Horizontal mitochondrial transfer as a novel bioenergetic tool for mesenchymal stromal/stem cells: molecular mechanisms and therapeutic potential in a variety of diseases
Intercellular mitochondrial transfer (MT) is a newly discovered form of cell-to-cell signalling involving the active incorporation of healthy mitochondria into stressed/injured recipient cells, contributing to the restoration of bioenergetic profile and cell viability, reduction of inflammatory processes and normalisation of calcium dynamics. Recent evidence has shown that MT can occur through multiple cellular structures and mechanisms: tunneling nanotubes (TNTs), via gap junctions (GJs), mediated by extracellular vesicles (EVs) and other mechanisms (cell fusion, mitochondrial extrusion and migrasome-mediated mitocytosis) and in different contexts, such as under physiological (tissue homeostasis and stemness maintenance) and pathological conditions (hypoxia, inflammation and cancer). As Mesenchimal Stromal/ Stem Cells (MSC)-mediated MT has emerged as a critical regulatory and restorative mechanism for cell and tissue regeneration and damage repair in recent years, its potential in stem cell therapy has received increasing attention. In particular, the potential therapeutic role of MSCs has been reported in several articles, suggesting that MSCs can enhance tissue repair after injury via MT and membrane vesicle release. For these reasons, in this review, we will discuss the different mechanisms of MSCs-mediated MT and therapeutic effects on different diseases such as neuronal, ischaemic, vascular and pulmonary diseases. Therefore, understanding the molecular and cellular mechanisms of MT and demonstrating its efficacy could be an important milestone that lays the foundation for future clinical trials.
Keywords: Mitochondria, Horizontal mitochondrial transfer, Mesenchymal Stromal/Stem cells, Tunnelling nanotubes, Extracellular vesicles, Neuronal diseases, Ischemic vascular diseases
Mitochondria are highly dynamic and multifunctional compartments that play a pivotal role in oxidative bioenergetic metabolism and in maintaining cellular homeostasis. In addition to energy production, mitochondria perform many other key functions regulating fatty acid β-oxidation, iron homeostasis, Ca2+ metabolism, heme and steroid hormones biosynthesis, innate immunity, redox homeostasis, and cellular waste management [1–3]. They are key sensors of multiple types of cellular stress [4–6] and crucial determinants in cell survival and death [7]. To fulfill its multiple tasks, mitochondrial network is dynamically redesigned by distinct processes (collectively referred to as mitochondrial quality control, MQC), including fission and fusion events, intracellular mitochondrial movement, selective removal of damaged mitochondria through mitophagy [8–10], and mitochondrial biogenesis, that orchestrate the overall shape, size, distribution, and connectivity of mitochondria. As signaling platforms, mitochondria communicate extensively with other cellular compartments [11] and can operate outside their intracellular confines exchanging information between cells, and even across organ systems [12]. Interestingly, different forms of circulating mitochondria (e.g., wrapped Mitos, mitochondria transported in vesicles or in platelets; free Mitos, mitochondria without a protective vesicle membrane) with distinct effects on immune cells have been found in both blood and cerebrospinal fluid (CSF) [8, 13–16].
It is not surprising that the concomitant increase of mitochondrial DNA (mtDNA) mutations and reactive oxygen species (ROS) generation, during the aging process, exacerbate mitochondrial dysfunction and dysregulates MQC, contributing to the pathogenesis of multiple age-associated diseases [17, 18], recently termed as non‐communicable diseases (NCDs) [19].
In this view, mitochondria-targeted therapeutic approaches have been explored over the past decade as potential treatments for tissue revitalization and homeostasis.
The phenomenon of intercellular transport of mitochondria between mammalian cells, also known as horizontal transfer of mitochondria, has recently attracted a renewed attention from the scientific community, representing an intriguing reparative strategy [20, 21]. Intercellular mitochondrial transfer (MT) is a novel form of cell-to-cell signalling involving the active incorporation of healthy mitochondria into stressed/injured recipient cells [22], contributing to restore the bioenergetic profile (ATP and mitochondrial membrane potential) and cell viability, to increase the mtDNA content, as well as to reduce inflammatory processes and normalize calcium dynamics [23].
MT can occur via multiple distinct molecular mechanisms, including tunnelling nanotubes (TNTs), extracellular vesicles (EVs), gap junction channels (GJCs), and other non-traditional routes, such as cell fusion and mitochondrial extrusion [24, 25] (Fig. 1).
The mitochondrial donation from mesenchymal stromal/stem cells (MSCs) has gained increasing attention in recent years [26], suggesting its potential in stem cell therapy [21]. MSCs-mediated MT is emerging as a critical regulatory mechanism for cell and tissue regeneration, and damage repair, where a remarkable restoration of cellular bioenergetics and a reduction in oxidative stress have been reported [26, 27]. MSCs, as reported by Viswanathan et al., are to be better referred as Mesenchymal Stromal/Stem Cells (MSCs) given the prevalence of stromal cells with respect to the smaller fraction of true "mesenchymal stem cells” (10.1016/j.jcyt.2019.08.002). MSCs have the ability to maintain tissue homeostasis and renewal via regulating functional activity of parenchymal cells, including haematopoietic stem cells (HSCs) [28, 29]. They can be isolated and expanded in vitro from almost all tissues, including bone marrow (BM), dental pulp (DP), umbilical cord (UC), adipose tissue (AD), placenta, and other sources [30–33]. Moreover, they have pleiotropic ability to differentiate into mesodermal lineage cells, including adipocytes, chondrocytes and osteocytes, or ectodermal lineage cells, such as neuronal and neuroglial cells [34]. In physiological conditions, the original microenvironment characteristics of MSCs (intercellular matrix, other cell types, soluble factors and humoral components) dictate their cellular fate [35]. On the other hand, MSCs reciprocally interact with the microenvironment through distinct processes, including immunomodulation and tissue repair. Therefore, MSCs display remarkable tissue regenerative properties given their ability to receive distinct signals from the surrounding tissues (e.g., damage-associated molecular patterns (DAMPs) and ROS) and constantly set up intricate intercellular communication networks with stressed/damaged cells.
Mitochondrial exchange and its beneficial effects were firstly reported by Spees and colleagues [36] in A549 ρ° cells with mtDNA deletion or defects. Specifically, the acquisition of healthy mitochondria in A549 ρ° cells following active interaction with human MSCs resulted in increasing levels of oxygen consumption, membrane potential, and intracellular ATP. Subsequent investigations in different in vitro and in vivo models have revealed the possibility of using mitochondria derived from MSCs for therapeutic purposes (e.g., stroke, lung and acute respiratory disorders, brain injury, muscle sepsis, and cardiac diseases), thus representing a novel strategy to treat many NCDs. Features such as immune privilege, significant migratory capacity to injured sites, fine-tuned redox balance, and low oxidative damage levels, as well as low energy demand (mitochondria in MSCs are quiescent and exhibit low activity level) render MSCs as elective donor cells in delivering functional mitochondria to diseased cells [37–42]. Furthermore, MSCs isolated from different tissue types (BM; AD; DP; UC; Wharton’s jelly, WJ) exhibit distinct bioenergetic signatures that influence their MT capacity, with AD-MSCs and BM-MSCs showing higher MT levels to cardiomyocytes than DP-MSCs and WJ-MSCs [37].
On the other hand, it has also been shown that MSCs-mediated MT plays a critical role in pathological states. In this sense, recent evidence indicates a dark side of MSCs in increasing malignancy of cancer cells, supporting tumor microenvironment and cancer progression, and providing metabolic flexibility and chemotherapy resistance [37, 43–45].
In this review, we highlight the therapeutic potential of MT from MSCs in restoring the bioenergetic metabolism and cell functionality/viability into stressed/injured recipient cells. Therefore, we provide a comprehensive overview of the molecular signals triggering MT, as well as the major routes and mechanisms mediating MT. Finally, recent advances in MSCs regenerative properties through MT process in different NCDs are proposed.
MT has been demonstrated in various contexts under physiological (e.g., in tissue homeostasis and stemness maintenance) and pathological conditions (e.g., hypoxia, inflammation, and cancer). MT is a highly regulated multistep process that requires the spatiotemporal orchestration of many factors, including distinct triggering events/signals, molecular processes governing the formation of transfer machinery, and regulatory elements that modulate mitochondrial transfer speed and duration. In this section we will discuss the current state of knowledge regarding molecular stress signals and modes of MT, as well as signaling pathways regulating the different patterns of mitochondrial exchange.
Depending on different forms and activation states, mitochondria can affect the functions of neighbouring cells performing complex activities as mediators of regenerative and anti-inflammatory effects or trigger factors of inflammatory reactions [8, 46–48]. Multiple stress conditions, including oxygen–glucose deprivation [49, 50] drug-induced oxidative stress [51, 52] and inflammation [53, 54] generate harmed and fragmented extracellular mitochondria that trigger MT between cells. Therefore, the tissue regenerative properties of MSCs via the acquisition of active mitochondria by injured cells may be prompted by stress signals such as dysfunctional mitochondria and DAMPs of mitochondrial origin (mitoDAMPs), including ROS, DNA, cardiolipin, ATP, N-formyl peptides, transcription factor A mitochondria (TFAM), Cytochrome C (Cyt-C), succinate, and Ca2+(Fig. 1).
Consistent with this hypothesis, in an in vivo model of myocardial infarction, mitochondria released from damaged cells operate as potential DAMPs in MSCs, inducing the activation of eme-oxygenase-1 (HO-1) signaling pathway and mitochondrial biogenesis. These events enhance MT and consequently potentiate the rescue ability of MSCs to damaged cells [55]. In acute myeloid leukemia (AML) cells, NADPH oxidase-2 (NOX2)-derived ROS regulates MT from the BM-MSCs to the AML through TNTs [56]. In line with this, the generation of oxidative stress in HSCs triggers MT from BM-MSCs. Specifically, infection by Gram-negative bacteria drives MT from the BM-MSCs to HSCs through a ROS-dependent mechanism involving the activation of p53 and its downstream Akt/PI3K/mTOR pathway, and Connexin 43 (CX43) Gap Junctions [57, 58]. Of note, the p53-dependent activation of Akt/PI3K/mTOR pathway also leads to the overexpression of Tumor Necrosis Factor Alpha Induced Protein 2 (TNFαip2) and the formation of TNTs. As described above, MSCs also promotes the transfer of depolarized mitochondria to macrophages in response to ROS generation [58].
In addition to ROS, the Cyt-C released in stressed cells can also stimulate MT. Therefore, in UV-damaged PC12 cells at early stage of apoptosis loss of Cyt-C from injured mitochondria activates MT to healthy cells through TNTs formation, thereby leading to the recovery of apoptotic PC12 cells [59].
CD38 is a multifunctional transmembrane glycoprotein responsible for the biosynthesis of two calcium-mobilizing second messengers, cyclic ADP-ribose (cADPR) and nicotinic acid adenine dinucleotide phosphate. Interestingly, CD38 may also play a role in promoting MT in different cellular models. Therefore, by a calcium-dependent mechanism involving CD38 and cADPR signalling, astrocytes promote the mitochondrial exchange to nearby neurons via microvesicles (MVs), thereby contributing to activation of neuroprotective and neurorecovery mechanisms after ischemic stroke [60]. In this case, CD38 expression may be related to the excessive release of excitotoxic glutamate from ischemic neuron, suggesting for this neurotransmitter a potential role in promoting astrocytes-mediated MT. A subsequent study further demonstrated that CD38/cADPR signaling and mitochondrial Rho GTPases (Miro) 1 and 2 contribute to the MT between astrocytes, and from neuronal cells into astrocytes [61]. CD38 has been indicated to exert a critical function during bone formation [62]. Very recently, it has been demonstrated that CD38/cADPR signaling plays a crucial role in mediating the differentiation and maturation of osteoblasts and osteoprogenitor cells through stimulating the secretion of mitochondria and mitochondrial-derived EVs [63].
Multiple in vitro and in vivo models have highlighted the elective role played by TNTs in MT. Specifically, MT can occur from MSCs to lung macrophages via TNTs, resulting in higher phagocytic activity of macrophage cells in a model of acute respiratory distress syndrome [64]. In an in vitro simulated ischemia/reperfusion model, MSCs move mitochondria to injured H9c2 cardiomyoblasts via TNTs, protecting cardiac cells against the apoptosis [65]. A bidirectional MT between MSCs and vascular smooth muscle cells via TNTs upregulates MSCs proliferation [66]. Also, MT via TNTs occurs from human BM-MSCs and endothelial cells experiencing chemotherapy stress, resulting in recovery of injured endothelial cells [67]. Mitochondria move between BM-MSCs and myeloma cells via TNTs, as described above [56, 68]. TNTs-mediated mitochondria exchange also form between donor cells other than MSCs, including PC12 cells [59], astrocytes [69], lung epithelial cells [70], cancer cells [71], retinal pigment epithelium [72], and trabecular meshwork cells [73].
According to the initial description, TNTs are open-ended membranous channels directly connecting cytoplasm of cells over long distances in a homotypic and heterotypic fashion. They are non-surface adherent and have a structure of 50–900 nm (with an average of 200 nm) in width and an average length between 20 and 100 mm.
TNTs contain F-actin cytoskeletal filaments that allow the bidirectional and unidirectional exchange between cells of various‐sized cargoes, including small molecules, nucleic acids and proteins (e.g., tau, α-synuclein, and huntingtin), organelles (e.g., vesicles, lysosomes, endoplasmic reticulum (ER), mitochondria, and autophagosomes), and even virus (herpesvirus and SARS-CoV-2) and bacteria [74–78], suggesting their role in in coordinating metabolism and signalling events in a wide-range of physiological processes and pathological conditions. Of note, this transfer function represents a crucial distinctive characteristic between TNTs and filopodia. The membrane structure of TNTs is very heterogeneous and cell type-specific. Therefore, close-ended TNTs (exhibiting gap junctions), that allow transfer of electrical signals, have also been identified in different in vitro and in vivo conditions [72, 79–81]. In this regard, 3 types of close-ended TNTs have been observed, such as “hand-shake”, “invaginated”, and “resting” [82].
In addition, two types of open-ended TNTs according to the different distribution of cytoskeletal elements have been described: “thin” TNTs (< 700 nm in diameter), containing only F-Actin, and “thick” TNTs (> 700 nm in diameter), with F-Actin, microtubule, and intermediate filaments [83]. In neuronal and stromal cell lines, cryo-correlative light and electron microscopy (cryo-CLEM) coupled with tomography have also revealed individual TNTs (iTNTs), a bundle of small open-ended tubes (up of two to 11) that run parallel, and sometimes braided together, along the entire length of the TNTs [82, 84]. These structures exhibit N-Cadherin as molecular linker connecting adjacent iTNTs. In addition to confer mechanical stability, it acts as guidance during growing iTNTs. Differently from single TNT, iTNTs allow for a bidirectional transfer of cargoes, including vesicles and mitochondria.
Thus far, we have learned that the formation of TNTs can be generated via two different mechanisms that can occur simultaneously, or change dependently on the environmental conditions, though the underlying molecular process still remains under active investigation. These mechanisms include the “cell dislodgement”, wherein cells initially in contact leave behind a tubular connection when they move apart, and the “protrusion-elongation mechanism” where the cell extends a filopodia-like protrusion to another cell located at some distance. In this regard, the main mechanism responsible for TNTs biogenesis seems to be cell dislodgement, as reported by live imaging analyses in different cell types [85, 86]. However, the protrusion-elongation mechanism is typical of post-mitotic cells with low migratory phenotype, such as neurons and epithelial cells [87].
As mentioned above, the biogenesis of TNTs involves stress-signaling (e.g., p53 and MAP kinase) and pro-survival (e.g., EGFR, Akt, ROCK, PAK, MAP/ERK, PI3K or mTOR) pathways. Downstream effectors of these signaling pathways are proteins related to both membrane recycling and cytoskeletal remodeling.
Interestingly, the Rab family of small GTPases regulates many steps of membrane trafficking and also participate in actin cytoskeleton remodeling [88, 89]. Therefore, transport and recycling of vesicles regulated by the small GTPases Rab11a and Rab8a promote TNTs formation in different cell systems [88, 90]. Specifically, VAMP3 (vesicle-associated membrane protein 3, often called v-SNARES) operates downstream of Rab8a to regulate TNTs biogenesis [88]. Depletion of Ras association domain-containing protein 1A, a master regulator of cellular homeostasis and cytoskeleton, results in Rab11 accumulation and the subsequent release of exosome, thereby leading to TNTs formation [91]. Furthermore, Rab35 and its downstream effectors, such as ACAP2, ARF6-GDP, and EHD1 operate in a cascade mechanism to promote TNTs biogenesis in neuronal cells [92].
F-actin remodeling processes are responsible for initiating TNT protrusion by acting against the tensile strength of the plasma membrane, thus allowing it to deform and generate forces for tube growth. In this regard, distinct mechanisms involving specific sets of actin regulatory proteins are implicated, such as M-Sec (also known as TNFAIP2) [93–95] the exocyst complex, leukocyte specific transcript 1 (LST1) [96], the unconventional Myosin X (Myo10) [97], the actin bundler epidermal growth receptor substrate 8 (Eps8) and insulin receptor tyrosine kinase substrate protein 53 kDa (IRSp53), and small GTPases (e.g., Miro1/2, Rac1, Cdc42 and RalA) [98].
The cytosolic protein M-Sec acts as a key regulator of TNTs biogenesis through interaction with the small GTPase Ral in the mouse macrophage-like cell line RAW264.7 [93]. In this context, the transmembrane MHC class III protein LST1 may operate as a membrane scaffold for the generation of multi-molecular complex that controls the formation of TNTs. Thus, LST1 promotes TNTs formation by recruiting RalA and the actin-crosslinking protein filamin to the plasma membrane. It also stimulates the binding of RalA with two components of exocyst complex, Sec5 and Exo84, thereby leading to actin cytoskeletal remodeling and membrane protrusion. Meanwhile, the interaction of LST1 with M-Sec, myosin, and myoferlin may be involved in the process of mitochondrial anchoring and transfer [96]. In this regard, the recruitment of cytosolic M-Sec to the plasma membrane during the initial phase of TNT formation may occur through its direct binding to phosphatidylinositol (4,5)-bisphosphate [PI(4,5)P2] or phosphatidylinositol (3,4,5)-trisphosphate [PI(3,4,5)P3] [94]. Nucleolin, an RNA-binding protein, is essential for TNT formation in various types of mammalian cells [99]. Specifically, the binding of nucleolin to 14-3-3ζ mRNA modulates phospho-cofilin levels, leading to its inactivation, F-actin polymerization, and TNTs biogenesis.
In addition to the Rab35 signaling, Wingles-related integration site (Wnt)/Ca2+ pathway, an intracellular cascade that is implicated in actin cytoskeleton remodeling, promotes TNTs biogenesis and stability in CAD (mouse catecholaminergic neuronal cell line) cells and primary neurons through the modulation of the interaction between the β isoform of Ca2+/calmodulin‐dependent protein kinase II and actin [100].
The I-Bin/Amphiphysin/Rvs (BAR)-domain protein IRSp53 is an essential spatio-temporal coordinator of plasma membrane protrusions (through promoting negative membrane curvature) that couples Rho-GTPase signaling to cytoskeleton remodeling and membrane dynamics. As a signaling platform, IRSp53 operates under the control of activatory (Cdc42, Eps8) or inhibitory (14-3-3) inputs as well as downstream effectors, and recruits to the plasma membrane various actin regulatory proteins, such as Eps8, Ena/VASP4, Wiskott-Aldrich syndrome protein (WASP), N-WASP, mDia2, WASP family verprolin-homologous 2 (WAVE2), and others [101, 102].
Therefore, Delage and colleagues [98] have shown that Cdc42/IRSp53/VASP network negatively regulates TNTs formation and vesicular transport in neuronal cells, suggesting that Cdc42-dependent pathways are mainly involved in filopodia formation, rather than TNTs generation. By contrast, Eps8 (that reduces filopodia formation) is a positive regulator of TNTs biogenesis as its overexpression leads to increases in the extent of TNT connections and cargo transfer. Interestingly, the inhibition of actin related protein 2/3 complex (Arp2/3; a seven-subunit protein complex that mediates the formation of branched networks of filamentous actin, crucial for both filopodia and lamellipodia formation) promotes TNTs biogenesis and actin polymerization over longer distance in neuronal cells [82]. In particular, Arp2/3 inhibition enhances the co-expression of Eps8 and IRSp53, thus favoring their synergistic interaction. This event promotes the reorganization of the actin cytoskeleton, thereby leading to the switch of the system towards the extension of straight F-actin formation rather than the formation of branched networks [87] (Fig. 2). In addition, another recent study has demonstrated that the inhibition of ROCK, a downstream effector of Rho/Rac/Cdc42, stimulates TNTs formation via Myosin II mediated F-actin modulation [103].
However, the same Rho GTPase signaling pathways operating in neuronal cells may act in a different way depending on the cell type. Indeed, observations in macrophages indicate that the blockage of Arp2/3 results in a decrease of TNTs formation [104]. Moreover, pathways converging on Arp2/3, including Cdc-42-WASP and Rac1-WAVE2 contribute together to TNT biogenesis. Consistent with this, in human trabecular meshwork cells (HTMCs; endothelial-like cells), the inactivation of Arp2/3 complex causes a reduction in TNT number and length, as well as vesicles transfer [73]. On the contrary, disassembly of actin stress fibers formation using Rho kinases inhibitors increased the number of TNTs and vesicle transfer.
Myo10 is an unconventional myosin that has critical functions in TNTs and filopodia formation [97, 105]. In particular, neuronal TNTs biogenesis requires both the N-terminal head (the myosin motor domain that can bind to F-actin) and C-terminal tail domains (that can bind to several molecules, such as PI(3,4,5)P3, microtubules and β-integrins) of Myo10. Myo10 operates downstream of Cdc42 and promotes TNTs formation independent of VASP (a strong inducer of dorsal filopodia) and through a mechanism independent of integrins and substrate attachment, suggesting that Myo10-driven TNTs may arise from a different mechanism than the one that drives the formation of dorsal filopodia [98, 105, 106].