The Functions, Methods, and Mobility of Mitochondrial Transfer Between Cells

Transfer via TNTs

Tunneling nanotubes (TNTs) are a type of long membrane structure of 100–800 nm in width and 100 μm in length. Based on F-actin as their framework, TNTs are wrapped by a phospholipid bilayer extending from the cell membrane, connecting the cytoplasm of two cells (57). TNTs allow bidirectional and unidirectional substance transport, including a variety of small molecules, proteins, organelles, and even virus particles (58–60). The formation of TNTs has three stages: Formation of a membrane protrusion, elongation of the membrane protrusion, and fusion of the membrane protrusion with the target cell membrane. Transformation of the membrane protrusion is the direct result of actin cytoskeleton remodeling ( Figure 2 ). TNTs exist extensively in a variety of physiological and pathological cells, such as kidney cells (61), PC12 pheochromocytoma cells (62), astrocytes (10), and myocardial cells (17, 63).

Studies have demonstrated that TNTs act as a channel for mitochondrial transfer between various cells ( Figure 1A ; Table 1 ). Under condition of oxidative stress, intracellular p53 expression is upregulated and the AKT-PI3K-mTOR signaling pathway is activated, leading to the formation of TNTs from stressed cells to non-stressed cells, and mediating the transcellular transport of four organelles (ER, Golgi, endosome and mitochondria), including mitochondria (10). Interestingly, an experiment conducted two years later contradicted this view, concluding that TNTs did not depend on p53 activation (64). In the bone marrow microenvironment of multiple myeloma, TNT-mediated transcellular transfer of mitochondria was found to rely on the expression of CD38 (55). Although the upstream activation signals may be distinct, there should be a common downstream mechanism guiding the formation of TNTs.

A variety of molecules are involved in the process of TNT formation, including M−Sec (also known as TNF alpha induced protein 2), the exocyst complex, small GTPases, and leukocyte specific transcript 1 (LST1). The exocyst complex, an octameric protein complex, mediates the fusion of secretory vesicles derived from the Golgi body with the plasma membrane during exocytosis of yeast, mammals, and other eukaryotes (65–68). In recent years, it has been found that the exocyst complex also plays a significant role in various processes involved in the morphological transformation of the cell membrane, such as neurites, cilia, filopodia, and EVs (67, 69, 70). The components of the exocyst complex interact with several small GTPases (Rho1, Rho3, Cdc42, and RalA) to promote actin cytoskeletal remodeling, which is closely related to the function of the exocyst complex (69, 71, 72). The exocyst complex also participates in the formation of TNTs, and this process is controlled by M-Sec, a protein that has been proven to control the formation of TNTs (73, 74).

M-Sec can promote the assembly of the exocyst complex and interact with it, as well as with RalA (or Cdc42), leading to actin cytoskeleton remodeling, which is the key step of TNT formation (75). Studies have shown that in the initial stage of TNT formation, the N-terminal polybasic region of M-Sec directly integrates with phosphatidylinositol (4, 5)-diphosphate, so that M-Sec is fixed on the cell membrane. In addition, the interaction between M-Sec and RalA requires a positively charged surface at the C-terminal of M-Sec (76). Filopodia transformation is also a possible mechanism for TNT formation (62), as Cdc42, whose activated protein Ral binding protein 1 (RalBP1) regulates actin remodeling during filopodia formation (77), also participates in TNT formation; however, Cdc42 seems to be only related to the prolongation stage of TNTs (75).

The MHC class III protein, LST1, is a transmembrane protein that can recruit RalA to the submembrane region and accelerates interaction between RalA and the exocyst complex. Meanwhile, LST1 recruits the actin cross-linked protein filamin. LST1, along with M-Sec, RalA, and the exocyst complex, together promote the remodeling and cross-linking of actin filaments, leading to cell membrane protrusion and fusion, ultimately leading to TNT formation (78) ( Figure 3 ). However, the mechanism of the fusion between the membrane protrusion and the target cells remains unclear.

The transport and recycling of vesicles between biomembranes is also thought to be related to TNTs, as both the small GTPase Rab11a, which controls this process, and the downstream Rab8a, are involved in the formation of TNTs (79). In bladder cancer cells, inhibition of RalGPS2, the guanine nucleotide exchange factor (GEF) of RalA, also leads to a large reduction in TNT formation; moreover, RalGPS2 also interacts with LST1 and RalA and participates in the formation of multi-molecular complexes (80). In neurons, the unconventional molecular motor myosin X is also necessary for the formation of TNTs, because TNTs can be developed from filopodia driven by myosin X (81). The concentration gradient of S100 calcium binding protein A (S100A) in nervous tissue determines the extension direction of TNTs. Under oxidative stress, a decrease in S100A levels in neurons leads to a concentration gradient of S100A between astrocytes and neurons. TNTs then extend from neurons with a low concentration S100A and are received by astrocytes with a high concentration S100A (82).

TNT-mediated mitochondrial transfer can act as a survival pathway for cells under stress, for example, saving ischemic damaged cells (83), protecting the alveolar epithelium from injury (49), and repairing tissue (2, 6, 84). TNTs also exist in cancer cells, in which they are related to the survival and drug resistance of cancer cells (41, 85).

Transfer via Extracellular Vesicles

There are two types of EVs: Exosomes (30–150 nm in diameter) and microvesicles (30–1000 nm in diameter) (86). As a form of intercellular communication, EVs can deliver various substances, including small molecules, organelles, and membrane proteins, and thus play a role in mitochondrial transcellular transfer in certain types of cells.

Under conditions of evacuating serum in vitro and stimulating ischemic stroke in vivo, astrocytes were able to release EVs containing mitochondria for phagocytosis and utilization by nearby neurons to reduce nerve damage (25). In a model of allergic airway disease, EVs containing mitochondria derived from airway myeloid-derived regulatory cells (MDRCs) could be ingested by peripheral T cells, resulting in modulation of the bioenergetics of T cells, which is related to immunomodulation and the regulation of inflammatory responses of the disease (33, 87). Retinal ganglion cells can produce cell membrane processes containing axon mitochondria and release vesicles, which are phagocytized and degraded by astrocytes in the optic nerve papilla (27). EVs containing mitochondria were also found in the plasma of patients with melanoma (88).

The NAD+/CD38/cADPR/Ca²⁺ pathway is the key in EV-mediated mitochondrial transcellular transfer. CD38, a transmembrane protein expressed on various cell membranes, can produce cyclic ADP-ribose (cADPR) using co-enzyme nicotinamide adenine dinucleotide (NAD) as a substrate. cADPR is a second messenger that can open the intracellular Ca²⁺ pool. cADPR acts on ryanodine receptors (RyRs) (a calcium ion channel) on the endoplasmic reticulum, which promotes the release of Ca²⁺ stored in the endoplasmic reticulum. Thus, the transient Ca²⁺ concentration in the cytosol increases, leading to a series of downstream responses (89).

Glioma cells under starvation can accumulate a large amount of NAD+ and release it into the extracellular region. Through the NAD+/CD38/cADPR/Ca²⁺ pathway, the increasing calcium in the cytosol promotes actin cytoskeleton remodeling, leading to cell membrane invagination, thus completing the endocytosis of EVs (28) ( Figure 1B ). Inhibition of endocytosis resulted in a reduction of mitochondrial transfer from bone marrow mesenchymal stem cells (BMMSCs) to wounded lung epithelial cells (24). Mitochondrial transcellular transfer by EVs between astrocytes and neurons has also been shown to depend on the NAD+/CD38/cADPR/Ca²⁺ pathway (25, 26). This pathway controls not only the endocytosis stage of EVs, but also seems to be involved in the exocytosis stages, because the amount of EVs is significantly reduced, not increased, after small interfering RNA (siRNA) inhibition of CD38 expression (25).

The exocyst complex, the critical molecule of actin skeleton remodeling and membrane protrusions formation, as mentioned in section 2.1.2, might also participate in the downstream mechanism of the CD38-mediated EV pathway, because inhibition of the composition of the exocyst complex caused a remarkable reduction in EV formation (70). In addition, the calcium signal that promotes EV formation also participates in TNT formation mediated by the exocyst complex (75). Therefore, we believe that: The exocyst complex can be assembled and activated by the NAD+/CD38/cADPR/Ca²⁺ pathway and interacts with certain types of small GTPase and unknown regulatory factors, leading to the release or endocytosis, or both, of EVs.

Transfer via Gap Junctions

The GJ protein connexin is a widely distributed in various cells. Six rod-shaped connexins form a connexon or hemichannel across the plasma membrane, with a nanoscale hydrophilic channel in the middle. The hemichannel, as a precursor structure of GJs, can also exist independently as a channel connecting the inside and outside of the cell (90, 91). A head-to-head connection between two hemichannels forms a structure called a gap junction channel (GJC) that connects the cytoplasm of two cells. GJCs allow the exchange of ions (especially calcium ions) or small molecules between the two cells (92), facilitating the exchange of chemical or electrical signal between the two cells. GJCs can provide nutrient substances to surrounding cells, protect cells from injury (93), and are involved in clathrin-dependent EV endocytosis (94, 95).

Cx43 is a type of connexin that has been proven to be involved in ischemia/reperfusion injury of myocardial and cerebral tissue, and can also protect myocardial and cerebral tissue from ischemia/reperfusion injury under certain conditions (96). Cx43 seems to have an important association with mitochondrial function. Cx43 hemichannels on mitochondria participate in maintaining mitochondrial calcium homeostasis and can lead to mitochondria damage and cell apoptosis by passing Ca²⁺ (97, 98). Cx43 also participates in the transcellular transfer of mitochondria. In a model of LPS-induced acute lung injury, bone marrow stromal cells (BMSCs) delivered mitochondria to the injured alveolar epithelium. This process relies on Ca²⁺ exchange between the two cells through CX43-GJCs (49).

Cx43 has been proven to be an important regulator involved in the formation and function of TNTs. In a model of airway allergic inflammation, the formation of TNTs connecting induced pluripotent stem cell (IPSC)-MSCs and airway epithelial cells depends on Cx43-GJCs, while downregulating Cx43 expression significantly inhibited the metastatic transfer between the two groups of cells, and simultaneously blocked the protective effect of IPSC-MSCs on lung tissues (12). Cx43 and its related signaling pathways in breast cancer cells, such as Rho associated coiled-coil containing protein kinase (ROCK), protein kinase A (PKA), focal adhesion kinase (FAK) and p38, have novel non-canonical roles of regulating TNT formation (99). Knocking out the CX43 genes of human trabecular meshwork cells also led to a significant reduction in TNTs (100). Cx43-GJCs might play an important role in the formation of TNTs. Cx43-GJCs participate in the coordination mechanism of the two cells. This is likely to be regulated by intercellular ion exchange, especially Ca²⁺, as calcium signal exchange exists between the two cells connected by TNTs (75). Moreover, functional GJs have been found at the end of the membrane extension of TNTs (101), which also supports this hypothesis.

In addition, Cx43-GJCs participate in the intercellular exchange of ROS (102), and there may even be direct mitochondrial transfer via Cx43 GJCs (9), which might also be a mechanism for TNT formation and the intercellular transport of mitochondria ( Figure 1C ).

Transfer via Other Routes

TNTs, EVs and Cx43-GJCs are the main routes that mediate transcellular transfer of mitochondria in most studies. However, there are also some other routes, such as mitochondrial extrusion and cytoplasmic fusion.

Naked mitochondria or mitochondrial components can also be extruded and internalized without carrier, which are also called exocytosis and endocytosis (5). Cytoplasmic vacuoles derived from plasma membrane engulf mitochondria and then fuse with cytomembrane to extrude the damaged mitochondria out under the condition of tumor necrosis factor-α (TNFα)-induced apoptosis in a actin and tubulin cytoskeletons-dependent manner (103). Endocytosis of naked mitochondrial was also detected in chloramphenicol (CAP)- and efrapeptin (EF)-sensitive mammalian cells, which internalized free mitochondria isolated from CAP- and EF-resistant fibroblasts in a coincubation medium (104). By endocytosis, vitality and bioenergetics of recipient cells restore (105).

Cytoplasmic fusion is a common phenomenon in which the membrane of two or more cells fuse together and the organelles are shared, in which injury and inflammation may induce this process (106). Cell fusion can regulate the potential of stem cells, which plays an important role in regeneration and oncogenesis (107). Mitochondrial transfer may participate this process (108). However, the extent to which mitochondrial transfer promotes the potential of stem cells is still unknown.

Movement of Mitochondria Based on Microtubules and Microfilaments

The mitochondrial transport complex consists of mitochondrial Rho small GTPase (MIRO) and motor proteins based on microtubules or microfilaments. In certain cases, adaptors such as TRAK are also needed ( Figure 4 ). MIRO, a member of the Ras superfamily, is located on the mitochondrial outer membrane, attached by its C-terminal hydrophobic transmembrane domain (109, 110), and binds with motor proteins directly or indirectly (via adaptor proteins) (43). Transport complex allows mitochondria immobilized on cytoskeleton and provides the power driving the movement of mitochondria.

MIRO1 and MIRO2 are two subtypes of MIRO in humans. The combination of drosophila MIRO (dMIRO) and Milton mediates the transport of mitochondria based on microtubules in Drosophila (111). Inhibition of dMIRO expression leads to mitochondrial transport disorder in the dendrites and axons of Drosophila neurons (112, 113). As the mitochondrial receptor, MIRO forms the transport complex with Milton and Kinesin to mediate the anterograde transport of mitochondria in axons, in which Milton act as a motor adaptor and Kinesin is a type of motor protein based on microtubules (114). Trafficking Kinesin Protein 1 (TRAK1) and TRAK2, the human Milton orthologs, have also been proven to play an important role in the process of mitochondrial transport by binding with MIRO to form MIRO/TRAK/Kinesin complex (115, 116) ( Figure 4 ). Indeed, not only kinesin, but also other microtubule motor proteins, such as kinesin superfamily KIF5 and dynein can also interact with TRAK and MIRO to form different mitochondrial transport complexes based on microtubules (43, 117, 118). MIRO contains two GTPase domains (located at the N-terminus and C-terminus, respectively) and two EF-hand Ca2+-binding domains (109, 110). Based on the two EF-hand Ca²⁺-binding domains, Ca²⁺ can change the configuration of the MIRO/adaptor/motor complex to regulate mitochondrial fixation on microtubules. When the cytosolic Ca²⁺ concentration increases, the MIRO/TRAK/motor complex is separated from the microtubule, causing the mitochondria fall off the microtubule (43, 119, 120) ( Figure 4 ).

Microtubules are responsible for the long-distance transport of mitochondria, such as the transport in neuron axon, while the short-distance transport mainly depends on the actin microfilament system (43). Mitochondrial transport based on microfilaments is mediated by a transport complex containing actin motor—myosin. Currently known myosin family members that adapt to mitochondria include myosin II, myosin V, myosin VI, and myosin XIX.

Myosin II promotes mitochondrial contraction by inducing deformation of the microfilament skeleton during mitochondrial division (121). Myosin V and VI exert a negative regulatory effect on the microtubule-based axon transport of mitochondria. Downregulating the expression of myosin V and VI can significantly accelerate the speed of mitochondrial transport, which is possibly caused by myosin V and VI mediating the actin-based docking of mitochondria (122). By slowing down the speed of mitochondria and anchoring mitochondria to the cytoskeleton, mitochondrial docking maintains the needed quantity of stationary mitochondria in the regions that need energy and Ca2+-buffering capacity (43). Thus, these regions obtain sufficient ATP. Although the docking receptor has not been identified, it is established that myosin mediates mitochondrial docking (43). This process might also assist the transcellular transport of mitochondria, because the actin-based docking of mitochondria can slow down the speed of mitochondrial movement, which might be a transitional stage in which mitochondria move from the cytoplasm to TNTs or EVs via actin microfilaments ( Figure 4 ).

Myosin XIX Drives the Movement of Mitochondria in TNTs

Myosin XIX is a high-duty ratio actin motor with strong affinity for mitochondria (123, 124), which is fixed to the mitochondrial outer membrane via a 30–45-residue motif (125), while MIRO can recruit and stabilize myosin XIX to the mitochondrial outer membrane as the mitochondrial motor adaptor protein, thus driving the actin-based movement of mitochondria by interacting with myosin XIX (126, 127). This process is regulated by the local ATP/ADP ratio (128). We hereby propose a hypothesis: The movement of mitochondria in TNTs is driven by the MIRO/myosin XIX transport complex, of which myosin XIX is the motor protein ( Figure 4 ).

This hypothesis is based on the following evidence: a. Under the induction of starvation culture or ROS, cells produce abundant filopodia, in which mitochondria exist, and the process of mitochondrial transfer to the top of filopodia has been proved to be mediated by myosin XIX (125, 128, 129); b. As mentioned above, filopodia have many similarities with TNTs: Filopodia and TNTs are both generated by membrane protrusions under oxidative stress and use actin filaments as skeletons. In certain conditions, filopodia can also be converted into TNTs (62); c. Evidence shows that myosin drives TNT-based transcellular material transport (130); and MIRO also mediates TNT-based transcellular mitochondrial transport from mesenchymal stem cells to epithelial cells (21, 131) and nerve cells (132, 133). This process depends on a unique residue (a class specific tryptophan) of the myosin XIX motor domain (129). However, direct evidence of myosin XIX’s presence in TNTs is still needed. Besides, myosin XIX controls actin-based mitochondrial movement during cell mitosis to ensure the symmetrical distribution of mitochondria to daughter cells (134).

Coordination of MIRO in Transcellular Mitochondrial Transport

MIRO, the connecting component of the transport complex, interacts with microtubules and actin motor proteins such as kinesin and myosin, adaptor proteins TRAK1 and TRAK 2, PTEN induced kinase 1 (PINK1) (regulates mitochondrial autophagy) (135), hypoxia up-regulated mitochondrial movement regulator protein (HUMMR) (involved in mitochondrial movement in neurons under hypoxia) (136), and abundant types of cytoskeleton binding proteins. MIRO can also be used as part of the ER-mitochondria encounter structure (ERMES) complex to fix mitochondria on the endoplasmic reticulum ( Figure 4 ) (137, 138). By connecting mitochondria with different transport complexes, MIRO regulates mitochondrial movement based on microtubules and actin, and the transformation between the two forms of mitochondrial movement (126, 127). MIRO not only modulates the intracellular movement of mitochondria, but also mediates the intercellular movement of mitochondria in TNTs.

The evidence above suggests that MIRO is the key protein that modulates intra- and intercellular mitochondria distribution. MIRO might also mediate the transitional stage during which mitochondria move from the cytoplasm to the transcellular transfer structures, such as TNTs or EVs. Mitochondria are separated from the microtubule binding site, and then anchored on an actin filament. After that, mitochondria move to enter TNTs driven by myosin, where mitochondria are anchored on the actin filament of TNTs for subsequent intercellular transport. All these processes are mediated by the transport complexes composed of MIRO and its adaptor proteins and motors ( Figure 4 ). Similarly, MIRO might also play an important role in the process of mitochondrial entry into EVs.