Although numerous studies have revealed several routes of mitochondrial release/transfer, including TNTs and MVs, the specific molecular mechanisms remain to be elucidated. In this article, we have summarized the current understanding of the molecular mechanisms involved in the mitochondrial release and/or transfer from cells (Table 1).TNTs, a unique intercellular membrane structure, were first reported in rat pheochromocytoma PC12 cells [70]. TNTs have a diameter of 50–200 nm, which allows the transport of cell membranes, plasma, and organelles, such as mitochondria, between cells through intercellular connected tubes (Figure 2A). A variety of stimuli that cause mitochondrial damage can induce the formation of TNTs. Mechanistically, p53 is considered to be an important upstream regulator in the process of TNT formation [71]. Activated p53 triggers the epidermal growth factor receptor activation, followed by the upregulation of the downstream Akt-phosphatidylinositol-3 kinase-mTOR pathway, leading to the high expression of M-sec (TNFαIP2). M-sec is one of the key regulators in the formation of TNTs [72], which is responsible for mediating the assembly of exocyst complexes by interacting with RalA to induce F-actin polymerization. CD38, a transmembrane glycoprotein known as cyclic ADP ribose hydrolase, is expected to be a key regulator of TNT-mediated mitochondrial transfer. In the co-culture model of primary multiple myeloma cells and BMSC, TNT-mediated mitochondrial transfer to multiple myeloma cells is inhibited by the shRNA-mediated knockdown of CD38 [49]. Wang et al. demonstrated that enhanced CD38 expression caused by the activation of extracellular regulated protein kinases 1/2 (ERK1/2) promoted mitochondrial transfer from astrocytes [73]. The intercellular concentration gradient of the S100 calcium-binding protein A4 (S100A4) is known to be closely associated with the process of TNT elongation. Namely, a previous report demonstrated that TNTs expand from the donor cells with lower concentrations of S100A4 to receptor cells with higher concentrations [74]. In addition, Cdc42 has been suggested to be involved in the prolongation of TNTs [72]. Gap junction protein connexin 43 (CX43) mediates mitochondrial transfer between bone marrow-derived stem cells and LPS-damaged alveolar epithelial cells [32], possibly through the promotion of TNT formation [33]. On the other hand, He et al. demonstrated that CX43 is absent in the TNTs between neonatal rat CMs and cardiac fibroblasts (CFs) [43], suggesting that further studies are needed to determine the specific role of CX43 in the formation of TNTs. Recent studies have revealed that Miro1, a mitochondrial outer membrane protein, plays an important role in regulating the intercellular movement of mitochondria along TNTs [36,37,68,69]. As Miro1 is a type of calcium-sensitive Rho-GTPase [75], it can be bound to TRAK1/2, which results in the mediation of the movement of mitochondria by recruiting a series of kinesins. In addition, Miro1 interacts with mitofusins (Mfn1 and Mfn2), thereby facilitating the transport of mitochondria along TNTs [76]. Shen et al. demonstrated that mitochondrial transport along TNTs is mediated by KIF5B, a member of the kinesin superfamily, in the co-culture system of neonatal rat CMs and CFs [44].EV is another important transporter of mitochondria (Figure 2B). EVs are membranous structures derived from the endosomal system or shed from the plasma membrane, including exosomes and microvesicles, which differ in size and origin [77]. Previous studies have shown that cells can secrete EVs with mitochondrial components. Although the molecular mechanism needs to be further elucidated, a potential link between the mitochondria-containing EV and the mitochondrial-derived vesicle (MDV), a unique vesicle capable of transporting mitochondria to other organelles has been revealed [16,24,78]. MDVs are generated through the selective packing of mitochondrial content and transported to lysosomes to be degraded early on in mitochondrial damage, causing them to be considered an early response of mitochondrial quality control [79]. A recent study demonstrated that the packaging of mitochondrial proteins into EVs requires the expression of Snx9 and OPA1, while PRKN facilitates the delivery of MDVs containing damaged mitochondrial components to lysosomes rather than EVs, thus preventing the release of mitochondrial DAMPs [78]. Consistently, the inhibition of lysosomal activity promotes the release of mitochondria containing EVs in adipocytes [16,24]. However, according to previous studies, oxidatively damaged mitochondrial proteins are enriched in both MDVs and EVs isolated from stressed adipocytes, suggesting some dysfunction in the selective packaging process.In addition to participating in TNT-mediated mitochondrial transfer, CD38 is involved in the extracellular release of mitochondrial particles by astrocytes [39]. As mentioned above, astrocytes promote the survival of neurons by producing mitochondrial EVs after a stroke, while the inhibition of CD38 was shown to significantly decrease the number of mitochondria containing EVs, suggesting that a CD38-dependent pathway is required for the ejection of these EVs.