TNTs, also called membrane nanotubes, are long tubular membrane structures (Figure 1). TNTs were discovered as unique structures for intercellular communication for the first time by Rustom and coworkers in 2004 (16). Recent studies have shown that cells in cardiovascular systems (such as cardiomyocytes, cardiac fibroblasts, endothelial cells, and vascular smooth muscle cells) can exchange mitochondria with their neighboring cells via TNTs (Table 1) (8–12, 15, 17–24). The intercellular transfer of mitochondria through TNTs could be unidirectional or bidirectional. The diameter of TNTs ranges from 50 to 1,000 nm (8, 17, 21, 22, 24). The length of TNTs, which differs in various types of cells, is usually 5–120 μm (8, 9, 17). Actin is the principal component of TNTs, and filamentous actin (F-actin) polymerization is necessary for the assembly of TNTs (15, 18, 23). Besides actin, another cytoskeleton component, microtubule, is also found in some TNTs (23). Both F-actin and microtubules could act as cytoskeletal tracks for the movement of mitochondria. TNTs containing both microtubules and F-actin are large in diameter (>0.7 μm) and responsible for the long-distance delivery of mitochondria. TNTs containing only actin are small in diameter (<0.7 μm) and in charge of the short-distance transport of mitochondria (19, 21, 25–28).

There are two different mechanisms involved in the formation of TNTs. (1) Cell dislodgment mechanism. Cells contact each other, then quickly migrate in opposite directions, retaining a thread of membrane between these two detached cells which finally develop into TNTs (9, 18). (2) Actin-driven protrusion mechanism. Filopodia-like membrane protrusions extend beyond the cell and elongate in an F-actin polymerization-dependent manner. Then the elongated protrusions connect with the target cells or other protrusions to form the TNTs (9, 12, 29). Unlike other cellular protrusions, these filopodia-like protrusions do not anchor to the substratum but suspend in the culture medium, which makes it possible for long-distance communication between cells (30).

M-Sec, also known as tumor necrosis factor α-inducible protein 2 (TNFαIP2), is reported as a key trigger of TNTs formation. Studies have shown that interaction of M-Sec with RalA can induce the assemble of exocyst complex and then initiate F-actin polymerization, while Cdc42 may be required for the extension process of TNTs (31). The expression of M-sec expression is regulated by many stimuli. Oxidative stress can activate p53, which in turn upregulates M-Sec expression by enhancing epidermal growth factor receptor expression or activating Akt/PI3K/mTOR pathway (32). Treatment of mesenchymal stem cells (MSCs) with TNF-α can increase M-Sec expression and trigger TNTs formation with cardiomyocytes via the NF-κB signaling pathway (15).

Some studies have demonstrated that gap junction protein connexin 43 (CX43) is necessary for TNTs formation and TNT-mediated intercellular mitochondrial transfer in non-cardiovascular systems (33–35). However, new evidence has shown that there is no Cx43 exists in the TNTs between cardiomyocytes and cardiac fibroblasts (9). Wang and coworkers have found that Cx43 only anchors at one end of TNT between two human umbilical vein endothelial cells (36). Since gap junctions do not allow the passage of large molecules (>1.2 kDa), Cx43 in the TNTs might only mediate intercellular electrical coupling but not cell-to-cell mitochondrial delivery (36). The role of Cx43 in the formation of TNTs and TNT-mediated mitochondrial transfer in the cardiovascular system still needs to be further explored.

Recently, mitochondrial Rho GTPase 1 (Miro1) has been reported to play a critical role in mediating mitochondrial movement along the TNTs (14, 15). Miro1 is a tail-anchored mitochondrial outer membrane protein. After combing with the adaptor protein TRAK1/2, Miro1 can recruit motor proteins (such as kinesin) and initiate microtubule-based mitochondrial movement (37). In a cardiomyocytes and cardiac myofibroblasts co-culture system, mitochondrial transport along microtubules in TNTs is mediated by KIF5B, which is a membrane of the kinesin superfamily (23). Recent studies have shown that Miro1 can also mediate actin-based mitochondrial transport via binding with motor protein Myo19 within individual mouse fibroblasts (38). However, whether Miro1 and Myo19 are involved in the mitochondrial movement along F-actin in TNTs still needs to be further investigated.

Another pathway for cell-to-cell mitochondrial transfer is through EVs (Figure 2). The properties of EVs that transfer intact mitochondria or mitochondrial components in the cardiovascular system are listed in Table 2.

EVs are phospholipid membrane-bound microparticles released by cells. Exosomes and microvesicles are two major forms of EVs. Exosomes are small EVs (30–150 nm in diameter) that originated from the endosomal networks and are considered to deliver lipids, RNAs, and mitochondrial components (such as mtDNA). Microvesicles derived from cellular plasma membranes are larger than exosomes (100–1,000 nm in diameter) (43). Since mitochondria are elongated organelles with a diameter of 500–1,000 nm, the intact mitochondria more likely exist in the microvesicles but not in the exosomes (44). Many harmful stimuli, such as lipopolysaccharide, can induce endothelial cells to release EVs. Then the EVs are taken up by the recipient cells and cause inflammatory responses. The inflammatory responses might be due to the pro-inflammatory effect of mtDNA in the EVs (39, 41). On the contrary, hypoxia-injured cardiomyocytes can uptake the EVs containing respiratory-competent mitochondria to increase their rescue ability (40).

Recent studies have suggested that selective packaging of mitochondrial content into EVs depends on optic atrophy 1 (OPA1) and sorting nexin 9 (Snx9) proteins (45), but the exact mechanism is unclear. The formation of exosomes is initiated via membrane invagination to generate multivesicular late endosomes. Then the multivesicular late endosomes fuse with the plasma membrane, leading to the release of exosomes into the extracellular space. The biogenesis and release process of microvesicles is different from exosomes. Microvesicles are generated via membrane blebbing and then released into the extracellular environment by separating from the plasma membrane in Ca²⁺-dependent enzymatic machinery (46). Integrins on the surface of EVs have been widely reported as major regulators of anchoring EVs on recipient cells (47, 48). Once attaching the recipient cells, EVs can directly fuse with the recipient cell membrane or be engulfed by recipient cells through multiple pathways, including clathrin-dependent endocytosis, caveolin-mediated endocytosis, lipid raft-mediated endocytosis, phagocytosis, and micropinocytosis (49–51).

In 2020, a new type of mitochondria-containing EVs called exophers was discovered in hearts by Nicolas et al. (42). The structure of exophers from cardiac tissues is similar to that of neural exophers of C. elegans, which mainly contain misfolded proteins and damaged mitochondria (52). Different from the traditional EVs, cardiac exophers are large membrane-surrounded microparticles with an average diameter of 3.5 μm, which allows intact mitochondria to be packed in (42). The formation of cardiac exophers is motivated by the cardiac-specific autophagy mechanism. A large number of exophers extruded by cardiomyocytes can be engulfed by cardiac-resident macrophages via Mertk-mediated endocytosis. Such kind of crosstalk between cardiomyocytes and immune cells is required for the maintenance of mitochondrial fitness and cardiovascular health (42).

Many studies have shown that the intact respiratory competent mitochondria exist in healthy human and animal blood which might be released by resting or activated platelets (53, 54). Likewise, mitochondria can also be released into the environment in the form of naked organelles by many normal or abnormal cells beyond platelets (54). For example, extracellular mitochondria are found in the endothelial progenitor cells culture system under physiological conditions (55). Monocytic cells can extrude naked mitochondria after being attacked by lipopolysaccharide (41). It has been proven that intact cell-free mitochondria are released from platelets through an actin-dependent but microtubule-independent mechanism (56).

However, the uptake mechanism of cell-free mitochondria by recipient cells has not been fully clarified. A few previous reports have shown that MSCs engulf platelet-derived functional mitochondria through clathrin-mediated endocytosis and enhance their pro-angiogenic activity (49). Some evidence has demonstrated that autologous mitochondria can be internalized into cardiomyocytes through actin-dependent endocytosis. Neither caveola-mediated nor clathrin-mediated endocytosis is involved in the mitochondrial internalization into cardiomyocytes (57). It has been reported that H9C2 rat cardiomyocytes can recognize and engulf exogenous mitochondria released from human uterine endometrial gland-derived MSCs in a co-incubation system. The uptake of mitochondria by cardiomyocytes is mainly via micropinocytosis (58). During the mitochondrial internalization process, cells can discriminate intact mitochondria from other similar microparticles and only engulf mitochondria (59). So the internalization mechanism of naked mitochondria might be different according to the types of recipient cells and the origin of naked mitochondria.

Other pathways, such as cell fusion, are also found to be involved in intercellular mitochondrial transfer. In 2003, bone marrow-derived MSCs was reported to donate their mitochondria to cardiomyocytes through cell fusion. The cell fusion between MSCs and skeletal muscles is at a very low rate, suggesting that cell fusion is a kind of cell-specific machinery for cell-to-cell mitochondrial transfer (60).

The cell-to-cell mitochondrial transfer has been detected in the cardiovascular system under physiological conditions (Table 3). In 2005, the unidirectional mitochondrial transfer from neonatal cardiomyocytes to endothelial progenitor cells was observed for the first time by Koyanagi et al. (8). After receiving donor mitochondria, endothelial progenitor cells acquire a cardiomyocyte-like phenotype through reprogramming (8). Meanwhile, a bidirectional mitochondrial transfer has been detected between cardiac myocytes and MSCs in a co-culture system (22). Migration of mitochondria from MSCs into fully differentiated cardiomyocytes can reprogram the adult cardiomyocytes and regress them to a progenitor-like state (20). Likewise, the mitochondrial transfer from embryonic cardiomyocytes to MSCs initiates stem cells differentiation toward cardiac cells, which might be an essential mechanism of stem cell-based therapies for cardiovascular disorders (22). Studies have also shown that mitochondrial transfer between vascular smooth muscle cells and MSCs is required to promote stem cells proliferation (10). These results suggest that intercellular mitochondrial transfer might play an important role in the regulation of cardiovascular system development.

Although cardiomyocytes account for about 70–85% of the adult myocardial tissue volume (72), non-myocytes in cardiac tissues are essential for heart health. Cardiomyocytes and cardiac fibroblasts are the two most abundant cell types in mammalian hearts. Recent studies have demonstrated that mitochondrial exchange between cardiomyocytes and fibroblasts is a distinct intercellular communication pattern, which might be indispensable for normal cardiac function (9). But the exact molecular mechanism remains unclear. In 2020, Nicolas-Avila and colleagues found that cardiomyocytes can eliminate their abnormal mitochondria by delivering them to heart-resident macrophages under physiological conditions. Harmful stimuli, such as ischemia or isoproterenol challenge, can enhance the efficiency of mitochondrial transfer and accelerate the clearance of dysfunctional mitochondria (42). The mitochondrial transfer from cardiomyocytes to macrophages is beneficial to maintain the mitochondrial fitness of cardiomyocytes, reduce the accumulation of pro-inflammatory material, and prevent the activation of inflammasome (42). Studies have also demonstrated a low mitochondrial transfer between heart-resident macrophages and other non-myocytes (such as endothelial cells), suggesting intercellular mitochondrial transfer within the heart has a highly cell-specific feature. These studies demonstrated that cell-to-cell mitochondrial transfer might be essential for maintaining normal cardiac homeostasis.

Under pathophysiology conditions such as ischemic cardiomyopathy, damaged cells can not only release dysfunctional mitochondria to ask for help but also take up exogenous functional mitochondria to rescue their own mitochondria network (Table 3). The transfer of healthy mitochondria toward injured cells has multiple protective mechanisms include the following. (1) Improvement of mitochondrial biogenesis. The perturbation of mitochondrial biogenesis is known as the fundamental mechanism of cardiovascular diseases (6). Transfer of healthy mitochondria to the injured cardiomyocytes or endothelial cells can increase cellular ATP levels through elevating oxidative phosphorylation and tricarboxylic acid (TCA) cycle and reducing glycolysis (12, 15, 57, 61, 69). The improvement of mitochondrial biogenesis is due to the renewal of damaged mitochondrial DNA and increased expression of mitochondrial respiration-related protein through activation of peroxisome proliferator-activated receptor-gamma coactivator 1-alpha (PGC-1α)-mediated pathway (40, 57, 70, 71). It has been reported that this beneficial effect of the mitochondrial transfer can last for a long time (at least 28 days) in ischemic cardiomyocytes (73), which is in contrast to the short-term improvement of energy metabolism found in normal cardiomyocytes (74). (2) Enhancement of antioxidant capacity. Mitochondria are vital organelles that regulate redox balance via their pro-oxidant and antioxidant functions. Oxidative stress-induced injury is involved in the pathogenesis of many cardiovascular diseases, including atherosclerosis, myocardial ischemia-reperfusion injury, and hypertension (75–77). Inflammatory response, triggered by excessive ROS level, is also associated with vascular dysfunction in many pathophysiology conditions (78, 79). Recent studies have shown that the delivery of healthy mitochondria to cardiac cells or endothelial cells can protect them against oxidative damage (14, 70). Transplantation of MSCs to a doxorubicin-induced animal cardiomyopathy model also alleviates cardiac inflammation via mitochondrial transfer (15). The protective mechanism might be due to the overexpression of heme oxygenase-1, which has well-known properties of anti-oxidative and anti-inflammatory activities (14). (3) Reduction of apoptosis. Apoptosis is one of the most common patterns of programmed cell death in the cardiovascular system (80). Cardiomyocytes and endothelial cells are prone to apoptosis under various cellular stress (such as hypoxia, chemicals, and metabolic stress). Many studies have shown that transfer of healthy mitochondria to these injury cells can reduce apoptosis (11, 15, 18, 23, 69). The anti-apoptotic effect of mitochondrial transfer has been shown to have a gender-specific characteristic in pregestational diabetes mellitus-exposed offspring (61). The mechanism of mitochondrial transfer-induced anti-apoptosis might involve the decrease of Bax/Bcl-2 ratio and the inhibition of caspase-3 activity (11, 61).

In short, a series of studies implied the significance of mitochondrial transfer in the cardiovascular system. In physiological conditions, cardiac fibroblasts and cardiomyocytes show frequent intercellular communication through bidirectional mitochondrial transfer, which is critical in maintaining normal cardiac function (9). Although this phenomenon is observed in an in vitro model, whether it exists in vivo has not been confirmed. Meanwhile, transferring distressed mitochondria to macrophages is also critical to the fitness of cardiomyocytes (42). It is undoubted that mitochondria containing the information of donor cells once internalized into recipient cells can trigger a cascade of response, which in turn acts on donor cells. For instance, cardiomyocytes and endothelium suffered ischemia/reperfusion injury deliver mitochondria as signals to MSCs to ask for help. After receiving mitochondria, MSCs enhance the biogenesis of mitochondria and promote the capacity of anti-apoptosis, then generously donate functional mitochondria to distressed cells (14).

MSCs and progenitor cells have been recognized as the preferred mitochondria donors for the treatment of cardiovascular diseases due to their abundant sources and high mitochondrial respiratory activity. Animal experiments and clinical trials have shown that transplantation of MSCs can successfully repair injured myocardium in ischemic cardiomyopathy (81–84). The protective mechanism of MSCs therapy has been proven mainly via mitochondrial transfer in vitro or in vivo studies. For example, MSCs can prevent vascular endothelial cells injury in an ischemia-reperfusion model by enhancing aerobic respiration and reducing apoptosis through transferring mitochondria to endothelial cells (12). Mitochondria transferring from MSCs can protect cardiomyocytes against oxidative stress-induced injury by improving mitochondrial respiration function (11, 14, 15, 17, 70). MSCs transplantation has also been reported to reduce myocardial fibrosis, alleviate left ventricular dilatation, and improve cardiac function in an animal model of anthracycline-induced cardiomyopathy through mitochondrial transfer (15).

MSCs used in clinical trials can be originated from widely various tissues. Paliwal and coworkers have reported that the MSCs derived from pulp and Wharton's jelly have lower mitochondrial transfer abilities but higher mitochondrial respiration capacities than those of MSCs from bone marrow and adipose (70). Compared with bone marrow-derived MSCs, human-induced pluripotent stem cell-derived MSCs have higher efficiency of mitochondrial transfer due to their higher expression of Miro1 and TNFαIP2 (15). So the difference in mitochondrial transfer capacity and effectiveness of these tissue-specific MSCs need to be considered in cell-based therapy for cardiovascular diseases.

Whether the beneficial effect of MSCs transplantation is mainly due to mitochondrial transfer remains controversial. A few previous studies have reported that MSCs can rescue damaged cells through paracrine mechanisms (96). However, many in vitro studies have proven that the cardiovascular protective effect of stem cell-based therapy is mainly dependent on the transfer of functional mitochondria rather than the secretion of paracrine factors (12, 15). Since MSCs have limited trans-differentiation abilities into cardiomyocytes or other vascular components in vivo, it seems that the protective effect of MSCs transplantation is also unlikely due to their differentiation capacity (97). However, recent studies have shown that the emergence of many safety issues such as undesired differentiation, pro-arrhythmia, and microcirculation occlusion limit the clinical use of stem cell-based therapy (98–100).

Naked mitochondrial transplantation refers to the transplantation of isolated and uncoated mitochondria to the injured tissues through circulation delivery or local injection (85, 86). The transplantation of naked mitochondria for the treatment of cardiovascular diseases can be traced back to 2009 when McCully's laboratory at Harvard demonstrated for the first time that intramyocardial injection of respiration-competent mitochondria could reduce infarct size and promote postischemic functional recovery in an animal model of heart ischemia-reperfusion injury (85). The optimal dose of mitochondria needed for efficiently protecting against ischemia-reperfusion injury ranges from 2 × 10⁵ to 2 × 10⁸ per gram wet weight (62, 71, 73, 87). Autologous mitochondria isolated from healthy cardiac or skeletal muscle are the dominant sources of donor mitochondria (85, 87, 101). The uptake of autologous mitochondria by cardiomyocytes is usually within minutes via internalization (62, 73). Exogenous mitochondria can also enter into cardiomyocytes without being cleared by lysosomes or autophagosomes. However, it takes more than 8 h for cardiomyocytes to engulf the xenogeneic mitochondria (73). In 2017, the first clinical use of mitochondrial transplantation was performed in five pediatric patients who suffered from cardiac ischemia-reperfusion injury. Four in five patients who accepted intramyocardial injection of autologous mitochondria have shown an improvement in ventricular function without any short-term side effects (such as arrhythmia, mitochondrial autoimmune response, and intramyocardial hematoma) (101).

Although intramyocardial injection of mitochondria has been proven efficient and safe in the treatment of myocardial ischemic disease, there are still a few limitations. For example, only a small number of donor mitochondria is allowed to inject within the myocardium per shot, and the percentage of mitochondrial internalization at each injection site is also pretty low (about 3–7%) (71, 73, 87). So multiple injections are required to ensure the extensive distribution of donor mitochondria throughout the ischemic heart, which increases the difficulty of operation. In addition, the need for thoracotomy prior to the intramyocardial injection may also be a huge obstacle limiting the clinical application of mitochondrial transplantation for potential patients (101).

Since 2016, researchers have begun to test the feasibility of intracoronary delivery as an alternative mitochondrial transplantation method (88). Both autologous and exogenous mitochondria can rapidly spread throughout the whole heart within 10 min of coronary perfusion rather than occur in clusters as found in intramyocardial injection. Furthermore, intracoronary delivery also results in a higher mitochondrial internalization into cardiomyocytes (~23%) than intramyocardial injection (88). In animal models of regional ischemia-reperfusion injury, both preischemic or postischemic intracoronary injection of autologous mitochondria can decrease infarct size, enhance coronary blood flow, and increase cardiac function (88–91). Intracoronary injection of healthy mitochondria also has powerful cardiac protection against globally ischemic injury of donor hearts and global ischemia/reperfusion injury of diabetic heart (88, 92–94). No signs of microvascular obstruction and cardiac arrhythmia are observed after intracoronary injection of mitochondria (91). The safety and efficacy of intracoronary delivery of mitochondria make it a promising treatment for myocardial infarction through percutaneous coronary intervention.

Considering the clinical use of mitochondrial transplantation, intravenous injection of mitochondria might be more feasible than intramyocardial and intracoronary administration. Intravenous delivery of mitochondria has been used as a promising therapeutic method for fatty liver and Parkinson's disease (102, 103). Recent studies have shown that intravenous injection of viable exogenous mitochondria for 3 weeks can improve right ventricular function in an animal model of pulmonary hypertension (86). After systemic administration, the mitochondria are observed in various tissues, including heart (102, 103). The stability of naked mitochondria in the bloodstream and tissue-specific delivery may be the key factors for successful therapy in ischemic or non-ischemic cardiovascular disorders.

Besides ischemic heart injury, recent studies have also found that mitochondrial transplantation can delay the progression of right heart failure (86, 95) and improve myocardium metabolism of offspring born to diabetic mothers (61). Although studies have confirmed the efficiency of naked mitochondrial transplantation, how to obtain high yields of good-quality mitochondria is still a challenge. Enhancing the efficiency of mitochondrial internalization into target cells and maintaining the viability of transferred mitochondria are also crucial problems to assure the efficient clinical application of naked mitochondria transplantation.

Another method of cell-free therapy is mitochondria-rich EVs transplantation. EVs have been recognized as a powerful platform for mitochondrial delivery (40, 104, 105). MSCs from different tissues are the main sources of EVs (106). Although many studies have proven that MSCs-derived EVs can serve as a potential therapy for the treatment of cardiovascular disease, whether the protective effect is mainly dependent on their mitochondria cargo is unclear (107). In 2021, Ikeda's laboratory at Stanford isolated some mitochondria-rich EVs from human-induced pluripotent stem cell-derived cardiomyocytes. The diameter of these EVs ranges from 98 to 677 nm (40). The outer lipid bilayer of EVs usually serves as a security guard that keeps the mitochondria maintaining their morphological and functional integrity. The mitochondria encapsulated within EVs are found more stable than naked mitochondria under extracellular environmental stress, such as calcium overload and oxidative stress (40). In vitro and in vivo studies have demonstrated that transplantation of mitochondria-rich EVs can restore intracellular bioenergetics, prevent post-ischemic left ventricular remodeling, and improve myocardial contractility (40). Since the diameters of MSC-derived EVs are usually <10 μm (108), intravenous or intracoronary injection of EVs has no risk of microvascular obstruction. The intramyocardial injection of mitochondria-rich EVs into the peri-infarct region does not induce cardiac arrhythmia (40), which supports the opinion of Adamiak and coworkers that MSC-derived EVs are safer than MSCs (109). The cargo content of EVs mainly includes mitochondria and their components, nucleic acids, lipids, and proteins, which can be altered according to their cellular origins and isolation methods (107). The complex composition of different EVs makes them have distinctive mechanisms and effects on various diseases. Ikeda and coworkers have found that the beneficial effect of mitochondria-rich EVs transplantation on myocardial ischemia-reperfusion injury was not only due to mitochondria cargo but also due to non-mitochondrial cargo (40). In order to reduce the heterogeneity of EVs and guarantee the therapeutic effect, it is necessary to set up a standardized EV isolation protocol. Meanwhile, how to improve the targeting specificity of EVs is also an issue needed to be further investigated (110).

In conclusion, many in vivo studies have proven the effectiveness of mitochondrial transplantation in the treatment of different cardiovascular diseases. Ischemia/reperfusion injury is one of the most common diseases in the cardiovascular system. Researchers delivered mitochondria to rescue damaged cardiac tissue through different administrational routes in various species (14, 40, 69, 73). Comfortingly, all of these studies have confirmed the significant improvement of cardiac function after mitochondrial transplantation. In a clinical trial, autologous mitochondria from skeletal muscle were injected into the damaged cardiomyocytes of pediatric patients suffering ischemia/reperfusion injury, which effectively promote the recovery of postischemic myocardium without adverse short-term complications (101). Functional mitochondrial delivery has been used for some other cardiovascular diseases (including heart failure, anthracycline-induced cardiomyopathy, pregestational diabetes-induced cardiac malfunction, heart transplantation) and to some extent improve the prognosis (15, 61, 70, 94).