Mitochondria-Enriched Extracellular Vesicles (EVs) for Cardiac Bioenergetics Restoration: A Scoping Review of Preclinical Mechanisms and Source-Specific Strategies

Data were independently charted by two reviewers (Tadahisa Sugiura and Dhienda C. Shahannaz) using a standardized charting form. Key variables extracted included: (1) EV source (e.g., iPSC-CM, MSC, CPC) (2) Engineering or modification method (e.g., passive loading, overexpression, electroporation). (3) Mitochondrial cargo type (e.g., mtDNA, ATP5a1, PGC-1α). (4) Recipient models and detection methods (e.g., flow cytometry, fluorescence microscopy). (5) Functional endpoints (e.g., ATP restoration, ROS reduction, infarct size, LVEF). (6) Safety and biodistribution metrics, if available.

Outcomes were extracted as reported, including fold change (ATP), percentage changes (infarct size, ROS), or directional improvements (LVEF). Where multiple time points or measurements were provided, the most translationally relevant data were prioritized. No statistical conversions or imputations were applied. Missing or unclear fields were recorded as "not specified" and only included if primary inclusion criteria were met.

Although formal grading tools (e.g., GRADE) were not applied due to the preclinical scope and heterogeneity of designs, consistency and methodological rigor across studies were qualitatively evaluated. Bias risk was assessed using a custom-modified SYRCLE™ tool (Redbound University Medical Center, Nijmegen, The Netherlands), with attention to randomization, blinding, quantification of EV dose, and endpoint validation. No automation tools were used, and no authors were contacted for data clarification.

Due to the methodological diversity across included studies—ranging from EV isolation protocols and mitochondrial labeling to recipient models—a narrative synthesis was employed. Studies were categorized by: EV origin (iPSC-CM vs. MSC); Mitochondrial content (whole organelles vs. isolated proteins or mtDNA); Outcome domains (bioenergetic function, ROS reduction, LVEF, infarct size); Delivery mechanism (e.g., macropinocytosis, TNTs, clathrin-mediated endocytosis).

Where applicable, data were harmonized to percent or fold-change relative to controls for visual clarity. No meta-analyses or pooled estimates were performed.

Tables in Section 3 summarize comparative findings across included studies. Figure 1 (PRISMA flow diagram) outlines the full selection process. No sensitivity analyses or statistical heterogeneity tests were conducted, as this review does not aggregate results quantitatively. Internal consistency in the directionality of functional improvements (e.g., increased ATP or reduced infarct size) was considered qualitatively during synthesis.

Out of 137 initially identified records, 122 remained after duplicate removal. Following title and abstract screening, 34 articles underwent full-text review. Of these, 18 met all inclusion criteria and were included in the final synthesis. Screening was performed independently by two reviewers, and discrepancies were resolved by discussion and consensus.

Sixteen studies were excluded during full-text review for the following reasons:Non-cardiac model (n = 7): e.g., EV-based mitochondrial therapies in glioblastoma, kidney injury, or NSCLC [23,24,25,26,27,28,29].Non-EV mitochondrial delivery system (n = 5): e.g., liposomal, nanoparticle, or MITO-Porter approaches [30,31,32,33,34,35].Lack of mitochondrial relevance despite EV or cardiac focus (n = 4): e.g., general EV studies without mitochondrial functional assays [36,37,38].

Reasons for exclusion are described, and excluded studies are cited accordingly for transparency.

EVs were predominantly isolated from conditioned media using differential ultracentrifugation, tangential flow filtration, or size exclusion chromatography (Studies 2, 10, 17). Density gradient methods yielded high-purity vesicles, whereas tangential flow systems allowed larger-scale processing. Surface engineering approaches included ligand conjugation and PEGylation to enhance cardiac targeting (Studies 16, 18). Mitochondrial enrichment was achieved through donor cell preconditioning, such as hypoxia or mitochondrial stimulation (Study 16). No studies in this review reported direct electroporation of mitochondria into EVs; instead, functional mitochondrial loading was achieved through endogenous packaging or overexpression strategies (Studies 10, 18).

Five studies (Studies 2, 4, 8, 10, 18) profiled mitochondrial cargo beyond miRNA content. iPSC-CM–derived EVs were shown to contain mitochondrial proteins (TOM20, COXIV, TFAM) and intact mtDNA, as confirmed by immunoblotting, qPCR, and proteomics. Fluorescence imaging [e.g., MitoTracker^TM (Thermo Fisher Scientific, Waltham, MA, USA)] and immunogold electron microscopy demonstrated mitochondrial cargo internalization and colocalization with host cell mitochondria. In contrast, MSC-derived EVs displayed a more heterogeneous mitochondrial cargo profile, with incomplete complex representation reported in Study 9.

Furthermore, recent mechanistic reviews have underscored the potential cross-talk between mitochondrial-derived vesicles (MDVs) and classical exosomal pathways, offering insight into the origins and selectivity of mitochondrial cardiac encapsulation (Study 14). These findings may explain observed heterogeneity in mitochondrial complex representation among EVs from different sources.

Six studies (Studies 1, 2, 3, 6, 17, 18) investigated uptake mechanisms, identifying clathrin-mediated endocytosis and macropinocytosis as primary routes. These were validated through pharmacological inhibitors (e.g., EIPA, chlorpromazine) and live-cell imaging. Labeling techniques such as MitoTracker^TM, mito-GFP reporters, and Cy5-tagged mtDNA (Studies 4, 10, 11) enabled visualization of mitochondrial transfer. Bioenergetic restoration was confirmed via respirometry, membrane potential (Δψm) assays, and NAD+/NADH quantification (Studies 2, 5, 10). iPSC-CM–derived EVs restored up to 90% of mitochondrial respiration capacity, whereas MSC-EVs achieved ~50–60% restoration (Studies 9, 16).

ATP levels increased by 2.5–3.2-fold in cardiomyocytes treated with mitochondria-enriched EVs (Studies 2, 3, 10), while ROS levels were reduced by 35–50%, measured by DCFDA and MitoSOX^TM (Thermo Fisher Scientific, Waltham, MA, USA) (Studies 1, 2, 11). In vivo studies involving myocardial infarction models (Studies 5, 6, 7) demonstrated an average infarct size reduction of ~28% and an LVEF increase of up to 16% following EV treatment. Histological analysis (Study 7) confirmed reduced fibrosis and increased angiogenesis in peri-infarct zones.

Table 2 summarizes the outcome measures across all included studies, including ATP production, ROS reduction, LVEF improvement, and infarct size reduction, where reported. Due to the heterogeneity of reporting and lack of precision estimates in many studies, only fold-change and percent differences are presented. Confidence intervals, SDs, or p-values were not consistently reported across experimental studies, which is typical for preclinical EV research.

Due to the heterogeneity and narrative nature of the synthesis, formal sensitivity or subgroup analyses were not applicable.

Eight studies (Studies 2, 3, 8, 9, 12, 13, 15, 16) conducted direct comparisons between iPSC-CM and MSC-derived EVs. iPSC-CM EVs consistently showed superior mitochondrial loading, higher delivery fidelity, and enhanced bioenergetic recovery. Studies 3 and 16 reported a 1.7-fold increase in mitochondrial integration efficiency (p < 0.01), and elevated ATP levels were sustained beyond 72 h post-treatment. In contrast, MSC-derived EVs provided modest structural and antioxidant benefits but demonstrated inconsistent mitochondrial-specific functional rescue (Studies 9, 13).

Although formal risk-of-bias appraisal is not a mandatory component of scoping reviews, we conducted a structured evaluation to contextualize the internal validity of included studies. A modified SYRCLE tool was applied to assess key domains relevant to preclinical research, including randomization, blinding, extracellular vesicle (EV) quantification, and endpoint validation. Summary ratings are provided in. Most in vivo and in vitro studies showed moderate-to-low risk across major domains. In contrast, conceptual or modeling papers were marked as high risk due to the lack of direct empirical evidence. This quality mapping helps interpret study heterogeneity and inform future preclinical standardization efforts. Supplementary Table S1

Among the included studies, 13 were experimental (in vitro or in vivo), while five (5) were mechanistic or conceptual reviews. Publications ranged from 2017 to 2025, underscoring a recent surge in interest toward non-cellular, mitochondria-based therapies. EVs were primarily isolated from iPSC-CMs, MSCs (including adipose-derived stem cells, ADSCs), cardiac progenitor cells (CPCs), fibroblasts, and endothelial cells. Recipient systems included iPSC-CMs, primary cardiomyocytes, and multiple myocardial infarction (MI) animal models, with 13 studies reporting in vivo functional validation.

iPSC-CM-derived EVs emerged as high-fidelity carriers of mitochondrial cargo, consistently delivering intact organelles and respiratory proteins (e.g., ATP5a1, TOM20) with robust cardioprotective outcomes as mentioned in MISEV (Minimum Information for Studies of Extracellular Vesicles) guidelines [56]. In contrast, MSC-EVs showed greater variability in mitochondrial content but demonstrated broader anti-apoptotic and immunomodulatory effects, particularly via miRNA- and SIRT6-enriched cargo.

Mitochondrial cargo types varied significantly by source. iPSC-CM and cardiac-derived EVs most often contained intact mitochondria or key respiratory proteins. MSC- and ADSC-derived EVs largely delivered miRNAs (e.g., miR-144, miR-9-5p, miR-210) or epigenetic regulators such as SIRT6, which indirectly modulated mitochondrial function.

Proteins such as Miro1, involved in mitochondrial trafficking, and TFAM, a key regulator of mitochondrial DNA transcription and packaging, were also identified in select proteomic datasets—supporting the delivery of functional and transcriptionally competent organelle components.

Targeting strategies were typically passive. However, engineered EVs in some studies used peptide ligands for selective cardiac uptake, confirming myocardial tropism via fluorescence tracking. Surface modifications, including PEGylation and ligand conjugation, are increasingly being applied to enhance EV targeting, as shown in drug delivery and regenerative contexts [57]. These modifications may include functionalization with targeting peptides, aptamers, antibodies, or PEG moieties to modulate tissue tropism, immune evasion, and half-life circulation [56,57,58].

EV-mediated mitochondrial transfer was achieved via fusion, macropinocytosis, and receptor-mediated endocytosis, with tunneling nanotubes (TNTs) and connexin-43 channels also discussed in mechanistic reviews. Notably, Phinney et al. demonstrated that mesenchymal stem cells outsource mitophagy and deliver mitochondrial content to recipient cells via TNT-connected vesicular structures, revealing an early mechanistic basis for EV-linked organelle rescue [58]. Fluorescent dyes, mito-GFP reporters, and immunogold techniques confirmed mitochondrial co-localization and internalization, although few studies reported quantitative uptake. A seminal proof-of-concept study by Islam et al. demonstrated that mitochondria delivered by bone marrow-derived stromal cells could restore alveolar bioenergetics and function in acute lung injury models, establishing a foundational paradigm for EV-mediated mitochondrial rescue in vivo [59].

Mitochondrial labeling strategies included MitoTracker^TM dyes (Thermo Fisher Scientific, Waltham, MA, USA), mito-GFP constructs, and Cy5-conjugated mtDNA, which enabled high-resolution visualization of organelle internalization in vitro and in vivo using confocal microscopy, flow cytometry, and immunogold electron microscopy. These internalization mechanisms and post-uptake co-localization steps are visually summarized in Figure 2, illustrating how EVs from different cell sources deliver mitochondrial cargo to damaged cardiomyocytes via TNTs, macropinocytosis, and clathrin-mediated routes, culminating in bioenergetic rescue.

Heyn et al. (2023) [51] further highlighted the relevance of mitochondrial-derived vesicles (MDVs) in cardiovascular contexts. These vesicles, generated via mitochondrial fission and stress pathways such as ESCRT or PINK/Parkin, may contribute to EV mitochondrial loading. This is further supported by earlier mechanistic work showing that MDVs can selectively traffic damaged mitochondrial components via Snx9- and Parkin-regulated vesicle formation, forming a parallel quality control system linked to endosomal sorting and potential EV packaging [60]. Understanding this convergence between mitochondrial quality control and EV biogenesis could inform the design of next-generation vesicles with enhanced targeting and cargo specificity.

Functional readouts demonstrated that M-EVs improve ATP production, restore mitochondrial membrane potential, and reduce ROS in both in vitro and in vivo cardiac models. These effects translated into reduced infarct size, improved LVEF, and enhanced cardiomyocyte survival. Several studies also reported improved calcium homeostasis and field potential normalization, supporting an electrophysiological role in cardiac recovery.

One study also reported improvements in field potential duration, indicating that M-EVs may stabilize not only metabolic but also electrophysiological parameters post-injury—an underexplored but potentially critical aspect of cardiac remodeling.

An important unresolved question is how long EVs persist in circulation and how durable their functional rescue is once taken up by target tissue. Preclinical biodistribution and pharmacokinetic studies indicate that EVs typically have a short half-life in plasma—on the order of 5 to 30 min in small animal models, with clearance by most tissues by ~6 h [61,62,63] after intravenous administration. In non-human primates, some EVs remain detectable in plasma for longer periods (≈40 min), suggesting species differences [62]. Meanwhile, functional rescue of cardiac or cardiomyocyte injury by mitochondria-enriched EVs has been observed as early as 3 h post treatment, with signs of enhanced mitochondrial biogenesis and contractile improvement at ~24 h [16]. Together, these data suggest a discrepancy: EVs may be cleared from circulation quickly, but their delivered mitochondrial cargo can exert lasting effects in injured myocardium—at least for 24 h, and possibly longer, though evidence >24–72 h is limited in the literature to date. Thus, while your manuscript reports longer rescue effects (if it does), it is important to note this gap in the published data. Future studies should include longer follow-ups (e.g., 48–72 h, 7 days) with tracking of both EV presence and mitochondrial function.

Comparative studies consistently favored iPSC-CM EVs in terms of mitochondrial content, uptake efficiency, and bioenergetic restoration. Their enhanced respiratory complex integrity and sustained ATP elevation (>72 h) suggest higher source fidelity. Moreover, the scalability and maturation potential of iPSC-CMs, as previously demonstrated in our optimization of human iPSC-CM production pipelines, support their viability as industrial platforms for therapeutic EV generation [64]. Preclinical studies have also shown that iPSC-derived EVs confer greater regenerative benefit and safety than iPSC transplantation itself, further supporting their therapeutic independence [65]. A visual comparison of these mechanistic domains is summarized in Figure 3, highlighting distinct advantages of iPSC-CM EVs in mitochondrial fidelity, uptake mechanisms, and clinical scalability.

iPSC-derived cardiomyocytes are known to exhibit an immature bioenergetic phenotype, with underdeveloped mitochondrial networks and a reliance on glycolysis [60,66]. As previously discussed in iPSC-based cardiac therapy models, immature mitochondria and oxidative vulnerabilities contribute to poor engraftment and necessitate energetic support [67]. This metabolic vulnerability may be partially rescued through M-EV supplementation.

Meanwhile, MSC-EVs, while versatile and immunomodulatory, showed inconsistent mitochondrial delivery and appeared more suited for paracrine or anti-inflammatory roles. Cardiac-derived EVs, by contrast, have demonstrated stronger infarct-targeted cardioprotection than MSC-EVs in certain contexts [68].

iPSC-CM-derived EVs exhibit superior mitochondrial packaging, delivering ATP5a1, TOM20, and TFAM through fusion and TNT-mediated uptake, resulting in ~90% restoration of mitochondrial respiration and sustained ATP elevation beyond 72 h. In contrast, MSC-derived EVs primarily deliver antioxidant miRNAs such as miR-144, which activates PTEN/AKT signaling and inhibits NOX4, leading to partial (~50–60%) ATP recovery via redox regulation rather than mitochondrial integration.

Functionally, iPSC-EVs reduce infarct size and enhance bioenergetic output with greater fidelity, positioning them as more translationally viable compared to MSC-EVs, whose effects are predominantly paracrine and variable in mitochondrial enrichment. These mechanistic disparities underscore the platform-specific strengths of iPSC-CM EVs as mitochondria-proficient, programmable vesicles for targeted cardiac repair.

An important unresolved question is whether EV mitochondrial enrichment is determined primarily by the intrinsic mitochondrial density of the donor tissue. Current evidence suggests that while donor tissue type contributes, it is not the sole determinant. For example, mitochondria-rich tissues such as brown adipose tissue (BAT) and skeletal muscle release EVs enriched in mitochondrial proteins and polarized mitochondria, often yielding higher amounts than mesenchymal stem cell–derived EVs [69,70,71,72,73]. However, metabolic status and stress also strongly influence mitochondrial cargo: BAT EV proteomes differ between lean and obese states, and thermogenically stressed BAT increases release of damaged mitochondria via EVs [69,74]. Similarly, mesenchymal stem cells under oxidative stress outsource mitophagy through vesicular release [58,75], and activated monocytes export mitochondria in EVs reflective of their activation state [76]. Bone marrow stromal cells have also been shown to restore alveolar bioenergetics in vivo via mitochondrial transfer, independent of basal mitochondrial density [59].

Together, these findings indicate that EV mitochondrial cargo is shaped by both donor tissue mitochondrial content and dynamic cues such as metabolic programming or stress responses [70]. Critically, few studies have directly compared high-density tissues (e.g., cardiomyocytes, muscle, BAT) with lower-density sources (e.g., MSCs, adipose stromal cells) under controlled conditions. Systematic head-to-head studies are therefore needed to clarify the relative contributions of tissue mitochondrial density versus stress-induced packaging mechanisms.

While the core of this review focuses on experimental evidence, two modeling and mechanistic studies provide valuable context. First, a 2023 PBPK modeling study offered a computational framework for predicting EV biodistribution, clearance, and cardiac uptake—paving the way for rational dosing strategies in future trials. These computational findings are built on prior in vivo biodistribution studies, notably by Wiklander et al., which demonstrated that EV organ tropism is significantly influenced by cell source, route of administration, and surface targeting ligands [77].

Second, foundational work by Sugiura et al. demonstrated that stressed mitochondria selectively form mitochondrial-derived vesicles (MDVs), which traffic damaged mitochondrial proteins toward endolysosomal pathways via Snx9- and Parkin-dependent mechanisms [78]. These MDVs have since been proposed to intersect with EV biogenesis under oxidative stress. As such, emerging research into MDVs and mitochondrial-endosomal interactions emphasizes the biogenetic pathways by which stressed mitochondria may be selectively sorted into EVs. Leveraging molecules such as Snx9, OPA1 [79], or Parkin could enhance mitochondrial payload specificity in engineered EVs. In parallel, fusion-mediating proteins such as mitofusin-1 and mitofusin-2 (MFN1/2), which regulate mitochondrial outer membrane tethering and inter-organelle connectivity, may modulate the fidelity of mitochondrial integration post-EV delivery, particularly in fusion-competent recipient cells [80,81]. These insights bridge the gap between molecular design and translational optimization of mito-EV therapeutics.

The collected evidence strongly supports the use of mitochondria-enriched EVs to restore cardiomyocyte function following ischemia–reperfusion or oxidative injury. By combining metabolic rescue with structural benefits, EVs offer a multipronged therapeutic modality. The synergy between miRNA cargo and mitochondrial content in some EVs suggests a layered repair mechanism that targets both gene expression and energy metabolism simultaneously.

The application of EVs for mitochondrial transfer represents a paradigm shift in cardiac regenerative therapy. Compared to whole-cell transplantation, EV-based approaches offer lower immunogenicity, fewer safety concerns (e.g., arrhythmia or teratoma formation), and simpler storage logistics [65], aligning with evidence that iPSC-derived EVs show superior safety and regenerative performance compared to their parental stem cells.

Despite these limitations, the consistency in bioenergetic restoration, infarct reduction, and LVEF improvement across diverse preclinical models reinforces the translational potential of mitochondria-enriched EVs.

This review also carries inherent procedural limitations, including exclusive reliance on published English-language data without author correspondence and the absence of formal certainty grading frameworks such as GRADE. While methodological rigor was prioritized throughout, these constraints may influence the depth of inferential confidence for future clinical translation.

Although a formal GRADE evaluation was not performed, confidence in the preclinical evidence was qualitatively interpreted based on model rigor, experimental replication, and functional endpoint consistency.

Additionally, publication bias cannot be ruled out, as many preclinical studies lacked negative controls or null-effect reporting. While no formal funnel plot or quantitative bias detection was conducted, this limitation reflects a common challenge in regenerative EV research.

Scalability and regulatory compliance are major hurdles for clinical translation. GMP-grade production of mitochondria-loaded EVs is not yet standardized, and long-term storage or shelf-life validation remains limited [85]. Regulatory definitions of EV-based biologics are still evolving, complicating approval pathways. Furthermore, tracking mitochondrial cargo post-infusion remains technically challenging in humans, necessitating non-invasive imaging advancements [86].

To advance the field, future research should focus on the following (Figure 4):Developing robust, scalable EV bioengineering platforms with consistent mitochondrial loading.Conducting clinical trials to assess safety and efficacy in post-MI or heart failure patients [87].Performing long-term studies on myocardial remodeling and electrical integration post-EV therapy.Exploring combination therapies—e.g., EVs integrated with gene editing tools, biomaterial scaffolds, or cardiac patches.Establishing standardized in vivo biodistribution assays using dual-labeled EVs (e.g., MitoTracker + Cy5-EVs) to assess cardiac targeting, retention, and clearance kinetics across delivery routes [77].Validating functional uptake via mitochondrial membrane potential recovery (Δψm), respiratory complex reconstitution (via Western blot for ATP5a1, COXIV), and single-cell OCR in recipient cardiomyocytes.Utilizing side-by-side comparisons of iPSC-CM- and MSC-EV-treated infarct models to define duration, tissue depth, and mitochondrial functional half-life in vivo.Investigating direct intercellular mitochondrial transfer as a complementary or alternative mechanism to EV-mediated delivery, including strategies to harness tunneling nanotubes, gap junctions, or cell-encapsulated mitochondria for clinical applications.The need for longer follow-up periods. Our review highlights that most published studies evaluate therapeutic effects at acute time points (e.g., within 24–48 h) following EV administration. While these studies demonstrate promising immediate bioenergetic restoration, the clinical relevance of such therapies relies on their sustained efficacy. Therefore, future research should incorporate longer-term follow-up assessments (e.g., 72 h, 7 days, or more) to track the long-term persistence of EV presence and their lasting impact on mitochondrial function.

In addition, mechanistic insights from non-cardiac preclinical models (e.g., lung, renal, and neural systems) may guide future optimization of EV-based mitochondrial delivery for cardiac applications. Together, these strategies will help transform mitochondria-enriched EVs from a preclinical curiosity into a clinically actionable therapy for cardiomyopathies.