Cartilage is responsible for dissipating mechanical forces and providing a near frictionless articular surface for locomotion (Sophia Fox et al., 2009). Injury to the articular surface leads to chondrocyte death and cartilage matrix degradation. This cellular and structural damage initiates cycles of inflammation and irreversible tissue degeneration, culminating in end-stage osteoarthritis (OA) (Madry et al., 2011), a leading cause of chronic pain and disability worldwide (Lawrence et al., 2008). Cartilage and other skeletal soft tissues such as meniscus, ligament and intervertebral disc have a low metabolic rate, poor vascularization and as such, possess limited healing capacity (Bianchi and Sinigaglia, 2012). Further, no current therapeutics can slow progression of OA after joint injury (Sarzi-Puttini et al., 2005).

Recent studies have shown that mitochondrial dysfunction is one of the earliest responses of chondrocytes to mechanical injury, implicating mitochondria (MT) as an important mediator of injury-induced OA (Blanco et al., 2011; Goetz et al., 2017; Delco et al., 2018a; Bonnevie et al., 2018; Liu et al., 2019). We have found that the mitoprotective peptide SS-31 can improve MT function, preserve chondrocyte viability, and prevent tissue loss following cartilage injury (Delco et al., 2018b; Bartell et al., 2020). These findings support the concept of protecting MT function after acute injury to prevent cartilage loss, however MT-targeted strategies to improve healing of degenerated articular cartilage have not been investigated.

Tissue healing is an energy intensive process, and MT dysfunction mediates many degenerative diseases, including Parkinson’s (Winklhofer and Haass, 2010), cardiomyopathy (Pacak et al., 2015), and pulmonary fibrosis (Watson et al., 2016). MT dysfunction has been less well studied in cartilage, in part because chondrocytes are largely glycolytic and do not heavily rely on MT for energy production during homeostasis (Bianchi and Sinigaglia, 2012). However, MT also act as signaling hubs in cellular processes such as apoptosis and autophagy, and are critical organelles for maintenance of cartilage homeostasis (Loeser, 2011). Inflammation that occurs in OA is associated with a shift towards glycolysis for energy production in chondrocytes (Zheng et al., 2021). This metabolic shift may represent an attempt to restore energy capacity after MT dysfunction reduces an already limited supply (Mobasheri et al., 2017). Therefore, replacing dysfunctional MT that have been lost to injury and/or disease may promote healing in avascular, metabolically quiescent tissues such as articular cartilage.

Intercellular MT transfer, a process whereby injured and dysfunctional cells recruit MSCs to donate functional MT, has been shown to restore MT function in energy-expensive tissues such as neurons (Berridge et al., 2016), cardiomyocytes (Pacak et al., 2015) and renal tubular epithelia (Konari et al., 2019), as well as more quiescent, poorly vascularized tissues such as the corneal epithelia (Jiang et al., 2016). In vivo murine models of acute lung injury revealed that MSC-mediated MT transfer protects against tissue degeneration after pulmonary injury (Islam et al., 2012; Jackson et al., 2016). Though the mechanisms mediating intercellular MT transfer have not been fully elucidated, evidence from a variety of cell and tissue types implicate processes requiring direct cell-cell contact, including filapodial extensions, tunneling nanotubules, gap junction formation, and cell-cell fusion events (Chinnery et al., 2008; Larsson et al., 2008; Murray and Krasnodembskaya, 2019).

In addition to direct cell-cell MT transfer, our group and others have documented apparent non-contact MT transfer via extracellular vesicles (EVs). EVs are small, membrane bound structures released by all cell types. EVs serve as vehicles for cell-cell signaling and orchestrate many biologic processes including inflammation, immunomodulation, homeostasis, metastasis, and tissue repair (Lai et al., 2010; Luwig and Giebel, 2012; Yuana et al., 2013; Kogure et al., 2020). While EV populations are heterogenous and remain largely uncharacterized, they are most commonly classified into three categories based on size and mode of biogenesis: exosomes, microvesicles (MVs), and apoptotic bodies (Gould and Raposo, 2013). Recent work has identified mitochondria-specific cargoes, including DNA, RNA, and proteins in EVs originating from many cell types, including fibroblasts (Sansone et al., 2017), neurons (Peruzzotti-Jametti et al., 2021), and MSCs (Phinney et al., 2015). Furthermore, intact MT have been identified in larger vesicles derived from adipocytes (Crewe et al., 2021) and MSCs (Phinney et al., 2015; Morrison et al., 2017). In one study, at least 60% of human plasma-derived EVs contained MT, and MT cargoes were ubiquitous in the very largest vesicles (Zhang et al., 2020). Further, EV-mediated MT transfer has been documented between MSCs and professional phagocytic cells; in vitro, macrophages were found to take up MSC-derived EVs containing MT and incorporate them into their mitochondrial networks (Phinney et al., 2015; Morrison et al., 2017). This process of stem cell to phage mitoEV-mediated transfer has also been documented from neural stem cells (NSCs) to mononuclear phagocytes, with transfer restoring ATP capacity, upregulating oxidative phosphorylation, and reducing inflammation (Peruzzotti-Jametti et al., 2021). However, MSC-derived mitoEV transfer has not been demonstrated to chondrocytes. Intriguingly, our group has reported MSC-chondrocyte MT transfer in direct coculture and observed MSCs apparently shedding MT into the extracellular space during co-culture with injured chondrocytes (Bennett et al., 2019; Fahey et al., 2022). However, it was unclear if these MT were packaged within EVs (mitoEVs). Therefore, the goals of this study were to characterize EVs produced by murine and human bone marrow derived-MSCs (including mitochondrial cargoes), determine if MSC-mitoEVs contain functional mitochondria, and investigate whether chondrocytes are capable of taking up MSC-derived mitoEVs. While live MSCs have shown promise as an orthobiologic therapy (Kingery et al., 2019), the therapeutic use of live cells presents significant clinical challenges (Musiał-Wysocka et al., 2019). Harvesting autologous MSCs from sources such as bone marrow or cartilage requires painful and/or invasive procedures, while the use of non-autologous cells raises concerns of immunogenicity and tumorigenesis (Hall and Watt, 1989; Charron, 2013). Additionally, culturing, expanding, transporting, and storing live MSCs requires a large investment of time, resources, and expertise (Asghar et al., 2014). These obstacles, alongside the high regulatory burden associated with the clinical use of MSCs, have greatly limited their widespread use and caused clinicians to seek more practical and readily available orthobiologics.

This study demonstrated that human MSCs release intact, functional MT in EVs. To the best of our knowledge, this work represents the first evidence of MSC to chondrocyte, EV-mediated MT transfer. Further, we confirmed that MSC-derived mitoEVs can serve as a vehicle for MT transfer in the absence of direct cell-cell interactions between MSCs and chondrocytes. Our findings are consistent with previous reports, where Phinney et al. found that MSCs undergoing oxidative stress produced vesicles containing dysfunctional, partially depolarized MT to outsource mitophagy. Macrophages engulfed these vesicles and repurposed functional components, improving bioenergetics for both the MSC and macrophage (Phinney et al., 2015). Morrison et al. used mitochondrial organelle staining of MSC-derived EVs and flow cytometry to show that mitoEVs released by MSCs can be incorporated by macrophages into their MT networks. MitoEV uptake by macrophages enhanced phagocytosis by promoting oxidative phosphorylation, and improved their ability to treat lipopolysaccharide-induced lung injury in vivo (Morrison et al., 2017). Consistent with our findings, evidence that cells release both functional and non-functional MT in EVs suggests mitoEV transfer is a dynamic, context-driven process, likely involving crosstalk (i.e., bidirectional mitoEV transfer) between donor and recipient cells (Morris and Hollenbeck, 1993; Plotnikov et al., 2008). Studies of contact intercellular MT transfer, involving direct interactions between donor and recipient cells in other tissues indicate that calcium signaling and gap junction proteins play an important role (Islam et al., 2012; Hayakawa et al., 2016). However, the mechanisms of EV-mediated MT transfer are unclear, and further investigation is warranted. In our proof-of-concept experimental design, we found a low rate of mitoEV uptake by chondrocytes in vitro. However, it should be noted that several factors, including culture time and MV numbers were not optimized for these experiments. In similar work, where chondrocytes and MSCs were directly co-cultured, the frequency of MT transfer was higher than reported here for mitoEV-mediated transfer; between 1–15% depending on experimental parameters, including culture time, cell ratios, and environmental conditions (Bennett et al., 2019; Fahey et al., 2022). Therefore, we anticipate that future work to optimize experimental parameters and identify mechanisms mediating MSC-chondrocyte mitoEV transfer will allow us to manipulate the rate of mitoEV release by MSCs and uptake by chondrocytes.

To investigate functionality of MT packaged into mitoEVs we used TMRM staining, where fluorescent intensity is directly proportional to the inner MT membrane potential (Brand and Nicholls, 2011). Using this technique, we found that a subset (at least 5%, up to 70%) of mitoEVs contain functional MT, meaning that they have a competent electron transport chain and are capable of producing ATP. We suspect that the high variability in TMRM-staining between experiments is reflective of the exquisite sensitivity of mitochondria to depolarization, and subtle differences in MT stress induced by EV isolation and analysis techniques, especially flow cytometry, although further studies would be required to confirm this. Our preliminary experience suggests that while fluorescence activated cell sorting (FACS) of mitoEV subpopulations is possible, this approach is damaging to mitoEV polarity, and too time, effort, and cost-intensive to be practical at this time (Kormelink et al., 2016). However, our data suggest the possibility of more scalable, size-dependent methods of mitoEV isolation; We found that more than 70% of MSC-MVs are mitoEVs, a higher percentage than previously reported (Morrison et al., 2017). This is likely due to several experimental differences, such as cell of origin, EV isolation technique, and EV identification method. However, further size analysis of our mitoEV population indicated that nearly all of the MVs >500 nm contain mitochondria. Further studies to investigate methods of isolating mitoEVs based on a size will be necessary to investigate the functional impact of mitoEV uptake on injured chondrocytes. Based on work suggesting MSC donation of functional MT to injured cells can rescue cell viability and promote tissue healing, optimizing functionality of mitoEVs during isolation would be an important consideration for the investigation of mitoEVs as a regenerative orthobiologic therapy (Pacak et al., 2015; Berridge et al., 2016; Jiang et al., 2016; Konari et al., 2019).

Our data suggest mitoEVs themselves are not a homogenous population, but rather a diverse collection of vesicles containing distinct mitochondrial cargoes, with a range of functionality. Distinct mitoEV sub-populations may play biologically distinct roles, including cellular rescue through intercellular mitochondrial transfer, outsourcing of mitophagy and diverse signaling functions (Phinney et al., 2015; Morrison et al., 2017; Sansone et al., 2017; Zhang et al., 2020; Peruzzotti-Jametti et al., 2021). DLS size data further supports this. We found that agents known to affect mitochondrial dynamics and function shift the size profile of EV subgroups released by MSCs; When MSCs were stimulated with either a mitoprotective peptide or a MT specific stressor, we observed a significant shift towards a larger 100–1,000 nm, “medium” peak, presumably MVs. Although our current experimental methods did not allow us to determine why both groups responded similarly, these data suggest mitoEV release is responsive to changes in cellular MT function. It is possible that following rotenone/antimycin-induced MT stress, MSCs increased packaging of dysfunctional MT into MVs, consistent with the previous work demonstrating stressed MSCs release unhealthy MT into EVs to outsource mitophagy (Phinney et al., 2015). Following mitoprotective treatment with SS-31, MSCs may similarly increase export of healthy, functional MT in mitoEVs. Further research into the origin, mechanisms of release, and function of mitoEV sub-populations is necessary to elucidate how MSC MT function is related to mitoEV biogenesis.

Measurement of particles smaller than 200 nm (i.e., exosomes) was limited to size and protein analysis. Neither of the MT proteins we probed for (COXIV an outer MT membrane protein and ATP5A1, an enzyme that catalyzes synthesis of ATP) was present in the exosome fraction. This may be due to the specific proteins we chose to investigate, rather than an indicator that MSC-exosomes do not transport MT components. Previous analysis has shown that exosomes may be more likely to contain miRNA and mtDNA, as opposed to the MT membrane proteins and intact MT seen in larger vesicles such as MVs (Phinney et al., 2015; Torralba et al., 2018; D’Acunzo et al., 2021). As techniques improve and high resolution, microparticle optimized flow cytometers become more widely available, characterization of individual mitoEV sub-populations will become more straightforward.

Anunderlying goal of this study was to characterize MSC-derived mitoEVs and investigate their potential suitability as vehicles for non-contact, acellular MT delivery in therapeutic applications. The properties that make EVs effective in long range endogenous communication, also make them excellent candidates for regenerative therapy. EV lipid bilayers are durable enough to retain structural integrity after freeze-thawing and remain intact following intraarticular injection (Li et al., 2019). EVs are non-replicating and have low levels of major histocompatibility complex proteins, posing minimal risk of immunogenicity or tumorigenicity (Yeo et al., 2013; Zhu et al., 2017; Tsiapalis and O’Driscoll, 2020). We confirmed the presence of functional MT in MSC-MVs, identified evidence of distinct sub-populations of mitoEVs, and found that, although MT are present in MVs of all sizes, large MVs are much more likely to contain MT. Further, we observed evidence that stressed chondrocytes will take up MT packaged in mitoEVs in the absence of physical interactions with the parent MSCs. The methods developed in this study will allow investigation into the effects of these mitoEVs after uptake and the mechanisms underlying this phenomenon, as well as those behind why distinct sub-populations such as functional and dysfunctional mitoEVs are released.