Mitochondrial Transfer by Mesenchymal Stromal Cells

By Benn Bluestein-Veyra, M.Sc. Organic Chemistry · Updated 20 October 2026

Mitochondrial transfer from MSCs is the movement of whole mitochondria or mitochondrial DNA from mesenchymal stromal cells into neighboring cells. Spees and colleagues reported in 2006 that human MSCs and skin fibroblasts restored aerobic respiration in cells lacking functional mitochondrial DNA.^[1] Tunneling nanotubes, extracellular vesicles, gap junctions and cell fusion have since been described as routes.^[2]

This reference covers MSC mitochondrial transfer as normal cell biology: the 2025 consensus vocabulary, the landmark co-culture experiments, each route and its molecules, and the assays that measure transfer. The findings come from co-culture, structural and physiology studies and from the reviews that collect them.^[3][4]

Defining mitochondrial transfer with the 2025 consensus nomenclature

Brestoff, Singh and a group of more than thirty investigators published a consensus statement on mitochondria transfer nomenclature in Nature Metabolism in 2025. It defines intercellular mitochondria transfer as an evolutionarily conserved process in which one cell delivers some of its mitochondria to another cell in the absence of cell division.^[3]

The statement notes that many transfer mechanisms had been described under a variety of names, which generated confusion. It defines the relevant terms and proposes a nomenclature framework for transfer and for mitochondria transplantation as a foundation for the field.^[3]

Donor and acceptor cells frame every description

The consensus vocabulary names a donor cell, which delivers mitochondria, and an acceptor cell, which receives them. Transfer was first shown to give metabolic support to acceptor cells, and later work added roles in donor mitochondrial quality and in tissue homeostasis and remodeling.^[3]

Berridge and colleagues describe horizontal transfer of mitochondria and mitochondrial DNA in several mammalian cell culture systems, and they propose that it may be a fundamental physiological process in development and tissue homeostasis.^[4] Torralba and colleagues add that mitochondria carry their own circular genome, so horizontal transfer moves genetic material between cells alongside the organelle.^[2]

MSCs restore aerobic respiration in cells lacking mitochondrial DNA

Spees and colleagues used rho0 cells, a human cell line whose mitochondrial DNA had been depleted with ethidium bromide, so they grew only in medium supplemented with uridine and pyruvate. After co-culture with human bone marrow MSCs or skin fibroblasts, colonies grew in restrictive medium without those supplements.^[1]

Using MSCs from six donors, the authors isolated more than a hundred such colonies. Restriction fragment length polymorphism and SNP-PCR assays found donor MSC mitochondrial DNA in the rescued clones, with 76 of 79 clones positive by SNP-PCR, and the clones raised ATP at least threefold over the rho0 cells.^[1]

Nuclear genotyping with a collagen gene polymorphism and nine microsatellite markers showed acceptor nuclei in every clone, which excluded cell fusion as the main mechanism. MSCs expressing a mitochondria-targeted DsRed2 sent cytoplasmic extensions toward the rho0 cells, and labeled mitochondria streamed through them.^[1]

Early co-culture studies saw mitochondria inside intercellular tubes

Plotnikov and colleagues co-cultured human MSCs with rat cardiomyocytes and observed intercellular contacts by light, fluorescence and electron microscopy. Mitochondria moved in one direction only, toward the cardiomyocytes, and transmission electron microscopy found mitochondria-like structures inside nanotubes of variable diameter.^[5]

Koyanagi and colleagues saw similar connections between human endothelial progenitor cells and neonatal rat cardiomyocytes. The tubes measured 50 to 800 nm across and 5 to 120 µm long, formed transiently, and carried MitoTracker-labeled structures between the cells.^[6]

Mitochondria move between cells through four described routes

Reviews of the field group the mechanisms into four routes: tunneling nanotubes, extracellular vesicles, cell fusion and gap junctions (Fig. 1).^[2] Each route has its own structure and molecular machinery, and one co-culture can use several at once.^[7]

Sinclair and colleagues tested several routes in one system and found that MSCs from three sources donated mitochondria through nanotubes and endocytosis, with a gap junction contribution in some co-cultures.^[7] The sections below take each route in turn.

Tunneling nanotubes carry mitochondria along open F-actin bridges

Rustom and colleagues described tunneling nanotubes in 2004 as thin structures formed de novo between cells, building complex networks. They carried membrane vesicles and organelles selectively and seemed to impede the flow of small molecules, which the authors proposed as a principle of cell-to-cell interaction based on membrane continuity.^[8]

Sartori-Rupp and colleagues used correlative FIB-SEM, light and cryo-electron microscopy to resolve neuronal nanotubes. Each was a bundle of open-ended individual tubes filled with parallel actin bundles, on which membrane-bound compartments and mitochondria appeared to move.^[9]

Austefjord and colleagues review nanotubes as diverse in morphology and composition, varying between and within cell systems.^[10]

Zurzolo notes that several TNT-like protrusions have been described and that most studies relied on low-resolution imaging. She identifies consistent proof of material transfer between two distinct cell populations as a central requirement for the field.^[11]

Stromal cells from several tissues build nanotubes in co-culture

In the Sinclair study, F-actin tubes extended from bone marrow, lung tissue and bronchoalveolar lavage MSCs to BEAS2B bronchial epithelial cells within 45 min of co-culture (Fig. 2). All three MSC populations donated mitochondria with similar efficiency, and the epithelial cells donated none back to the MSCs.^[7]

Cytochalasin D, which blocks the actin tubes, and dynasore, which blocks endocytosis, each reduced mitochondrial transfer in all three co-cultures.^[7] Transfer in this system therefore drew on nanotubes and on vesicle uptake together.^[7]

Immune cells take up stromal mitochondria in co-culture

Jackson and Krasnodembskaya describe a direct co-culture of human bone marrow MSCs with monocyte-derived macrophages, in which MSC mitochondria moved to the macrophages through nanotubes. Transfer raised macrophage oxidative phosphorylation and phagocytosis, and cytochalasin B, which blocks nanotube formation, abolished the phagocytosis effect.^[12]

Luz-Crawford and colleagues found that Th17 cells took up mitochondria from bone marrow MSCs within 4 h of co-culture. IL-17 production fell in a contact-dependent way, together with a rise in Th17 oxygen consumption and conversion toward regulatory T cells.^[13]

Court and colleagues measured dose-dependent transfer from mitochondria-labeled MSCs mainly to CD4+ T cells, with less to CD8+ T cells and CD19+ B cells. Isolated MSC mitochondria delivered artificially raised FOXP3, IL2RA, CTLA4 and TGFB1 transcripts and expanded a CD25+ FoxP3+ population.^[14]

M-Sec and Miro1 build and load the nanotube route

Hase and colleagues identified M-Sec as a protein that induces membrane protrusions de novo, some of which tether to adjacent cells and form TNT-like structures. RNA interference against M-Sec reduced endogenous nanotube formation in a macrophage cell line, and blocking its interaction with Ral and the exocyst complex attenuated tube formation.^[15]

M-Sec is a 73 kDa cytosolic protein also known as TNFAIP2 or B94, with sequence similarity to the exocyst subunit Sec6.^[16] The human TNFAIP2 gene maps to chromosome 14q32.32.^[17]

Kimura and colleagues found that the N-terminal polybasic region of M-Sec binds phosphatidylinositol 4,5-bisphosphate to reach the plasma membrane. Their crystal structure showed helix bundles in a straight rod, like membrane tethering complex subunits, with a positively charged C-terminal surface that recruits active RalA.^[16]

Miro1 links mitochondria to microtubule motors

Fransson and colleagues named Miro-1 and Miro-2, for mitochondrial Rho, in 2003: GTPases with tandem GTP-binding domains separated by a linker carrying calcium-binding EF hand motifs, conserved from yeast to mammals.^[18] A C-terminal transmembrane domain targets Miro to mitochondria, and Miro binds the kinesin-binding proteins GRIF-1 and OIP106.^[19]

MacAskill and colleagues showed that Miro1 links mitochondria to KIF5 motors, which move them along microtubules. Micromolar calcium binding to Miro1 inhibits this link, and Miro1 with mutated EF hands no longer stopped mitochondria in response to glutamate or neuronal activity.^[20] The human gene, RHOT1, maps to chromosome 17q11.2.^[21]

Torralba and colleagues review evidence that Miro1 levels in MSCs set the efficiency of intercellular mitochondrial movement, which ties the motor adaptor to the nanotube route.^[2] Smith and colleagues solved the human Miro1 N-terminal GTPase domain at 1.7 Å, bound to GTP in a non-catalytic configuration of the active site (Fig. 3). Small-angle X-ray scattering of the soluble Miro1 and Miro2 proteins allowed modeling of a crescent-shaped assembly.^[22]

Extracellular vesicles carry mitochondria and mitochondrial DNA

Phinney and colleagues found that MSCs manage intracellular oxidative stress by targeting depolarized mitochondria to the plasma membrane in microvesicles mediated by arrestin domain-containing protein 1 (ARRDC1). Macrophages engulfed these vesicles and reused the mitochondria through fusion, which raised macrophage bioenergetics.^[23]

The microvesicles carried TSG101 and ARRDC1, and two-color MitoTracker labeling followed MSC mitochondria fusing with the macrophage network.^[23] The same MSCs shed microRNA-containing exosomes that dampened macrophage Toll-like receptor signaling, and these smaller vesicles carried mitochondrial DNA without whole mitochondria.^[23]

Vesicle-borne mitochondria support tissue physiology

Rosina and colleagues found that thermogenically stressed brown adipocytes release extracellular vesicles carrying oxidatively damaged mitochondrial parts. Resident macrophages cleared them, and depleting the macrophages let the vesicles accumulate and blunted the thermogenic response to cold.^[24]

Nicolás-Ávila and colleagues found that macrophages in the healthy mouse heart take up mitochondria ejected by cardiomyocytes in membranous particles resembling neural exophers. The cardiomyocyte autophagy machinery drove ejection, and the phagocytic receptor Mertk supported clearance.^[25]

Gap junction proteins and partial cell fusion shape contact-dependent transfer

Gap junctions are aggregates of intercellular channels that pass ions and small molecules directly between cells, and in chordates each channel half is a hexamer of connexins.^[26] Lee and colleagues solved the human connexin-43 channel by cryo-EM in detergent and lipid nanodiscs, showing two docked hexameric hemichannels (Fig. 4).^[27]

The Lee structures captured three conformations of the connexin-43 N-terminal helix, gate-covering, pore-lining and a flexible intermediate, whose balance shifted with cholesterol analogs, C-terminal truncation and pH.^[27] Connexin-43 is encoded by GJA1 on chromosome 6q22.31.^[28]

Connexin-43 and its GJA1-20k isoform set the level of transfer

Fu and colleagues found that GJA1 mRNA is alternatively translated into N-terminally truncated isoforms, of which GJA1-20k is the most abundant. GJA1-20k associated strongly with mitochondria, concentrated at the interface between mitochondria and microtubules, and needed its microtubule-binding domain to support mitochondrial transport.^[29]

Irwin and colleagues co-cultured bone marrow MSCs carrying fluorescent mitochondria with immortalized human chondrocytes. Transfer was higher in direct contact than across a Transwell insert, and connexin-43 staining concentrated at the tips of MSC actin extensions that contained mitochondria.^[30]

Connexin-43 knockdown lowered transfer, and GJA1-20k expression in MSCs correlated with the incidence of transfer.^[30] Results vary by system: in the Sinclair co-cultures, the connexin-43 inhibitor GAP26 left mitochondrial transfer unchanged.^[7]

Partial cell fusion joins cytoplasm and keeps donor mitochondria

Acquistapace and colleagues co-cultured human multipotent adipose-derived stem cells with mouse cardiomyocytes. Hybrid cells formed mainly through partial cell fusion with intercellular structures of F-actin and microtubules, expressed cardiac progenitor markers and kept a mouse genotype.^[31]

Stem cell mitochondria entered the cardiomyocytes, persisted in the hybrids and were required for the shift toward a progenitor-like state. Human bone marrow stromal cells behaved like the adipose cells, and human fibroblasts did not.^[31]

Measuring mitochondrial transfer requires labels that follow the organelle

Every transfer assay labels the donor, co-cultures it with an acceptor and reads where the label ends up (Fig. 5). Berridge and colleagues describe three label families: mitochondria-targeting dyes, mitochondrially imported fluorescent proteins and mitochondrial DNA polymorphisms, each with its own limitations.^[32]

MitoTracker dyes need confirmation by a second label

Chen and colleagues note that MitoTracker Red accumulates in organelles with a membrane potential and can enrich partially at sites other than mitochondria. Compared with mitochondria-targeted GFP, the dye overstated transfer, and experiments with mitochondria-deficient cells confirmed its non-specificity.^[33]

Hole and colleagues labeled astrocyte mitochondria with both GFP and MitoTracker and found that the dye reached neurons independently of mitochondria, including from astrocyte-conditioned medium. They recommend alternative labeling wherever contact-independent transfer is reported.^[34]

Genetic labels and genotyping follow the organelle itself

Spees and colleagues imaged a genetically encoded mitochondria-targeted DsRed2, stably expressed in the donor cells, and combined it with RFLP and SNP-PCR assays of donor mitochondrial DNA and nuclear genotyping to exclude fusion.^[1] Irwin and colleagues gave donor and acceptor mitochondria different fluorescent proteins and gated transfer events against single-color controls by flow cytometry.^[30]

Berridge and colleagues note that mitochondrial DNA polymorphisms suit systems where acceptor mtDNA is depleted, and that they also report the long-term fate of transferred genomes.^[32] The rho0 design adds a functional readout, growth without uridine and pyruvate, that only respiring cells pass.^[1]

Co-culture design separates contact routes from secreted ones

A Transwell insert keeps donor and acceptor apart while letting medium pass, so comparing it with direct contact separates contact routes from secreted ones.^[30] Inhibitors add route detail: Sinclair and colleagues used cytochalasin D for actin tubes, dynasore for endocytosis, carbenoxolone for gap junctions and GAP26 for connexin-43.^[7]

Jackson and Krasnodembskaya label MSCs with a mitochondria-specific dye, wash them extensively, seed them with macrophages at a 1:20 ratio and read transfer by confocal microscopy and flow cytometry.^[12] Pairing such a protocol with a genetically encoded label answers the dye caveats described above.^[33][34]

Where to go from here

The ISCT minimal criteria reference defines the cells these co-cultures start from, and the source tissues reference compares bone marrow, adipose tissue, umbilical cord and placenta. The secretome reference follows the soluble factors and vesicles MSCs release, and the immunobiology reference covers the T cell and macrophage assays met here.

The exosome biochemistry references explain how vesicles are named, isolated and taken up, and the perinatal tissue donor eligibility reference covers the federal framework for perinatal material. Verified practices can request lot documentation, and our team is glad to discuss the methods described on this page.