In‐cytoplasm mitochondrial transplantation for mesenchymal stem cells engineering and tissue regeneration

Abstract Stem cell therapies are unsatisfactory due to poor cell survival and engraftment. Stem cell used for therapy must be properly “tuned” for a harsh in vivo environment. Herein, we report that transfer of exogenous mitochondria (mito) to adipose‐derived mesenchymal stem cells (ADSCs) can effectively boost their energy levels, enabling efficient cell engraftment. Importantly, the entire process of exogeneous mitochondrial endocytosis is captured by high‐content live‐cell imaging. Mitochondrial transfer leads to acutely enhanced bioenergetics, with nearly 17% of higher adenosine 5′‐triphosphate (ATP) levels in ADSCs treated with high mitochondrial dosage and further results in altered secretome profiles of ADSCs. Mitochondrial transfer also induced the expression of 334 mRNAs in ADSCs, which are mainly linked to signaling pathways associated with DNA replication and cell division. We hypothesize that increase in ATP and cyclin‐dependent kinase 1 and 2 expression might be responsible for promoting enhanced proliferation, migration, and differentiation of ADSCs in vitro. More importantly, mito‐transferred ADSCs display prolonged cell survival, engraftment and horizontal transfer of exogenous mitochondria to surrounding cells in a full‐thickness skin defect rat model with improved skin repair compared with nontreated ADSCs. These results demonstrate that intracellular mitochondrial transplantation is a promising strategy to engineer stem cells for tissue regeneration.


| INTRODUCTION
Mesenchymal stem cells (MSCs) are the so-called "first-generation stem cell type". The anti-inflammatory, immunomodulatory, angiogenic, proapoptotic and trophic activities of MSCs, in combination with the ease of isolation and amplification, have led to over 950 registered MSCs clinical trials listed with ClinicalTrials.gov. However, there are several drawbacks in using MSCs: (1) the therapeutic potential of MSCs is highly variable in a complex pathophysiological environment 1 ; (2) the overall efficiency of MSCs engraftment to tissue injuries is poor, especially when these cells are systemically administered. 2 Consequently, there is a need for innovative approaches to endow "next-generation MSCs" with enhanced features and functionalities.
Consequently, taking inspiration from chimeric antigen receptor (CAR) T cells, MSCs surface can be engineered by a IgM-derived anti-GD2 CAR to specifically redirect MSCs delivering proapoptotic cytokine tumor necrosis factor-related apoptosis-inducing ligand (TRAIL) against a GD2-expressing tumor, in an effort to prolong site-specific retention of MSCs. 3 By applying popular CRISPR/Cas9-adenoassociated virus serotype 6 (AAV6) platform, genetically engineered MSCs lines with exogenous DNA integration overexpressing vascular endothelial growth factor and platelet-derived growth factor provided reduced immune clearance in vivo resulting in superior therapeutic efficacy in the diabetic mouse wound healing model. 4 However, these technologies are labor intensive and restricted with clinical applications due to genetic manipulation.
Another engineering strategy to boost the potency of MSCs is to prime the MSCs by exposing them to low oxygen cultivation, 5 small molecules, 6-8 functional particles, 9,10 distinct types of biomaterials, 11,12 or modified culture conditions 13,14 before transplantation. Unfortunately, the positive effects are only retained for several hours to a few days after the transfer of primed MSCs to in vivo environments. 15 Therefore, more effective, consistent and long-lasting priming effects are necessary.
Mitochondria are the key intracellular organelles contributing to cell energetics and viability. Since the initial discovery demonstrated that mitochondria-deficient A549 lung cancer cells could obtain functional mitochondria from donor MSCs, 16 the "mito-healing" theory of mitochondrial transfer from stem cells to damaged cells has been subsequently supported by numerous studies in a variety of mammalian cells, including cardiomyocytes, 17 neurons, 18 immune cells, 19,20 and epithelial cells. 21,22 Remarkable elevation in adenosine 5 0 -triphosphate (ATP) levels, restoration of cellular bioenergetics, and reduction in oxidative stress have been observed in the rescued recipient cells after mitochondrial transfer. 23 The success of stem cell-derived mitochondria to multiple cell types raises the question whether "next generation stem cells" with superior self-functionality could be engineered by transferring additional mitochondria to MSCs. Direct mitochondrial transplantation has been explored in clinical practice. Purified respiration-competent mitochondrial particles can be delivered via intravenous injection or intravascular coronary injection. 24,25 A successful in vivo mitochondrial transplantation trial in pediatric patients with cardiac disease in 2017 implied that mitochondria exhibit the ideal cellular organelle for transplantation, as they continue to survive independently from the host cell and further mitigate the possibility of an autoimmune reaction because of "natural attributes". 26 In the current study, we sought to develop a strategy for energetic engineering of MSCs by transfer of exogeneous mitochondria.
Our hypothesis was that mito-transferred MSCs would demonstrate enhanced therapeutic efficacy by modifying the metabolic state thereby positively affecting cellular phenotypes and paracrine activities. To benefit the persistent priming effects, we assembled recipient MSCs derived from human adipose tissue (ADSCs) by loading allograft mitochondrial particles isolated from donor ADSCs. We used highcontent live-cell imaging to investigate the effective dose and the optimal timing for exogenous mitochondrial internalization. To track gene expression changes, we analyzed total RNA profiles of ADSCs modulated by exogenous mitochondria. To investigate their secretome changes associated with paracrine mechanisms, comprehensive proteomic analysis was employed by liquid chromatography tandem mass spectrometry (LC/MS-MS). Finally, we tested the therapeutic efficacy of the mito-transferred ADSCs by transplanting them into a rat full-thickness skin defect model and monitoring the rate of wound healing. The results of this study pave the way for energetic engineering of "next-generation stem cells" for tissue repair. Excessive homogenization gives rise to mechanical damage of mitochondria. We first investigated the integrity of purified mitochondria from ADSCs. The isolated mitochondria derived from Y40-ADSCs with differential centrifugation after mechanical lysis represented intact "rod-shaped bacteria" appearance ( Figure 1a). Further information about the mitochondrial structure formation was provided by dynamic light scattering (DLS) and zeta potential determination (Figure 1b,c). The size of the mitochondria ranged from 100 nm to The purity of mitochondria was also evaluated by western blotting to analyze the mitochondrial marker proteins in isolated mitochondria.
Small cytoplasmic proteins, such as 60S ribosomal protein L9 (RPL9) and β-actin, were present only in low levels in purified mitochondria ( Figure 1d), whereas mitochondrial proteins, including glutaminase (GLS) and mitochondrial import receptor subunit TOM20 homolog (TOMM20) were enriched in mitochondrial pellet. We conclude that mitochondrial structure is retained during rapid isolation of mitochondria from ADSCs (<40 min).
To determine the biological stability, ATP changes in isolated mitochondria from Y40-ADSCs were measured using a luciferasebased kinetic assay over time. Mitochondrial ATP was stable within 5 h for storage at 4 or 37 C, whereas it was reduced to about 90% of initial level after 24 h (Figure 1e), indicating the difficulty of respiration competent long-term maintenance without host cells. To meet the quantity requirement for isolated mitochondria, Y40-ADSCs were amplified in vitro. We compared the mitochondrial yield from three different donor cell numbers. As predicted, increased cell number produced more mtDNA and mito-protein ( Figure S3b,c), while ATP production per unit of mito-protein remained constant ( Figure S3d).
However, the best extraction yield (described as the percentage of isolated mitochondria by wet cell mass) was obtained using 10 6 cells/ml (Figure 1f), possibly due to the limited capacity of Potter-Elvehjem grinder. Taken together, isolated mitochondria from ADSCs need to be prepared in small batches and used freshly.

| Kinetic analysis of allotransplanting exogenous mitochondria into recipient Y74-ADSCs
To determine whether Y74-ADSCs can take up exogenous To quantify the extent of exogenous mitochondria delivered into ADSCs, we carried out real-time tracing of mitochondria transfer using high-content imaging. Twenty fields of view (FOVs) were captured per well, and each FOV contained approximately 290 cells ( Figure 2b). Isolated Y40-ADSCs derived mitochondria were added at a blank, low, media, and high dose, respectively (0-18.5 μg/ml per 10 4 recipient cells). The three-channel fluorescence time-lapse images were acquired over 12 h period. We found exogeneous mitochondriapositive Y74-ADSCs reached 90% at 18.5 μg/ml (Figure 2c), which was considered as the optimal mitochondrial dosage for further application. Dynamic mitochondrial uptake is shown as movie in Videos S1-S4. At the same time, the transport of donor mitochondria to the recipient cells showed a dose-dependent accumulation of MitoTracker Red CMXRos-fluorescence, which gradually increased over the first 6 h (Figure 2d). A 6-h incubation was sufficient to reach the maximum mitochondrial concentration internalized by recipient cells (Figure 2e). Next, we used high-throughput analysis to investigate the effects on the original mitochondria of Y74-ADSCs. When  Figure S4). All this suggested the internalized exogenous mitochondria could still potential provide ATP to meet the metabolic demands, without impairing the mitochondrial function of the recipient ADSCs.
Based on the finding above, we further performed the Seahorse XF real-time ATP rate assay to quantify metabolic switching in response to exogenous mitochondrial stimulation. A higher proportion of ATP was generated by glycolysis in mito-transferred ADSCs than in control ADSCs (55% vs. 49%, respectively) (Figure 2i), while a smaller proportion was generated by oxidative phosphorylation (OXPHOS) (45% vs. 51%, respectively). The total ATP production rate was 17% higher in the population of mito-transferred ADSC, suggesting that exogenous mitochondrial incorporation may increase overall metabolic activity of stem cells by glycolytic pathway, similar to the Warburg effect observed in cancer cells.

| Improved stress tolerance of mito-transferred Y74-ADSCs against serum starvation and oxidative stress in vitro
To confirm our main hypothesis of whether the proliferation of Y74-ADSCs could be sufficiently enhanced by mitochondrial transfer, ADSCs were transfected with exogenous mitochondria at different doses for 6 h, followed by assessing cell proliferation for additional 12 h under serum starvation via CCK-8 ( Figure 3a). The proliferation rate was significantly higher in Y74-ADSCs with exogenous mitochondria at 18.5 μg/ml than in nontreated Y74-ADSCs served as control from 6 to12 h. At 12 h, the results of a cell counter showed that mitotransferred ADSCs had up to twofold higher cell number compared to control ADSCs (Figure 3b). Increase cell density was also observed by phase-contrast microscopy (Figure 3b), indicating that mitochondrial transfer has notable self-renewal potential by ADSCs proliferation under serum starvation.
Subsequently, under the same condition-serum starvation, we conducted both scratch wound healing and cell migration using live cell imaging. We first performed the scratch assay as wound healing is a critical step of tissue homoeostasis and repair. Y74-ADSCs transferred with mitochondria displayed enhanced migration leading to accelerated gap closure of the cell-free area compared with control ADSCs ( Figure S5a,b). There was significantly greater healed area for mito-transferred ADSCs (45.8% vs. 12.6% for control) (Figure 3c,d).
Next, the results from transwell migration assays confirmed that mitotransferred ADSCs were more migratory as they migrated through the porous membrane in increased cell numbers to the surface of the basolateral chamber (Figure 3e,f and Figure S6a,b). This indicates that exogenous mitochondria augment the mobility of ADSCs.
To further investigate the effect of mitochondrial transfer on motility, the migratory behavior of mito-transferred ADSCs was analyzed by time-lapse microscopy for 3 days. Without mitochondrial transfer, ADSCs were observed to migrate individually and slowly toward the central area of wound. When mitochondrial stimulus was present, ADSCs migrated collectively and resulting in a more rapid wound closure (Video S5). Therefore, these data suggest that mito-transferred ADSCs are more resistant against the injury simulated by serum deprivation.
To elucidate the cellular response against ROS, the mitotransferred ADSCs were exposed to Dox (200 nM) and examined for SA-β-Gal activity, colony-formation ability, and multilineage differentiation potential. Prior to these experiments, we first checked the survival of ADSCs against Dox by measuring the changes in mitochondrial membrane potential using a JC-1 assay. The monomeric JC-1 displays green fluorescent and aggregates in a mitochondrial membrane potential-dependent manner to display red fluorescence. Therefore, the color difference of the JC-1 stained mitochondria can be used to assess the severity of mitochondrial damage caused by different treatments. As shown in Figure S7a

| Rapid tissue repair by transplantation of mito-transferred ADSCs
Promoted proliferation and migration functions of mito-transferred ADSCs in vitro could translate to a rapid tissue repair response in vivo.
The therapeutic effects were verified in a rat full-thickness skin wound model (Figure 6a). Similar to mito-transferred Y74-ADSCs, mito-transferred rat-ADSCs were engineered by delivery of exogenous mitochondria isolated from Rat A-ADSCs into ADSCs derived from Rat B using the serum starvation combined with co-culture method. We included the following experimental groups: (i) salinecontrol, (ii) rat B-ADSCs, and (iii) mito-transferred rat B-ADSCs, followed by implanting total 6 Â 10 6 cells to the skin fat layer by direct injection. The presence of mito-transferred ADSCs at the (e) Schematic illustration of highly proliferative Y74-ADSCs construction by incorporating exogenous mitochondria to accelerate the cell cycle. Exogenous mitochondrial transfer promotes the glycolysis, that efficiently produces ATP and stimulates recipient cell proliferation via enhanced CDK1/2. The activities of these CDKs are primarily regulated by the periodic expression of their cyclin-binding partners, temporally control sequential cell cycle transitions through G1 to S phase and G2 to M phase. The division cycle contains long growth phase (G1) followed by a DNA synthesis phase (S) that is followed by a short growth phase (G2) before the next round of mitotic division (M) F I G U R E 5 Cellular secretome changes that occur with mitochondrial transfer. (a) Venn diagram depicting unique and shared proteins in conditioned media (CM) derived from mito-transferred Y74-ADSCs versus control Y74-ADSCs (n = 3). (b) Volcano plot indicating protein distribution in CM after mitochondrial transfer, as determined by quantitative proteomics. Positive (red; upregulated) and negative (blue; downregulated) correlations showed the log 2 fold expression changes (log 2 FC > 2, <À2, and p < 0.05). (c) Gene Ontology analysis of functional annotations upregulated (top) and downregulated (bottom) by mitochondrial transplantation. (D) Several proteins involved in "immune response" and "cell growth" secreted in significantly different quantities by mito-transferred Y74-ADSCs versus control Y74-ADSCs. Data are presented as the average number of PSMs. Significantly different (one-way analysis of variance [ANOVA]): ns, not significant, *p < 0.05, **p < 0.01, and ***p < 0.001. CTGF, connective tissue growth factor; MYDGF, myeloid-derived growth factor; TGFBI, transforming growth factor beta induced; TIMP1 and 2, tissue inhibitor matrix metalloproteinase 1 and 2; IGFBP3-7, insulin-like growth factor binding proteins family 3-7 F I G U R E 6 Legend on next page.
that the therapeutic benefit of mito-transferred ADSCs not only offers advantages to stem cell themselves but may act as "mitochondrial delivery vehicles" to affect neighboring cells.
Wound size measurements demonstrated that saline-and control ADSCs-treated wounds were slow for closure even after 14 days but the wound sites by mito-transferred ADSCs were healed in 14 days (Figure 6e and Figure S8). Mito-transferred ADSCs significantly accelerated wound closure, compared with saline control and nontreated ADSCs, with 60% closure at Day 3 relative to 10% for control and 30% for nontreated ADSCs; 100% of the wound area was closed by

| DISCUSSION
Despite the cognitive dissonance between preclinical findings in animal systems and human MSC clinical trial outcomes, MSCs are still the most frequently used cell source for tissue repair and regeneration due to their engrafting ability and broad differentiation potential. 27 Thus, especially in the field of somatic stem cells, the focus has been on developing effective methods to maximize clinical potency of MSCs. 28 In this study, we have introduced a novel alternative bioengineering approach to potentiate the therapeutic efficacy of Genetically engineered MSCs (either viruses or noninsertional gene transfer) are usually designed to secrete highly expressed various biologicals targeted against specific disease. The clinical use of these is restricted by packaging problems and safety concerns. 29,30 In comparison, mitochondria are naturally occurring particles with low immunogenicity. Using serum starvation combined with co-culture system, we assemble direct evidence for an easy-to-perform, robust, and amenable to clinical-scale expansion of MSCs transferred with isolated mitochondria from healthy donor stem cells (Figure 1). To the best of our knowledge, this is the first study to follow in vitro acquisition of exogenous mitochondria in real time, and we optimize the effective mitochondrial dose (Figure 2 and Videos S1-S4). The endocytic existence of active and functional exogenous mitochondria was visualized and quantified using MitoTracker Red CMXRos dye, which accumulates in active mitochondria with an intact mitochondrial membrane potential. 31,32 With a mitochondrial dose of 18.5 μg/ml per 10 4 ADSCs, mito-transferred ADSCS displayed a 17% increase in ATP production. These cells retained high multipotency, which is a hallmark of "super" stem cells.
The energy metabolism of undifferentiated MSCs has been reported to have more glycolysis dependence under aerobic conditions compared with OXPHOS dependence. 33 This phenomenon is called as the "Warburg effect," that operates predominantly in highly proliferative cells. 34 Either energy or various types of cellular constituents, such as amino acids, lipids, or nucleotides, are essential for cell division. Indeed, glycolysis along with the pentose phosphate pathway could explain cellular constituents as well as ATP. 35 Therefore, glycolysis is expected to be benefit for self-renewal of the actively proliferating stem cells and "stemness" maintenance. Glycolytic metabolism produces fewer ROS in cells. 36 Interestingly, the elevation in the nonmitochondrial oxygen consumption was detected in mito-transferred ADSCs (Fig. 2h). Several studies suggested that high nonmitochondrial respiration promoted oxidative stress followed by increasing ROS generation, which is an indicator of impaired mitochondrial function. 37,38 Conversely, other researches evidenced that increased nonmitochondrial oxygen consumption was not necessarily associated with ROS, and this phenomenon could be due to the processes including protein folding, synthesis of lipid and collagen, demethylation, and hydroxylation, because mitochondrial oxidative phosphorylation is not the unique way to consume oxygen in cells. 39 To confirm this, we performed DNA damage analysis using 8-OHdG level detection in mito-transferred ADSCs (Figure 4s . (g,h) The functional scores of re-epithelization and collagen deposition was scored 0 to 4 (n = 8). Significantly different (one-way ANOVA): ns, not significant, **p < 0.01, and ***p < 0.001 production rate of mito-transferred ADSCs. We infer that the exogenous mitochondria may induce a metabolic switch from OXPHOS to glycolysis in the bioengineered stem cells, which is similar to the shift during induced pluripotent stem cell (iPSC) reprogramming.
To further encompass these bioenergetics consequences related to mitochondrial uptake, transcriptome analysis was performed on mito-transferred ADSCs populations. Bioinformatic analysis demonstrated that the most significant alterations in mRNA expression levels were found in cell cycle, particularly CDK1 and CDK2 (Figure 4c). The activities of these CDKs are primarily regulated by the periodic expression of their cyclin binding partners, temporally control sequential cell cycle transitions through G1/S phase and G2/M phase. 40 Stimulation of cells with hormones, growth factors, microRNAs, or small molecules induces expression of CDKs and cyclins and accelerates cell cycle progression. 41 At the mechanistic level, it is well known that CDKs contribute to the promoted self-renewal by activating the PI3K-Akt pathway. 42,43 Notably, the knockdown of CDK1, CDK2 or treatment with CDK-inhibitors was reported to trigger differentiation in stem cells, 44 whereas overexpression of CDKs coordinately controls cell proliferation and migration. 45,46 Our observation corroborates these findings, with mito-transferred ADSCs showing greater proliferation, migration, and multilineage differentiation compared with control ADSCs (Figure 3). In addition to our transcriptomic data, secretome analysis also sheds light on the molecular changes associated with mito-transfer, including cytokine levels and antiinflammatory potential ( Figure 5). We identified TGFBI and TIMP 1 to be highly represented in the conditioned medium of mito-transferred ADSCs. TGFBI is an extracellular matrix protein that known to modulate homeostasis by promoting cell adhesion and microtubule stabilization. 47 TIMP 1 is a metalloproteinases inhibitor reported to have an antiapoptotic effect. 48 This implies that mitochondrial transfer may improve the secretome of ADSCs to enhance cell therapy.

The substantial cell-to-cell variation is widely shown between
MSCs. Such heterogeneity develops further during in vitro culture and population expansion. Such pervasive variability also limits their therapeutic efficacy. 49,50 Our results indicated that intracellular mitochondrial transfer results in highly homogeneous ADSCs for a rapid in vitro healing (Video S5), which was subsequently observed in the in vivo setting. Interestingly, we unexpectedly found evidence that the engineered ADSC may also act as "carriers" to share mitochondria with surrounding cells after injection (Figure 6c

| Isolation of mitochondria from donor Y40-ADSCs
To isolate sufficient quantity of mitochondria, 15 cm cell culture dishes were used for Y40-ADSCs amplification. The cells were collected when they achieved 90% confluence and were re-seeded with a den-

| Characterization of isolated Y40-ADSCs mitochondria
To examine the structural integrity, isolated mitochondria were resuspended in mitochondria storage buffer (Qiagen) and mounted on a microscope glass slide to be immediately visualized by the microscope Leica DMI8 with 40Â objective. MitoTracker Red CMXRos accumulates in active mitochondria in response to high mitochondrial membrane potential. Isolated mitochondria maintaining red fluorescence therefore indicates its viability.
The mean size distribution and zeta potential of isolated mitochondria were evaluated by DLS (Nano ZS, Malvern; 173 scattering angle at 25 C; 1.40 refractive index). The measurements were completed following appropriate dilution of the mitochondrial particles in storage buffer. Yield of mitochondria extracted from Y40-ADSCs was calculated by the following equation: where W mito is the weight of isolated mitochondrial pellets and W cell is the weight of total cells before mitochondrial isolation.
ATP levels of isolated Y40-ADSCs mitochondria were measured by the ATP Determination Kit (Cat. No. A22066, Invitrogen) according to manufacturer's protocol. 56 Luminescence was measured by Spark 20M multimode microplate reader (Tecan). For normalization, mitochondrial protein was measured by bicinchoninic acid (BCA) assay (Cat. No. 23225, Thermo Scientific). For mitochondrial DNA quantitation, isolated mitochondria were required to be lysed in mitochondrial lysis buffer (Cat. No. C3601-4, Beyotime) followed by lyophilization (50 C for 60 min) and ethanol precipitation (À20 C for 10 min) to obtain DNA. By checking the purity at OD260/OD280 ratio ≥1.8, total mitochondrial DNA was measured using a NanoDrop One Microvolume UV-Vis Spectrophotometers (Thermo Scientific).

| Generation of mitochondria transferred Y74-ADSCs
The recipient Y74-ADSCs (1 Â 10 4 ) were seeded onto a 96-well plate and cultured in low-glucose DMEM (Gibco) with FBS (10%, v/v; Gibco) and P/S (1%, v/v; Gibco) of 3 h for adherence at 37 C with 5% CO 2 . Subsequently, serum-free low-glucose DMEM was replaced for another 2 h incubation to enhance mitochondrial uptake efficiency. After the removal of medium and washing with PBS, fresh isolated mitochondria with different concentrations (1.85-18.5 μg/ml) were plated on top of Y74-ADSCs with Minimum Essential Medium (MEM) without calcium for up to 12 h coincubation at 37 C with 5% CO 2 . After that, any mitochondria not transferred into cells were removed. Finally, the resultant cells were cultured in serum-containing medium of 12 h for recovery.
Optimization of these bioengineering parameters is further detailed on the following high-content analysis. A recent study pointed out that bioenergetics of recipient cells after mitochondrial transplantation significantly decreased with increasing passage number and returned to physiological levels. 57 For this reason, in our study, mitochondria transferred Y74-ADSCs were used for direct characterizations and therapeutics without further expansion. For quantitative analysis, image segmentation was performed by the GE Developer software. 58 Briefly, cells were segmented using the "find nuclei" by the Hoechst blue channel and "find cyto-   were fragmented in collision-induced dissociation with normalized collision energy of 30%, intensity threshold 1000, and one microscan. A tandem mass spectrum of specific fragment ions for each peptide was obtained after the detection, isolation, and fragmentation.

| Secretome analysis
For quantitative analysis, we employed a LIMMA-based analysis pipeline. 59 Briefly, proteomics data were normalized using the nor-malizeBetweenArrays function and a linear model for each protein was fit via the lmFit function. Proteins with foldchange >2 and p-value < 0.5 were defined as differentially expressed proteins. To reveal minor differences in secreted factors between mitochondria transferred Y74-ADSCs and control Y74-ADSCs from the GO terms, mass spectrometry-based spectral count approach by the number of peptide spectrum matches (PSMs) was employed for each sample as an expression of its relative abundance. The sum of PSM in each sample (n = 3) was used to normalize the number of PSM.  DESeq2 was applied to identify differential expression genes (DEGs). 60 The DEGs between two samples were chosen using the criteria as below: the logarithmic of fold change was >2 and the false discovery rate (FDR) was <0.05. To explore the functions of the differentially expressed genes, GO functional enrichment analysis was done by clusterProfiler. 61 GO terms and metabolic pathways at Bonferroni-corrected p-value < 0.05 were considered as significantly enriched among DEGs. To further analyze the biological impact of DEGs, protein-protein interaction networks were created using the STRING database.

| In vitro characterization of mitochondria transferred Y74-ADSCs
The conventional and mitochondria transferred Y74-ADSCs were seeded at density of 2000 cells/well under serum starvation conditions, respectively. The proliferation was evaluated by the Cell Cou- for 10 min at room temperature, followed by repeated washing.
Imaged cells were quantified using ImageJ. 63 Chondrogenesis was evaluated using U-bottom suspension culture system described previously. 64 Briefly, chondrocyte pellets were formed during 21 days culture in chondrogenic differentiation medium (Cat. No. SCM123, Sigma-Aldrich), followed by staining with Alcian Blue (Cat. No. A5268, Sigma-Aldrich). The optical density of the photograph was analyzed using ImageJ.

| The efficacy of mitochondria transferred ADSCs in wound healing
To avoid the immune response to human cells transplantation in animal model, mito-transferred rat-ADSCs were engineered for the following in vivo study using the optimized method mentioned above.
Briefly, adipose tissue was obtained from the right inguinal region of Sprague Dawley (SD) Rat A and B (male, 3 weeks, $200 g), respectively. Donor Rat A-ADSCs were expanded to isolate the mitochondria using the same method as described above. Mito-transferred rat-ADSCs were constructed by co-culture with exogenous mitochondria at 18.5 μg/ml per 10 4 of serum-starved Rat B-ADSCs for 6 h.
Wound healing experiments by full-thickness excision were performed based on a reported protocol. 65 SD rats (male, 3 weeks, $200 g, n = 12) were used. The Ethics Committee of Zhejiang University (Ethics Code. ZJU20200166) legally approved all animal operations and experimental procedures. Anesthesia of rat was induced with 7% chloral hydrate. After shaving, depilatory cream was applied on the dorsal surface of the skin to remove hair completely.
Two full-thickness dorsal excisional skin wounds were made for each rat using a sterile 10 mm-diameter biopsy punch. The rats were divided into three groups: a saline group, a nontreated ADSCs group, and a mito-transferred ADSCs group. Total 6 Â 10 6 cells suspended in 1 ml PBS were locally injected into the fat layer of wound (n = 8 for each group). Rats were returned to cage after the recovery from anesthesia and daily monitored to evaluate health.
The distribution of mito-transferred ADSCs after 6 h posttransplantation was examined using in vivo imaging (IVIS Spectrum) by detecting MitoTracker Red CMXRo labelled exogenous Rat Aderived mitochondria. The existence of exogenous Rat A-derived mitochondria was also checked its fluorescence using tissue section.
Briefly, the fixed adipose tissue was embedded in optimal cutting temperature (OCT) compound and frozen at À70 C. The frozen samples were sliced into 20 μm-thick sections using Leica CM 1950 cryostat and processed them for fluorescence scanning using Eclipse Ni-U microscope (Nikon). Later, adipocytes were further permeabilized with 0.5% Triton-X100 (Cat. No. T8787, Sigma-Aldrich) in PBS for 30 min and blocked with 5% bovine serum albumin in PBS for 1 h at room temperature. Cell nuclei were stained with DAPI (Cat. No. D9542, Sigma-Aldrich) in PBS for 30 min at room temperature, and filamentous actin was labeled by using phalloidin-iFluor 488 (Cat. No. ab176753, Abcam) for 1 h. Immunofluorescence images were acquired with Zeiss LSM 880 confocal microscope with 20Â objective.
To correct the distance between the animals and the camera, a ruler reference was placed when taking the images. After determining the area of the reference circle by ImageJ, the wound area was quan-

| Statistical analysis
The quantitative data were expressed as means ± SD. Two-tailed Student's t tests were conducted to analyze the statistical differences between two groups, while the statistical differences among multiple groups were analyzed by analysis of variance (ANOVA) using GraphPad Prism 7.04 (San Diego, USA). The p values are shown in the figures as *p < 0.05, **p < 0.01, and ***p < 0.001.

ACKNOWLEDGMENTS
The authors would like to thank Miss S.F. Xie at Westlake University of Hangzhou for LC/MS-MS instrumental support.