-cellular communication. Their dynamic interactions with lysosomes, peroxisomes, phagosomes, and endosomal compartments confer remarkable cellular plasticity, enabling rapid adaptation to external stimuli.

4. Generation of mitoEVs for the control of mitochondrial quality and quantity

While the formation of MDV is crucial for mitoQC and organelle communication, recent work has shown that cells are also capable of exporting whole mitochondria and mitochondrial components including MDVs through EVs under steady state and stress/disease conditions. Such EVs carrying intact mitochondria and their derived components are usually called mitoEVs to distinguish them from other types of EVs. Initially, Spees et al.⁵³ observed that mitochondria-containing EVs released from human mesenchymal stem cells (hMSCs) or fibroblasts could be taken up by mtDNA-deficient A549 ρ cells, rescuing their respiratory capacity and restoring ATP production, oxygen consumption, and cell growth. After that, several studies have confirmed the mitoEVs secretion from astrocytes⁵⁴, activated monocytes⁵⁵, human brain endothelial cells⁵⁶, and cardiomyocytes⁵⁷. At present, it is well recognized that mitochondria (either whole or partial components) can be shuttled between cells and organs via tunnelling nanotubes (TNTs)⁵⁸, gap junction channels⁵⁹, and many subsets of mitoEVs including microvesicles (produced by directly budding from plasma membrane)⁶⁰, exosomes (formed by endosomes-multivesicular bodies and then fused with plasma membrane)^61–63, exophers (very large vesicles with a size range of 3500–4000 nm)⁶⁴, migrasomes (produced from specific migrating cells)⁶⁵, mitophers (a peculiar type of microvesicles containing a single healthy mitochondria)⁶⁶, and mitolysosomes (a specific lysosomes engulfed intact mitochondria that are then released by exocytosis)⁶⁷. The following focuses on the major mechanisms underlying the formation of mitoEVs and their roles in the control of mitochondrial quality and quantity (Fig. 3A–E and Table 1).

Figure 3: Microvesicle and Mitochondrial extracellular vesicle (mitoEV) release.

(A) Mitochondria and mitochondria-derived vesicles (MDVs) are released either by plasma-membrane budding or through the multivesicular-body (MVB) pathway. LC3+ microvesicles and CD38-regulated MDVs deliver mitochondrial components via exosomes or larger vesicles. (B) Under defective autophagy, autophagosomes enclosing damaged mitochondria are secreted instead of degraded; Ambra1–LC3 interaction promotes mitoEV formation during stress. (C) Exophers containing dysfunctional mitochondria bud from the plasma membrane and are subsequently cleared by macrophages. (D) During cell migration, damaged mitochondria transported by KIF5B and anchored by Myo19 integrate into migrasomes for quality control. (E) During spermatid maturation, excess mitochondria are expelled as mitophers through SPE-8- and actin-dependent budding to preserve mitochondrial balance. TSG101, tumor susceptibility gene 101; ARRDC1, arrestin domain containing protein 1; CD38, cluster of differentiation 38; ADPR, adenosine diphosphate-ribose; AMBRA1, autophagy and beclin 1 regulator 1; KLF5B, Krüppel-like factor 5b; MYO19, myosin XIX; SPE-8, Spermatocyte protein-8; PETP, polyethylene terephthalate; MDT-15, mediator of RNA polymerase II transcription subunit 15; SEP-1, selenium-binding protein 1; SREBP1, sterol regulatory element-binding protein 1; FGF, fibroblast growth factor; RAS, rat sarcoma; MAPK, mitogen-activated protein kinase; DAF16, DAuer Formation 16; and FOXO, forkhead box O. Modified from Nabhan et al⁶⁹, Zhang et al⁷⁴, Cooper et al.⁷⁸, Sheng et al.⁸⁴ and Liu et al.⁸⁵

Table 1.

Characterization of mitochondrial-derived vesicles (MDVs) and mitochondrial extracellular vesicles (mitoEVs)

Vesicle Type	Size	Origin/Biogenesis	Cargo	Condition
MDVs	60nm-500nm	Derived from outer mitochondrial membrane (OMM), inner membrane (VDIMs), or both	Mitochondrial proteins, lipids, RNA, mtDNA,	Basal/stress
Mito/Microvesicles	150nm-1000nm	Outward budding of plasma membrane	Mitochondria, MDVs	Basal/stress
Mito/Exosomes	<150nm	Endosomal-MVB pathway	MDVs	Basal/stress
Mito/Secretory Autophagosomes	100nm-500nm	LC3-mediated autophagosome formation that escapes lysosomal degradation and is secreted from the cell	Dysfunctional mitochondria	Stress
Exophers	1μm-10μm	Large vesicles formed by plasma membrane pinching and engulfed by recipient cells for degradation	Injured mitochondria	Stress
Mitolysosomes	Size-variable	Lysosome engulfment and SNARE-dependent exocytosis	Injured mitochondria	Stress
Migrasomes	50nm-100nm	Cell periphery–derived vesicles released during migration	Fragmented, damaged mitochondria	Stress
Mitophers	Up to 1μm	Actin-dependent plasma membrane budding	Healthy mitochondria	Sperm development

4.1. MitoEVs produced via microvesicles and exosomes

One proposed mechanism involves directly outward budding of the plasma membrane that encapsulates mitochondria. In this context, Phinney et al.⁶⁸ reported that mesenchymal stem cells (MSCs) secrete polarized mitochondria by packaging them into LC3⁺-microvesicles formed via directly outward budding of the plasma membrane) (Fig. 3A). These microvesicles contain intact and functional mitochondria, which can be released through arrestin domain-containing protein 1 (ARRDC1), an accessory protein localized at the plasma membrane and subsequently, such microvesicles are transferred to recipient cells (i.e., macrophages) and thereby enhance their bioenergetic function^{68, 69} (Fig. 3A). In another study by Crewe et al.⁷⁰, adipocytes were shown to release damaged mitochondria and their components through small EVs (exosomes) generated through the endosomal-MVB pathway (Fig. 3A). Importantly, only small size MDVs (30–150 nm) could be packaged into exosomes, while larger MDVs may be released through microvesicles⁷⁰. Recently, Suh et al.⁷¹ showed that the CD38/cADPR (cyclic ADP ribose) signaling regulates the secretion of mitochondria and MDVs (<1000 nm) from mature osteoblasts via microvesicles, a process tightly coupled with mitochondrial fragmentation and donut formation (Fig. 3A). In this context, proteins such as Fis1 (fission machinery), Opa1/Mfn1/2 (fusion machinery), and autophagy regulators (e.g., LC3, BNIP3/3L, and LAMP1) act as modulators of mitochondrial morphology and vesicle loading⁷¹. Interestingly, Opa1 knockdown or Fis1 overexpression enhanced mitoEV secretion, whereas treatments with promoting mitochondrial fusion reduced it⁷¹. These results suggest that dynamic remodeling of mitochondria dictates the formation of mitoEVs and modulated by CD38/cADPR and mitophagy-related proteins.

4.2. mitoEVs generated via secretory autophagosomes

Increased evidence has suggested that mitoEV secretion and autophagic flux are coordinated processes that maintain cellular homeostasis⁷². Usually, autophagy helps to remove damaged or dysfunctional mitochondria through the formation of autophagosomes. During this process, ATG8 proteins (i.e. LC3 in mammals), are conjugated to autophagosome membranes to help recruit and engulf damaged mitochondria⁷². These newly formed autophagosomes then fuse with lysosomes where the mitochondria are degraded. However, when such autophagic flux is blocked (i.e, impaired fusion, lysosomal dysfunction), autophagosomes can escape from degradation and be secreted out of cells as mitoEVs (also called secretory autophagy)⁷³. Recently, Zhang et al.⁷⁴ provided new evidence showing that upon cardiac ischemia/reperfusion (I/R), autophagosomes were accumulated in cardiomyocytes, resulting in generation and secretion of mitoEVs. They further identified that Ambra1 (autophagy/beclin-1 regulator 1, localized at the outer mitochondrial membrane) is required for such mitoEV formation in I/R-cardiomyocytes through binding tightly to LC3, an autophagosome surface molecule (Fig. 3B). Consistently, Ambra1 is highly enriched on the surface of myocyte-derived mitoEVs and cardiac-specific down-regulation of Ambra1 inhibits the mitoEV release from I/R-injured cardiomyocytes.

4.3. Exophers and mitolysosomes for the disposal of damaged mitochondria

Along with mitoEVs generated though microvesicles/exosomes (Fig. 3A) and secretory autophagosomes (Fig. 3B), another mechanism for mitochondrial extrusion involves exophers, which are large, micron-scale vesicle-like structures (1–10 μm) first described in C. elegans neurons by Melentijevic et al.⁶⁴ (Fig. 3C). Exophers are released out of cells through the outward budding of the plasma membrane via a pinching-off mechanism⁶⁴. Recent studies from C. elegans have revealed that exopher formation is tightly regulated by mitochondrial dynamics⁷⁵, intermediate filaments⁷⁶, autophagy-related proteins (e.g., ATG-16.2)⁷⁷, and stress responses⁷⁸. For example, Wu et al.⁷⁵ found that EGL-1, the BH3-only protein in C. elegans, promotes exopher formation cell-autonomously by regulating mitochondrial fission-fusion dynamics through CED-9. Furthermore, various stresses such as fasting and oxidative stress have been demonstrated to activate exopher production through multiple cell-nonautonomous signaling pathways (i.e, PETP-1, MDT-15, SBP-1/SREPB1, FGF/RAS/MAPK, DAF16/FOXO)⁷⁸. Indeed, the formation of exophers in the nematode C. elegans is beneficial to support offspring development⁷⁹ and neuronal function⁶⁴, as exophers can mediate transfer of injured whole mitochondria from proteotoxically stressed neurons to neighboring cells for degradation. Notably, analogous process of exopher formation appears to be conserved across species. For instance, Nicolás-Ávila et al.⁵⁷ discovered that cardiomyocytes could eject exopher-like structure containing dysfunctional mitochondria, which were taken up by cardiac resident macrophages through the phagocytic receptor MerTK and thereby preventing the accumulation of damaged mitochondria in cardiomyocytes to maintain healthy hearts. Similarly, the generation of exophers in murine cardiomyocytes is also driven by autophagy machinery and enhanced during cardiac stress⁵⁷.

In addition to disposal of injured mitochondria by exophers, one recent study by Bao et al.⁶⁷ revealed that intracellular damaged mitochondria can be expelled through exocytosis following their whole engulfment by lysosomes in neurons and astrocytes upon prolonged treatment with flunarizine (FNZ), a piperazine derivative with anticonvulsant properties that acts as a calcium antagonist. This process is termed mitolysosome exocytosis which occurs via a mechanism depending on vesicle v-SNARE, VAMP2 and plasma membrane t-SNARE STX4⁸⁰. Such extracellular secretion of mitolysosomes as mitoEVs may be an alternative mitoQC pathway for mitochondrial clearance. Recently, Dayalan Naidu et al.⁸¹ showed that the depletion of Nrf2 in lung cancer cells results in accumulation of damaged mitochondria within mitolysosomes and impaired their exocytosis, whereas upregulation of Nrf2 facilitates mitolysosome processing thereby ensuring timely disposal of injured mitochondria. However, it remains elusive how to regulate mitolysosome biogenesis and whether they involve cell-cell communication, which warrants further investigation.

4.4. Migrasomes and mitocytosis for mitoQC

Mitocytosis represents a newly identified dimension of mitoQC process discovered by Ma et al.⁸². In this process, migrating cells such as macrophages and neutrophils eliminate damaged mitochondria by packaging them into specialized small vesicular structures termed migrasomes (Fig. 3D). These vesicles, with diameters ranging from 50 to 100 nm, form at the cell periphery and are left behind during cell migration, thereby facilitating the efficient removal of dysfunctional mitochondria from the cell. Such migrasome-mediated selectively removal of damaged mitochondria is called mitocytosis. Migrasomes, initially observed in migrating cells, are now known to be presented in a variety of in vitro and in vivo cells including normal human, rat or mouse cells, as well as cancer cells⁸³. Mitocytosis is a dynamic process that usually takes approximately 40–200 min to complete through three key steps including mitochondrial translocation, fragmentation, and migrasome formation⁸⁴. Upon mild mitochondrial stress induced by carbonyl cyanide 3-chlorophenylhydrazone (CCCP), the damaged mitochondria are transported to the peripheral edge of the cell, which is driven by the motor protein KIF5B, propelling mitochondria along microtubules (Fig. 3D). Once the injured mitochondria reach the vicinity of cell membrane, they are anchored by the motor protein Myo19 and bind them to cortical actin, where mitochondrial fission is triggered, leading to the generation of smaller mitochondrial fragments (Fig. 3D). Subsequently, these fragments are integrated with plasma membrane to form migrasomes during cell migration⁶⁵. Migrating cells remove such minor mitochondrial damage predominantly through mitocytosis, which is different from mitophagy-mediated clearance of severe damaged mitochondria⁶⁴. However, both mechanisms can work together to effectively prevent the accumulation of damaged mitochondria and thereby ensuring the health of migrating cells.

4.5. Mitophers for the removal of healthy mitochondria during cell development

Recently, Liu et al.⁸⁵ discovered a specific mitochondria-exporting mechanism in male C. elegans that removes excess healthy mitochondria from spermatids during sperm development (Fig. 3E). They observed that spermatids can release single-membrane vesicles that enclose individual healthy mitochondria with a diameter ranging 490–1100 nm, termed mitophers. Similar to aforementioned microvesicle generation, they are formed by directly outward budding of plasma membrane. However, the process of mitopher formation is a distinct and rapid budding/shedding that occurs within a few seconds. Mechanistical analysis reveals that the biogenesis of mitophers requires normal actin-filament dynamics to drive mitochondrial movement toward the cell membrane, and the tyrosine kinase SPE-8 partially mediates the extracellular protease-triggered extracellular protease signaling for their shedding off (Fig. 3E). Notably, blockade of mitopher formation results in excessive accumulation of mitochondria in mature sperm, jeopardizing sperm viability and fertility. Hence, generation of mitophers is pivotal for maintaining appropriate quantity of mitochondria during cell development. Nonetheless, future research will need to investigate the conserved role of mitophers in mammals and their post-secretory fate in non-autonomous functions such as mediating intercellular communication.

In summary, upon different physiological and stress/diseases conditions, cells have evolved diverse and complex conventional and unconventional routes to generate distinctive mitoEVs, as summarized in Fig. 3A–E and Table 1. Additionally, these subsets of mitoEVs can act locally or remotely to regulate mitochondrial homeostasis and intercellular communication for restoring cell function or promoting cell dysfunction (discussed below).

5. MitoEV-mediated crosstalk between cells and organs

As mentioned above, mitoEVs can shuttle healthy or injured mitochondria and associated components between cells and thereby modulating cellular and tissue homeostasis. This is particularly evident in brain cells, adipocytes, skeletal muscles, and cardiac cells, where continuous mitochondrial turnover is required for growth and function. For example, brain astrocytes can transfer healthy mitochondria through mitoEVs to local endothelial cells and pericytes under physiological conditions and during ageing to compensate mitochondrial deficit⁸⁶. Similarly, brown adipocytes expel oxidatively damaged mitochondria through EVs to interstitial space where they are engulfed by tissue resident macrophages⁸⁷. This process not only prevents mitochondrial dysfunction but also sustains adaptive thermogenesis in adipocytes.

In cardiovascular system, thrombin-stimulated platelets also release mitoEVs that are internalized by monocytes and macrophages. Following the uptake of mitoEVs, the recipient cells exhibit increased mitochondrial respiration, ATP production and spare respiratory capacity⁸⁸. This is in stark contrast with free mitochondria released by platelets, which failed to induce respiratory capacity and metabolic activity of polymorphonuclear neutrophils despite uptake⁸⁹. Interestingly, Allan et al.⁹⁰ recently showed that platelet-released mitoEVs can transfer mitochondria into neutrophils and increase their metabolic capacity; however, such internalization of platelet mitochondria rendered neutrophils unable to subsequently engulf bacteria, demonstrating reduced phagocytic capacity. In addition, activated platelets in atherosclerotic lesions can release mitoEVs that stimulate inflammatory response in endothelial cells and neutrophils⁹¹. Likewise, LPS-stimulated monocytes release mitoEVs that contain TOM22 protein and mitochondria-associated TNFα and RNAs. Such mitoEVs are subsequently taken up by endothelial cells, leading to triggering type I interferon- and TNFα-mediated inflammatory cascade and thereby augmenting inflammatory responses⁵⁵. In a similar manner, upon myocardial infarction (MI), cardiac fibroblasts shuttle damaged mitochondrial components via mitoEVs to macrophages and promote their inflammatory response⁹². Furthermore, mitoEVs released by cardiomyocytes during myocardial I/R transfer their carried mitochondrial components such as VDAC1, cytochrome c oxidase subunit 4 (Cox-IV) and mtDNA to neighboring fibroblasts and consequently, promoting fibroblast activation and proliferation⁹³. Together, these studies suggest that mitoEVs secreted from different types of cells are heterogeneous and diverse in terms of their receptor repertoire and EV cargo and thereby inducing distinctive consequences in recipient cell function.

On the other hand, the release of mitoEVs into blood stream can make them reach distant organs and thereby regulate their function. For instance, a recent study by Crewe et al.⁷⁰ showed that adipocyte-released mitoEVs enter systemic circulation and sequestered by cardiomyocytes, where induces transiently release of ROS and activates compensatory antioxidant signaling that shields cardiomyocytes from oxidative insults and limits cardiac ischemia injury in mice. This phenomenon is of translational relevance as higher levels of circulating mitoEVs are also found in metabolically unhealthy patients⁷⁰. Whether they have detrimental or beneficial effects on other organs in addition to the heart warrants further investigation.

In summary, mitoEV-mediated intercellular communication enables cells to exchange intact mitochondria and signaling molecules for mitochondrial quality control. They can also transfer mitochondrial portions and/or mitochondrial damage-associated molecular patterns (mtDAMPs) (e.g., mtDNAs and mtROS) to modulate the metabolic and inflammatory responses of recipient cells. Yet, mechanisms of cargo selection, targeting, and fate after uptake remain unclear. Delineating these gaps would be significant for translating mitoEV-based communication into therapeutic strategies.

6. Isolation and characterization of small and large mitoEVs

The isolation of small mitoEVs has long been challenging due to their nanoscale size, heterogeneity, and overlap with other small EV populations. Initially, researchers relied on classical sucrose step-gradient ultracentrifugation, which separates vesicles by buoyant density^{94, 95}. While this method enables partial enrichment of vesicular populations, it often lacks specificity and reproducibility, particularly in distinguishing small mitoEVs from intact mitochondria or contaminating vesicles. To overcome this limit, D’Acunzo et al.⁹⁶ developed a comprehensive protocol for isolating small mitoEVs from brain tissue (Fig. 4A). Their workflow begins with gentle enzymatic digestion to loosen the extracellular matrix and release all EVs while protecting vesicle integrity. After removing cells and debris by low and medium speed centrifugation and filtration, they applied a high-resolution iodixanol discontinuous density gradient to achieve superior separation of vesicle subpopulations, including microvesicles and exosomes. The authors compared iodixanol and sucrose gradients and showed that iodixanol medium offers better resolution of vesicle subsets. They validated the isolated fractions using nanoparticle tracking analysis (NTA), electron microscopy (EM), mitochondrial protein markers (e.g. PDHE1α, COX IV, VDAC), and functional assays (ATP production measured ex vivo). Remarkably, mitoEVs isolated via this method retained electron transport chain activity, confirming their bioenergetic competence⁹⁶. Recently, other protocols for collecting small mitoEVs have been developed that take different approaches^{97, 98}. For example, in brown adipose tissue (BAT), small mitoEVs are isolated from tissue culture supernatants using sequential differential centrifugation followed by ultracentrifugation⁹⁷. The isolated small mitoEVs are characterized by immunoblotting and flow cytometry to confirm their identity and mitochondrial content. Western blot analysis detects EV markers such as CD63 along with mitochondrial proteins (e.g., OXPHOS components), while cytoplasmic proteins like tubulin serve as negative controls to ensure purity. In addition, single-vesicle flow cytometry can quantify EV abundance across fractions and further identify those positive for MitoTracker Green (MTG) fluorescence, confirming the presence of mitochondrial components within these EVs derived from brown adipose tissue⁹⁷. This workflow omits gradient separation for speed and accessibility, but co-sedimentation of naked mitochondria with large EVs remains a limitation, which may require TEM for clear discrimination⁹⁷.

Figure 4: Isolation strategies for mitochondrial extracellular vesicles (mitoEVs).

(A) Tissue is enzymatically digested and centrifuged to collect EV-containing supernatants, which are separated by sucrose or iodixanol gradients to enrich exosomes, microvesicles, and mitovesicles. (B) Tissue- or media-derived mitoEVs are sequentially isolated by differential ultracentrifugation at increasing speeds (10 −110K g) and further purified by size exclusion chromatography (SEC). (C) Tissue-derived exophers are filtered and sorted by fluorescence-activated flow cytometry using mitochondrial and cell-type markers, followed by morphological and molecular validation. CD38, cluster of differentiation 38; VDAC, voltage-dependent anion-selective channel; COX, cytochrome c oxidase subunit; PDHE1α, pyruvate dehydrogenase E1 alpha; CM, cardiomyocyte. Modified from D’Acunzo et al.⁹⁶, Lou et al.⁹⁸ and Nicolás-Ávila et al.⁹⁹

For the isolation of functional mitoEVs, one recent study by Lou et al.⁹⁸ introduced an alternative strategy for isolating tissue-derived mitoEVs directly from skeletal muscle (Fig. 4B). Instead of relying on gradient fractionation, their protocol combined enzymatic tissue digestion with sequential centrifugation and ultracentrifugation at defined speeds (30,000g to enrich mitoEVs, and 110,000g for smaller non-mitoEVs), followed by additional washing steps to improve purity. To further refine and purify these small mitoEV preparations, they also incorporated size exclusion chromatography (SEC). Validation was carried out using a multi-step approach, including EM to confirm vesicular morphology, Western blotting for EV-specific (TSG101, CD63) and -negative markers (calnexin, GM130), mitochondrial protein markers (TOM20, COX-IV), mtDNA detection, and functional assays assessing mitochondrial genome transfer and ATP production⁹⁸. This comprehensive workflow enables the isolation of intact and functional tissue-derived mitoEVs and provides evidence that mitoEVs may serve as natural mitochondrial replenishment therapies for regenerative medicine.

Regarding large mitoEVs such as exophers (often >3 μm) that package damaged mitochondria and protein aggregates for disposal, their isolation requires a distinct protocol. For example, Nicolás-Ávila et al.⁹⁹ utilized fluorescence-activated cell sorting to purify cardiomyocyte-derived exophers from cardiac-fluorescent-reporter mouse strains (Fig. 4C). Specifically, heart tissue is enzymatically digested to release cardiomyocytes and their vesicles, followed by low and medium speed centrifugation to remove debris meanwhile preserving exophers. Unlike nanoscale vesicles, their purification relies on flow cytometry-based approaches, using fluorescent reporters and exclusion dyes to distinguish them from apoptotic bodies or EVs derived from non-cardiac cells. Validation employs imaging (confocal, EM) to confirm morphology, immunoblotting for mitochondrial and autophagy markers (TOM20, LC3), and functional uptake assays showing that exophers are engulfed by macrophages⁹⁹.

Together, these complementary protocols highlight the diversity of methodological approaches currently available for the isolation of small/large mitoEVs, each with distinct advantages and limitations. Gradient-based methods, such as iodixanol density separation, offer superior resolution of vesicle subpopulations but are labor-intensive. In contrast, ultracentrifugation provides a more accessible workflow yet risks co-isolating intact mitochondria and other non-EV contaminants, which may confound downstream functional analyses. To overcome these limitations, SEC is often applied as an additional purification step following ultracentrifugation. SEC is recognized as one of the most reliable methods for isolating EVs, effectively removing soluble proteins and debris without introducing new contaminants, thereby improving sample purity and experimental reproducibility¹⁰⁰. Importantly, the introduction of tissue-derived and cell-based systems expands the field beyond traditional cell culture approaches, enabling further exploration of their translational potential. The continued refinement of these techniques, along with standardized validation criteria, will be essential to ensure reproducibility and to fully uncover the physiological and therapeutic significance of mitoEVs.

7. Pathogenic effects of mitoEVs in cardiovascular disease

Under normal physiological conditions, mitoEVs are integral parts of MitoQC. However, during stress and diseases, cargo composition loaded in mitoEVs gets altered, which may activate inflammatory response or stimulate innate immune response. Such pathogenic effects of mitoEVs have been reported in autoimmune diseases, cancers, cardiovascular diseases, metabolic disorders, and neurodegenerative diseases. The following section will focus on discussing the pathogenic influence of mitoEVs in cardiac inflammation and adverse cardiac remodeling (Fig. 5).

Figure 5: Pathogenic effects of mitochondrial extracellular vesicles (mitoEVs) in cardiac inflammation and adverse cardiac remodeling.

(A) Cardiac fibroblast–derived mitoEVs interact with macrophages. (B) Cardiomyocyte-derived mitoEVs are transferred to cardiac fibroblasts. (C) Monocyte-derived mitoEVs communicate with endothelial cells. TOM20, translocase of the outer mitochondrial membrane 20; ATP, adenosine triphosphate; IL, interleukin; ARG1, arginase-1; MRC, mannose receptor C-type 1; NLRP3, NACHT, LRR and PYD domains-containing protein 3; SMA, spinal muscular atrophy; TNF, tumor necrosis factor; MCP1, monocyte chemoattractant protein-1; TGFβR1, transforming growth factor β receptor 1; HMGB-1, high mobility group box 1; cGAS, cyclic GMP-AMP synthase; STING, stimulator of Interferon Genes; ICAM, intercellular adhesion molecule; and VACM, vascular cell adhesion molecule. Modified from Zhao et al.⁹², Zhang et al.⁷⁴ and Puhm et al.⁵⁵

In this regard, Zhao et al.⁹² reported that mitoEVs released by resident cardiac fibroblasts during post-MI could drive excessive cardiac inflammation with upregulation of pro-inflammatory genes (e.g., IL-1β, IL-6, IL-18) and down-regulation of anti-inflammatory genes (e.g., IL-10, MRC1, Arg1) as well as increased immune cell infiltration, leading to aggravated cardiac dysfunction and exacerbated pathological ventricular remodeling in post-MI mice. Mechanistically, such mitoEVs are uploaded with mitochondrial components (i.e, TOM20, ATPβ, PDH) and mtROS, which are subsequently engulfed by local macrophages where activate the NLRP3-Caspase-1 inflammatory signaling pathway (Fig. 5A). Accordingly, administration of NLRP3 inhibitor CY-09 abolished these pathogenic effects triggered by fibroblast-derived mitoEVs in post-MI mice.

Additionally, similar inflammatory responses could be also activated by cardiomyocyte-released mitoEVs during myocardial I/R. For example, Zhang et al.⁷⁴ demonstrated that mitoEVs derived from post-I/R cardiomyocytes carry damaged mitochondrial components and mtDNAs which are internalized by cardiac fibroblasts and thereby activating cyclic GMP-AMP synthase (cGAS)-stimulator of interferon genes (STING) pathway to promote expression of inflammatory cytokine genes (i.e., IL-1β, TNFα, MCP-1, and HMGB-1) and fibrotic genes (i.e., α-SMA, collagen 1, and TGFβR1), resulting in severe cardiac fibrosis and ventricular dysfunction (Fig. 5B). Consistently, either pre-clearance of endogenous mtDNAs from mitoEVs or pre-inhibition of cGAS-STING in fibroblasts effectively block such cardiomyocyte-mitoEV-induced pathological activation of post-I/R cardiac fibrosis. Most importantly, generation of such pathogenic mitoEVs in I/R-cardiomyocytes is dependent on the presence of Ambra1. Accordingly, myocardial-specific knockdown of Ambra1 inhibits mitoEV formation in cardiomyocytes and consequently, blocking cGAS/STING activation and reduced expression of fibrotic genes and inflammatory cytokines as well as improved cardiac function in myocardial I/R mice. Similar findings were also observed in diabetic murine hearts, where diabetic cardiomyocytes secreted mtDNA-enriched mitoEVs, which were then engulfed by cardiac fibroblasts, resulting in the activation of TLR9 signaling and the cGAS-STING pathway to initiate pro-fibrotic process and adverse cardiac remodeling¹⁰¹. Therefore, specific interventions on such mitoEV formation and efflux from the injured cardiomyocytes may be a new strategy to prevent or treat disease/stress-caused myocardial fibrosis and heart failure.

In addition to mitoEVs released from stressed cardiac fibroblasts and cardiomyocytes that exert pathological effects, it is plausible that other cardiac cell subpopulations, such as monocytes/macrophages and endothelial cells, may also secrete pathogenic mitoEVs. A prior work by Puhm et al.⁵⁵ showed that LPS-treated monocytes secrete mitoEVs that carry TOM22 together with TNFα, IL-1β, and oxidized mtRNAs (Fig. 5C). These monocyte-derived mitoEVs are then taken up by endothelial cells, leading to increased expression of IL-8, ICAM-1 and VCAM and severe vascular inflammation (Fig. 5C). Nonetheless, it remains to be clarified whether mitoEVs released from other cell types including monocytes/macrophages, neutrophils, platelets, endothelial cells, and adipocytes also contribute to the development of cardiovascular pathologies in vivo during various cardiovascular disease and stress settings.

Overall, the pathogenic role of mitoEVs in driving adverse cardiac remodeling appears to stem from the encased mtDAMPs (i.e., mtDNAs, mtRNAs, mtROS, ATP, etc) which are originated from injured mitochondria of parental cells. Therefore, identifying and targeting mitoEV-formation machinery may provide therapeutic avenues for the treatment of inflammatory and cardiovascular diseases.

8. Protective and therapeutic potentials of mitoEVs in cardiovascular disease and ageing

Due to the heterogeneous composition of mitoEVs and different cell origins, it is not surprising that mitoEVs might produce cardioprotective effects against stress/disease-induced cardiovascular injury and ageing as well as have therapeutic potential for the treatment of cardiovascular diseases. Indeed, recent multiple studies have demonstrated that natural mitoEVs collected from both healthy tissues/cells and artificial mitoEVs can restore/enhance metabolic processes in recipient cells and resolve excessive local inflammation and consequently, offering cardio-protection in the settings of MI, myocardial I/R, doxorubicin-induced cardiomyopathy, and ageing as discussed below.

8.1. Beneficial effects of healthy tissue/cell-derived mitoEVs in MI/R hearts

Mitochondria constitute ≈ 30% of cardiomyocyte mass and is essential for maintaining cardiac energy metabolism and contractile function. During MI and MI/R, severe mitochondrial damage and dysfunction are the major causes of cardiac injury¹⁰². Hence, strategies that can effectively restore mitochondrial function represent immense benefit in repairing ischemia-induced cardiomyocyte damage. In this regard, a recent study by Liu et al.¹⁰³ demonstrated that mitoEVs harvested from healthy cardiac tissue can transfer ATP5a1, the α-subunit of mitochondrial respiratory chain complex enzyme 5, into I/R-injured cardiomyocytes, thereby restoring oxidative phosphorylation, stabilizing mitochondrial membrane potential, and suppressing excessive ROS generation (Fig. 6A). Accordingly, intramyocardial delivery of mitoEVs in MI/R-injured mice significantly improved ventricular contractility, reduced infarct size, and mitigated adverse remodeling by inhibiting ferroptotic cell death and preserving mitochondrial integrity in cardiomyocytes (Fig. 6A). Furthermore, they observed that ATP5a1 encased in these mitoEVs primarily originates from cardiomyocytes of healthy murine hearts. Consistently, mitoEVs collected from ATP5a1-silenced cardiomyocytes negated these cardio-protective effects induced by healthy cardiomyocyte-derived mitoEVs. To enhance their translational feasibility, they engineered adipose-derived stem cells (ADSCs) with ATP5a1 overexpressing to produce ATP5a1-enriched mitoEVs, which exhibited superior efficacy in preserving mitochondrial function and alleviating MI/R-induced cardiac injury. Collectively, this elegant work (Fig. 6A) highlights that healthy tissue-derived mitoEVs could act as natural boosters of mitochondrial function to promote repair of cardiac injury, and ATP5a1-loaded mitoEVs would be a promising mitochondria-targeted therapeutic platform for the treatment of ischemic heart disease. Nonetheless, the mechanism by which cells selectively load ATP5a1 into EVs in normal steady state requires further investigation.

Figure 6: Protective pathways of mitochondrial extracellular vesicles (mitoEVs) from different cellular sources in the heart.

(A) mitoEVs isolated from healthy cardiac tissue or ATP5a1-enriched mitoEVs derived from adipose-derived stem cells (ADSCs) are delivered to ischemia/reperfusion (I/R) injured hearts. (B) Human induced pluripotent stem cell-derived cardiomyocytes (iCMs) release mitoEVs transferred to hypoxia-injured cardiomyocytes. (C) Mesenchymal stem cell (MSC)-derived large mitoEVs (300–800 nm) interact with doxorubicin-injured cardiomyocytes. ATP, adenosine triphosphate; ROS, reactive oxygen species; 4-NHE, 4-hydroxynonenal; MDA, malondialdehyde; PTGS2, prostaglandin-endoperoxide synthase 2; GPX4, glutathione peroxidase 4; PGC-1α, peroxisome proliferator-activated receptor gamma coactivator 1-alpha; LVEF, left ventricular ejection fraction. Modified from Liu et al.¹⁰³, Ikeda et al.⁹³ and O’Brien et al.¹⁰⁷

In a similar way, Ikeda et al.⁹³ explored the use of human-induced pluripotent stem cell (iPSC)-derived cardiomyocytes (iCMs) as a source of mitoEVs for restoring cardiac bioenergetics and improving heart function in ischemic myocardium (Fig. 6B). Their in vitro studies showed that mitoEVs harvested from human iCMs contain functional mitochondria and non-mitochondrial mRNAs (i.e, PGC-1α) that are associated with mitochondrial biogenesis and energy metabolism. Importantly, these cargoes encased in human iCM-derived mitoEVs can be transferred into the recipient cardiomyocytes and fused to their endogenous mitochondrial networks together with mitochondrial biogenesis initiated by PGC-1α. Consequently, treatment of hypoxia-injured cardiomyocytes with these iCM-mitoEVs greatly restored intracellular ATP production, enhanced contractile profiles, and increased cell survival, in comparison with control-treated cells. More interestingly, in an in vivo mouse model of MI, intramyocardial injection of these human iCM-mitoEVs into the peri-infarct region significantly restored bioenergetics and activated PGC-1α-mediated mitochondrial biogenesis and thereby improving left ventricular (LV) ejection fraction and preventing post-MI pathological cardiac remodeling, compared to control group (Fig. 6B). However, whether human iCM-mitoEVs stimulate innate immune response in mice remains to be clarified and warrants future investigation, as mitoEVs can also be taken up by cardiac macrophages.

8.2. Cardio-protective effects of MSC-derived mitoEVs against doxorubicin injury

Doxorubicin (Dox)-induced cardiomyopathy is a significant source of death in cancer survivors¹⁰⁴. Dox-induced ROS production and dsDNA breaks lead to the collapse of mitochondrial membrane potential and cell apoptosis¹⁰⁵. Interestingly, MSCs have been shown to possess the ability to both prevent and reverse Dox-induced heart failure through mitochondrial transfer¹⁰⁶. A recent study by O’Brien et al.¹⁰⁷ describe an in vitro clinical trial evaluating MSC-derived mitoEVs for efficacy in treating Dox-induced cardiac injury using human patient-specific iCMs. They collected large EVs (size ranging from 300–800 nm) from MSCs and observed that these large EVs (L-EVs) are enriched with intact mitochondria. Incubation of Dox-injured patient-iCMs with L-EVs remarkably augmented ATP production and mitochondrial biogenesis and thereby, improving contractility and cell survival, compared to control cells treated with red blood cell-derived large EVs that lack mitochondria (Fig. 6C). This study suggests that MSC-derived mitoEVs would be an ideal cell-free therapeutic avenue in treating Dox-induced cardiac injury. However, future studies are needed to test whether MSC-mitoEVs can efficiently transfer intact mitochondria to Dox-treated hearts in vivo using a chronic Dox-cardiomyopathy model.

8.3. Artificial mitoEVs for the delivery of healthy mitochondria to ischemic hearts.

Natural mitoEVs produced by heathy tissue and other donor cells can transfer bioenergetic support to damaged cells, but challenges remain in low delivery efficiency, poor stability, and limited quantity. To address these concerns, Chu et al.¹⁰⁸ recently developed gelatin methacrylate (GelMA)-based gelated microvesicles modified with phosphatidylserine (PS) to serve as carriers for effectively delivering mitochondria obtained from ADSCs (Fig. 7A). These mitochondria-loaded PS-gelated-mitoEVs (named Mito@Microgels-PS) have several advantages over natural mitoEVs such as preserved mitochondrial integrity, controlled release, and widely uptake by most cells including cardiomyocytes through phosphatidylserine receptor (PSR)-mediated interactions. As a matter of fact, the in vitro experimental results showed that Mito@Microgels-PS effectively delivered mitochondria into hypoxia-injured cardiomyocytes and restored mitochondrial network architecture and subsequent ATP production, leading to enhanced electrophysiological activity and contractile function. In addition, Mito@Microgels-PS can be taken up by macrophages and effectively repolarize them from pro-inflammatory (M1) to anti-inflammatory phenotype (M2). The therapeutic effectiveness of Mito@Microgels-PS was further tested in a rat myocardial infarction model, demonstrating improved cardiac function together with reduced fibrosis, inflammation, and cardiomyocyte death. Hence, Mito-Microgels-PS could represent a promising method for delivering functional mitochondria to treat ischemic heart disease. In a similar manner, Dong et al.¹⁰⁹ recently generated another artificial mitoEVs that carry active mitochondria, using size-controlled extrusion of enucleated MSCs (referred to as Mito@euMVs, Fig. 7B). They observed that Mito@euMVs efficiently delivered mitochondria into macrophages and cardiomyocytes, thereby improving mitochondrial function and facilitating repair of ischemia-induced cardiac injury in a diabetic rat model with MI. Overall, these artificial mitoEVs provide a stable, biocompatible, and efficient delivery tool, which may have great potential to treat a range of cardiovascular diseases and other disorders caused by mitochondrial dysfunction, but future studies are required to validate this potential in a large animal model.

Figure 7: Schematic illustration of artificial mitochondrial extracellular vesicles (mitoEVs) for delivery of healthy mitochondria to the ischemia/reperfusion (I/R) hearts and subsequent repair.

(A) Gelatin methacrylate–based microgels modified with phosphatidylserine (Mito@Microgels) effectively encapsulate and deliver functional mitochondria derived from adipose-derived stem cells (ADSCs). (B) Denucleated mesenchymal stem cell (MSC)-derived microvesicles (Mito@euMVs) enable efficient mitochondrial transfer to macrophages, thereby enhancing cardiomyocyte repair. PS, phosphatidylserine; UV, ultraviolet; CX43, connexin 43; ROS, reactive oxygen species. Modified from Chu et al.¹⁰⁸ and Dong et al.¹⁰⁹

8.4. Beneficial effects of mitoEVs in ageing

In addition to their protective roles in cardiac injury, mitoEVs have emerged as important regulators of aging-associated mitochondrial dysfunction, a central driver of cardiovascular aging. Aging is characterized by progressive impairment of mitochondrial quality control, bioenergetic decline, and chronic low-grade inflammation, all of which contribute to age-related cardiovascular disease¹¹⁰. Using high-resolution multicolor flow cytometry, Zhang et al.¹¹¹ showed that circulating EVs from healthy individuals across the lifespan contain functional respiring mitochondria. Notably, both the abundance of mitochondrial cargo and its respiratory activity within multiple EV subpopulations, particularly those originating from immune and progenitor cells, progressively declined with aging. These age-associated changes in EV-linked mitochondrial function were closely associated with immunosenescence and inflammaging, processes known to contribute to age-related cardiovascular dysfunction. Consistent with an age-dependent decline in mitochondrial stress responsiveness, Guo et al.¹¹² identified mitochondrial-derived vesicles (MDVs), an early mitochondrial quality-control mechanism, as dynamic regulators of aging. In models expressing amyotrophic lateral sclerosis (ALS)–linked oxidizable SOD1 mutants, a neurodegenerative disease characterized by progressive motor neuron degeneration and profound mitochondrial oxidative stress, mitochondrial stress initially elicited enhanced MDV formation and compensatory quality-control responses at younger ages; however, this adaptive MDV response and overall mitochondrial responsiveness to oxidative stress progressively deteriorated with aging, coinciding with premature cellular senescence and accelerated aging phenotypes.

In parallel, Lazo et al.¹¹³ reported that EV-associated circulating mtDNA in human plasma declined with age based on both cross-sectional and longitudinal analyses of a middle-aged cohort, reflecting mtDNA encapsulated within EVs rather than free circulating DNA. Importantly, functional analyses within the same study revealed that EVs from young individuals preserved mitochondrial bioenergetics, as recipient cells treated with young plasma EVs exhibited significantly higher basal and maximal respiratory capacity compared with cells treated with EVs from older donors, indicating that EVs derived from young donors may enhance mitochondrial bioenergetics.

While studies regarding the use of mitoEVs in the context of aging remain limited, several studies employing EVs have demonstrated beneficial effects in aging by enhancing mitochondrial bioenergetics^114–116. Specifically, EVs derived from young mouse plasma were efficiently taken up by aged tissues, where they restored mitochondrial oxidative metabolism by upregulating PGC-1α–dependent mitochondrial biogenesis, improving oxidative phosphorylation capacity, and increasing ATP production¹¹⁵. These bioenergetic improvements were mediated, at least in part, by EV-enriched miRNAs, known as MitomiRs, that positively regulated mitochondrial metabolism¹¹⁴, whereas age-associated EV miRNAs exerted opposing effects by suppressing PGC-1α signaling and exacerbating mitochondrial dysfunction¹¹⁵. In addition, young EVs modulated antioxidant and autophagy-related pathways, suggesting improved mitochondrial quality control and reduced oxidative stress in aged cells^{115, 116}. Furthermore, a recent study by Peng et al.¹¹⁷ showed that small EVs derived from early-passage mesenchymal stem cells isolated from human exfoliated deciduous teeth (SHED) rejuvenated senescent cells by remodeling mitochondrial dynamics. Treatment with these stem cell–derived EVs promoted Drp1-dependent mitochondrial fission, facilitating Drp1 translocation to mitochondria, thereby restoring mitochondrial morphology, improving mitochondrial function, and alleviating bioenergetic deficits associated with cellular senescence¹¹⁷. In further support of EV-mediated regulation of mitochondrial function during aging, recent work by Zheng et al.¹¹⁸ demonstrated that engineered and serum-derived EVs can directly modulate mitochondrial energetics in aged or metabolically compromised recipient cells by delivering mitochondrial regulatory cargo, including bioenergetic enzymes and mitochondria-associated signaling molecules. In this study, EV treatment enhanced oxidative phosphorylation efficiency, restored mitochondrial membrane potential, and increased ATP production in aged bone cells, effects that were linked to improved mitochondrial network integrity and metabolic resilience. These findings further support the concept that EV-mediated transfer of mitochondrial-related cargo represents an effective strategy to counteract age-associated mitochondrial dysfunction.

Interestingly, studies of the aging brain have demonstrated that mitoEVs are accumulated more prominently in females than in males with advancing age, particularly those of neuronal origin¹¹⁹. This female-specific enrichment of neuronal mitoEVs suggests an increased reliance on vesicle-mediated mitochondrial disposal and intercellular signaling pathways during aging, potentially reflecting heightened mitochondrial stress or a compensatory response to age-associated impairment of mitophagic capacity¹¹⁹. Supporting the relevance of sex-dependent mitoEV biology in humans, plasma-based analyses of neuron-derived EVs from African American people living with HIV have revealed higher mtDNA content in males compared with females¹²⁰, consistent with increased vesicle-associated mitochondrial stress in male neurons under chronic disease conditions.

Collectively, these findings indicate that enhancement of mitochondrial energetics represents a recurring and central mechanism through which EVs exert rejuvenating effects in aging. However, although multiple lines of evidence demonstrate that EVs can improve mitochondrial bioenergetics, dynamics, and quality control in aged cells, direct evidence specifically employing mitoEVs, defined by the selective enrichment of mitochondrial components, remains limited. Given their inherent capacity to deliver mitochondrial cargo with higher specificity, mitoEVs represent a particularly promising but underexplored therapeutic avenue for targeting mitochondrial dysfunction in aging. Future studies directly comparing conventional EVs and mitoEVs, as well as defining their cargo composition, targeting specificity, and long-term safety, will be essential to fully establish mitoEV-based strategies for ageing and age-related cardiovascular disease.

9. Conclusion and Future Perspectives

In this review, we summarize the current up-to-date literature regarding: 1) the mechanisms underlying the generation of MDVs and mitoEVs upon physiological and pathological conditions; 2) MDV-mediated organelle crosstalk and mitoEV-mediated communications between different types of cells for maintaining cellular and tissue homeostasis; and 3) pathogenic effects of mitoEVs and their therapeutic potentials in cardiovascular diseases with focusing on ischemic heart disease and ageing.

Beyond their critical roles in maintaining mitochondrial health, both MDVs and mitoEVs can be detected in body fluids (i.e., blood, saliva, sweat). Importantly, recent studies have demonstrated that components loaded in MDVs and mitoEVs are highly selective and specific which are linked to both physiological and pathological conditions. Herein, their presence in body fluids may have potential application as biomarkers for the diagnosis of diseases like cancer, cardiovascular and neurodegenerative disorders. Future investigations should aim to dissect the molecular mechanisms governing MDV/mitoEV formation, MDV/mitoEV cargo selective loading, alongside establishing standardized isolation and characterization protocols. Integration of single-vesicle imaging, multi-omics profiling, and targeted delivery approaches will accelerate translation of these vesicular systems as both diagnostic biomarkers and mitochondria-based therapeutics. Ultimately, better understanding of the MDV-mitoEV axis may open new horizons for precision interventions aimed at preserving mitochondrial health and restoring cardiac energetics.