Mitochondrial Transplantation Attenuates Cerebral Ischemia-Reperfusion Injury: Possible Involvement of Mitochondrial Component Separation

Background Mitochondrial dysfunctions play a pivotal role in cerebral ischemia-reperfusion (I/R) injury. Although mitochondrial transplantation has been recently explored for the treatment of cerebral I/R injury, the underlying mechanisms and fate of transplanted mitochondria are still poorly understood. Methods Mitochondrial morphology and function were assessed by fluorescent staining, electron microscopy, JC-1, PCR, mitochondrial stress testing, and metabolomics. Therapeutic effects of mitochondria were evaluated by cell viability, reactive oxygen species (ROS), and apoptosis levels in a cellular hypoxia-reoxygenation model. Rat middle cerebral artery occlusion model was applied to assess the mitochondrial therapy in vivo. Transcriptomics was performed to explore the underlying mechanisms. Mitochondrial fate tracking was implemented by a variety of fluorescent labeling methods. Results Neuro-2a (N2a) cell-derived mitochondria had higher mitochondrial membrane potential, more active oxidative respiration capacity, and less mitochondrial DNA copy number. Exogenous mitochondrial transplantation increased cellular viability in an oxygen-dependent manner, decreased ROS and apoptosis levels, improved neurobehavioral deficits, and reduced infarct size. Transcriptomic data showed that the differential gene enrichment pathways are associated with metabolism, especially lipid metabolism. Mitochondrial tracking indicated specific parts of the exogenous mitochondria fused with the mitochondria of the host cell, and others were incorporated into lysosomes. This process occurred at the beginning of internalization and its efficiency is related to intercellular connection. Conclusions Mitochondrial transplantation may attenuate cerebral I/R injury. The mechanism may be related to mitochondrial component separation, altering cellular metabolism, reducing ROS, and apoptosis in an oxygen-dependent manner. The way of isolated mitochondrial transfer into the cell may be related to intercellular connection.


Introduction
Stroke is an acute cerebrovascular disease, including ischemic and hemorrhagic stroke, and is considered to be one of the leading causes of human death and disability worldwide [1][2][3][4]. Ischemic stroke accounts for over 80% of all strokes and is usually triggered by brain arterial embolism [3,5]. When blood flow is blocked, brain tissue in the area of blood supply becomes ischemic and hypoxic, which then leads to neurological dysfunction. Moreover, following blood reperfusion, the damaged brain tissue can be further harmed by restoration of oxygen-rich blood, causing a socalled ischemia-reperfusion (I/R) injury [3,[6][7][8][9]. Major contributors to the pathological process include overproduction of ROS, dramatically increased extracellular glutamate levels, and activation of neuroinflammation responses [6,7,9].
Among these, the dysfunction of mitochondria in neurons plays a pivotal role [3,9]. Currently, the main strategy to mitigate ischemic stroke injury is revascularization [4,8,10], which may lead to I/R injury. Despite remarkable progress that has been achieved for ischemic stroke, there seem to be no better options for I/R injury.
Recently, several studies have applied isolated mitochondria from various sources as an intervention in many diseases. Four studies focused on cerebral I/R injury have shown benefits of mitochondrial transplantation based on various phenotypes, such as behavioral assessment, infarct size, ROS, and apoptosis. However, the appropriate source of mitochondria, the mechanism of its therapeutic effect, and the fate of isolated mitochondria remain unclear. The clinical application of isolated mitochondria has just begun, and more safety and effectiveness assessments are needed.
In order to answer the above questions, we performed this study. Firstly, we evaluated the source of mitochondria and then assessed the therapeutic effects of mitochondrial transplantation in cellular and animal models. Finally, we mainly focused on the therapeutic mechanisms of mitochondrial transplantation and the fate of transplanted mitochondria.

Cells.
The mouse neural stem cell (mNSC) was obtained and cultured as previously described [43]. Adherent culture of mNSC was performed with Matrigel (Corning, NY, USA, Cat#354277). N2a (Cat#SCSP-5035) and induced pluripotent stem cell (iPSC) (Cat#DYR0100) were purchased from the National Collection of Authenticated Cell Cultures, Shanghai. 293T was provided by Dr. Gao Liu from Zhongshan Hospital, Shanghai Medical College, Fudan University. N2a and 293T cells were cultured in DMEM supplemented with 10% fetal bovine serum. The iPSC was cultured according to the manufacturer's protocol.

Animals.
Sprague-Dawley rats (7-8 weeks old, 250-300 g) were obtained from Shanghai Super-B&K Laboratory Animal Corp. Ltd. (Shanghai, China). All experimental procedures and animal care were approved by the Animal Welfare and Ethics Group, Laboratory Animal Science Department, Fudan University (ethical approval number 202006013Z) and were carried out according to the Guidelines for the Care and Use of Laboratory Animals by the National Institutes of Health. The rats were divided into three groups: sham, I/R, and I/R+Mito group (sham =sham-operated; I/R = MCAO+reperfusion with saline injection; I/R+Mito = MCAO+reperfusion with mitochondria injection).

Mitochondrial Isolation.
Mitochondria were isolated from N2a and mNSC using the mitochondria isolation kit (ThermoFisher Scientific, USA, Cat#89874) as previously described [32,44]. Briefly, after cultured cells were orderly digested (trypsin) and centrifuged (300 g, 5 min) and the supernatant was removed, collected cells were resuspended by mitochondrial isolation reagent A (800 μl) in a 2.0 ml microcentrifuge tube and vortexed for 5 s and then incubated for 2 min on ice. Then, the reagent B (10 μl) was further added into the tube and continuously placed in situ for 5 min. Following vortexed at maximal speed for 5 times (each time for 1 min), the reagent C (800 μl) was added into the tube and mixed. Subsequently, the mixed solution was centrifuged (700 g, 10 min, 4°C) and then the supernatant was obtained for further centrifugation (12000 g, 15 min, 4°C). Finally, fresh mitochondria were obtained and used for further experiments. For animal experiments, each rat received mitochondria isolated from 1 × 10 7 cells, and the protein content was about 180 μg-200 μg.

Transmission Electron Microscopy (TEM)
. Cells were fixed with 2.5% glutaraldehyde for 2 h at room temperature and then centrifuged (300×g, 5 min). Subsequently, cells were postfixed with precooled 1% osmic acid (2 h, 4°C) and then centrifuged again (300×g, 5 min). After gradient alcohol dehydration and penetration with a solution of acetone and epoxy resin at different proportions, the cell samples were further embedded into epoxy resin and solidified for 48 h. Subsequently, the embedded samples were sectioned (thickness: 60-100 nm) and then double-stained with 3% uranyl acetate and lead citrate. Finally, the stained sections were observed and imaged by TEM (Tecnai G2 20 TWIN, FEI Company, Oregon, USA).

Mitochondrial
Membrane Potential Analysis. The mitochondrial membrane potential (MMP/ΔΨm) was assessed by JC-1 dye (Beyotime Biotechnology, Shanghai, China, Cat#C2006) and detected by flow cytometry and confocal microscopy, according to previous methods [45,46]. For flow cytometry, single-cell suspensions of mNSC and N2a were prepared and then coincubated with JC-1 work solution for 20 min at 37°C. Next, sample cells were centrifuged (600 g, 4°C, 5 min) and washed with JC-1 buffer solution 2 2 Oxidative Medicine and Cellular Longevity times. Subsequently, resuspended cells were subjected to flow cytometry tests. For image, cells were seeded in glassbottom Petri dishes. 24 hours later, cells were coincubated with JC-1 work solution for 20 min at 37°C, washed with JC-1 buffer, and then examined by confocal microscopy.
2.6. Polymerase Chain Reaction (PCR). Absolute quantitative PCR was performed as previously described [47][48][49]. The ratio of mtDNA and nuclear DNA was used to assess relative mtDNA copy number. In this experiment, mt-ND1/β-globin and mt-RNR1/β-actin were used to represent abundance. The sequences of the primers are described in Table S1.  [54,55]. Anesthetized by 2% pentobarbital sodium (45 mg/kg), the rats were placed in a prone position. Then, the left common carotid artery, external carotid artery (ECA), and internal carotid artery (ICA) were exposed. Next, a silicon-coated monofilament suture was gradually inserted through the left ECA and was moved up into the left ICA to successfully occlude the left middle cerebral artery (MCA) and remained in situ for 120 min. Subsequently, the suture was carefully removed, the ECA was permanently ligated, and the incision was sutured. Sham group rats were subjected to the same procedure except for the 120 min occlusion of MCA. Experimental animals were then placed into individual cages and provided a standard diet and water. After 120 min occlusion, right before ICA reperfusion, the isolated mitochondria (from 1 × 10 7 cells, the protein content was about 180 μg-200 μg) or saline (10 μl) was injected into the ICA and all incisions were closed.
2.14. Neurobehavioral Evaluation. Neurobehavioral deficits were evaluated 24 h after mitochondrial transplantation using multiple scales, including the Clark general functional deficit score [56,57], the Clark focal functional deficit score [56,57], the modified neurological severity score (mNSS) [55,58], and the rotarod test [55,59]. Behavioral assessments were conducted by two skillful investigators who were both blinded to the animal groups.
2.15. Cerebral Infarct Area Detection. Triphenyl tetrazolium chloride (TTC) staining was used to display the area of cerebral infarction [60,61]. Briefly, 24 h after MCAO, the rats were deeply anesthetized and perfused with PBS transcardially, after which the rat brains were obtained and cut into 2 mm thick coronal sections. Subsequently, the brain sections were incubated with a 2% TTC solution at 37°C for 30 min in darkness. Then, stained slices were placed from the frontal to occipital order, and macroscopic images were obtained with a digital camera. Infarct areas were measured by Adobe Photoshop 21.0.0 (Adobe Systems Inc., San Jose, CA, USA).
2.16. Transcriptomic Analysis. RNA sequencing was performed as previously described [62,63]. Downstream analysis was performed by R (R Foundation for Statistical Computing, Vienna, Austria).  (f-i) Seahorse XF cell mitochondrial stress test; the performance of N2a and mNSC at 1 × 10 5 level (f), N2a exhibited higher oxidative respiratory activity (n = 6); quantitative analysis of basal (g) and maximal (h) OCR of two cells showed that N2a exhibits higher oxidative respiratory activity (n = 9). The performance of mitochondrial stress test form NSC at 1 × 10 5 (same data as in (f)) and 2 × 10 5 level showed mNSC reacted well and did not die in the comparison with N2a (i); we did the test with N2a 1 × 10 5 , mNSC 1 × 10 5 , and mNSC 2 × 10 5 at one time but presented them in two graphs due to the order of magnitude. * * p < 0:01.

4
Oxidative Medicine and Cellular Longevity labeled. The Lyso Dye (Dojindo Laboratories, Kumamoto, Japan, Cat#MD01)was used to label lysosomes according to the manufacturer's instruction.

Statistical Analysis.
Data that conform to a normal distribution with homogeneous variance are expressed as mean ± standard deviation (SD), and Student's t-test or one-way analysis of variance (ANOVA) was used to compare the differences between two groups or among multiple groups, respectively. Data with a nonnormal distribution are presented as median (25%, 75% quantiles), and Mann-Whitney U-test was taken into consideration. Statistical analysis and diagram generation were performed using GraphPad Prism 8.0.1 (GraphPad Software, Inc., San Diego,   5 Oxidative Medicine and Cellular Longevity CA, USA). * p < 0:05 and * * p < 0:01 were considered to be statistically different.

Mitochondrial Transplantation Improved Neurobehavioral Deficits and Reduced Infract Size of MCAO Rats.
To evaluate the effect of exogenous mitochondria on behavior and infarct size, we used the MCAO model. The Clark general/focal scale, the mNSS, and the rotarod test were used to assess neurological behavior deficits, and TTC staining was performed to measure infarct size. The results suggested that mitochondrial transplantation significantly improved neurological behavior deficits. For sham, I/R, and I/R+Mito groups, the sample size was 8, 9, and 7, and the Clark general scale was 0 (0,0), 3 (1.5,6), and 1 (0,1) (nonnormally distributed data are expressed as median, 25%, 75% quantile), which indicated that mitochondria improved neurological outcome (p < 0:05). The Clark focal scale and mNSS confirmed the following: Clark focal: 0 ± 0,

Effects of Mitochondrial Transplantation on Transcriptomic
Profile Imply Metabolic Alteration. In order to clarify the effects of mitochondrial transplantation on transcriptomic profile and its possible mechanism, we performed RNA sequence analysis. The results identified 14 upregulated genes and 12 downregulated genes between control and H/R groups, 27 upregulated genes and 98 downregulated genes between control and H/R+Mito groups, and 17 upregulated genes and 1 downregulated gene between H/R and H/R+Mito groups. The overlapping genes are presented in Figure 4(c). The following KEGG pathway enrichment analysis suggested exogenous mitochondria may affect metabolism-related pathways, especially lipid metabolism-related molecules and pathways such as the PPAR signal pathway, insulin signal pathway, fat intake and digestion-related pathway, cholesterol metabolism, glycolysis, and gluconeogenesis (Figure 4(d)). These results indicated that exogenous mitochondria may be

Pattern of Exogenous Mitochondrial Transfer Implies
Mitochondrial Component Separation. To better understand the mechanism of exogenous mitochondria, we first traced its pathway. MitoTracker™ Red CMXRos and MitoTracker™ Green FM were used to label mitochondria. When the mitochondria of N2a labeled with MitoTracker Green were isolated and added to the medium of another N2a which was labeled with MitoTracker Red, a few hours later, we found that the red and green were completely fused ( Figure 5(a)). This phenomenon implied that exogenous mitochondria can be fused with endogenous mitochondria of the host cell. Next, we isolated red dye-labeled mitochondria and added them to the medium of green dye-labeled cell. The result validated the previous (Figure 5(b)). To figure out if this fusion property of exogenous mitochondria is species-limited, we isolated red dye-labeled mitochondria from a human-derived cell line-U87 and added it to mNSC

12
Oxidative Medicine and Cellular Longevity medium. Mitochondria fused again ( Figure 5(c)). This suggested that the ability of exogenous mitochondria to fuse with host cell mitochondria is cross-species.
To further verify the fusion property of exogenous mitochondria, we constructed a 293T cell overexpressing COX8A N-terminal signal peptide-mCherry to labeled mitochondria. And this tag had little effect on cell and mitochondrial function ( Figure S5). When the red-mitochondria 293T cell was labeled with MitoTracker Green again, we found all the mitochondria have both red and green markers (Figure 6(a)). The mitochondria isolated from the double-labeled cell also showed the colocation of the two colors ( Figure 6(b)). When we added the double-labeled isolated mitochondria to the medium of a 293T cell labeled with MitoBright Deep Red (set to pink), we came to an interesting result. The pink mitochondria (endogenous mitochondria of host cell) completely colocalized with exogenous green mitochondria, while only a portion of pink mitochondria overlapped with exogenous red mitochondria. Moreover, the two-color marker of the same exogenous mitochondria was partially separated (Figure 6(c)). All of these suggested that the exogenous mitochondrial components segregate when cocultured with host cells and a specific part can fuse with the endogenous mitochondria of host cells and the rest part has another fate. During the mitochondrial transfer process, we found that the green component can be internalized immediately by the host cell within 1 h, while the red component was internalized in a much slower way (Figures 7(a)-7(e)). This again confirmed the different fate of different mitochondrial components. We further investigated the pattern of exogenous mitochondrial 13 Oxidative Medicine and Cellular Longevity transfer using double-labeled 293T as a mitochondrial donor and pink-labeled induced pluripotent stem cell (iPSC) as host. The results showed nearly all the green components colocalized with the pink endogenous mitochondria and the red component was concentrated at the edge of the cell clones or in the scattered cells (Figures 8(a)-8(c)). These suggested the internalization efficiency of red component may be related to intercellular connections.

The Fate of Exogenous Mitochondria Is Fusion and
Lysosomal Degradation. To prove the theory that a portion of exogenous mitochondria can fuse with endogenous mitochondria, we assessed the mitochondrial dynamics by WB analysis. Our results suggested that after H/R treatment, the expression levels of the mitochondrial fusion-related proteins MFN1 (control vs. H/R, p < 0:01) and OPA1 (control vs. H/R, p < 0:01) were dramatically reduced and mitochondrial fission-related protein DRP1 was significantly increased (control vs. H/R, p < 0:01). Exogenous mitochondrial intervention alleviated this process and increased the expression of MFN1 (H/R vs. H/R+Mito, p < 0:01) and OPA1 (H/R vs. H/R+Mito, p < 0:01) but did not significantly reverse DRP1 expression (Figures 9(b)-9(e)). The above data further confirm the fusion property of exogenous mitochondria.
To figure out the fate of the unfused, COX8A N-terminal signal peptide-mCherry fusion protein-labeled mitochondria, we did lysosomal staining with Lyso Dye. Interestingly, the red component of mitochondria is totally colocalized with lysosomes of the host cell (Figure 9(a)), suggesting the fate of unfused components of exogenous mitochondria is lysosomal degradation.

Discussion
There are four studies [30,31,64,65] focused on the treatment of cerebral I/R injury with isolated mitochondria, according to the latest review [66]. All of them showed a favorable outcome in behavioral assessment or cerebral infarct size with different mitochondrial donors by intravenous or intracerebroventricular injection. However, there is no more detailed information about the source of mitochondria, the mechanism, and the fate of isolated mitochondria. And this is the information we want to provide.
In order to apply mitochondrial transplantation therapy to the clinic, the first priority is the source and quality control of mitochondria. The ideal source of mitochondria is one that is readily available and can be amplified in large numbers. Both stem cells and tumor cells meet this requirement. Therefore, we chose N2a and mNSC as mitochondrial 14 Oxidative Medicine and Cellular Longevity source cells to assess a range of mitochondrial characteristics. Previous studies isolated mitochondria from the placenta [65], or human umbilical cord-derived mesenchymal stem cells [64], or pectoralis major muscle [30], or baby hamster kidney fibroblast [31], and evaluated them mainly by MMP or respiratory activity. The outcome is closely related to the isolation and preservation process, and the consistency may not be good among different batches. The four studies did not give the reason why they choose these cells as mitochondrial donors. However, in our study, we evaluated the mitochondria before isolation through multiple dimensions, including morphology, MMP, mtDNA copy number, respiratory activity, metabolomic profile, and tumorigenicity. We hope to provide reference data when choosing a mitochondrial donor in future research. For the first time, we have identified the oxygen dependence of therapeutic effects of isolated mitochondria. This reminds us of the application scenario of isolated mitochondria, where incorrect application may lead to serious consequences. It is generally believed that exogenous mitochondria have a relatively intact function and can replace the damaged mitochondria in the host cell [67]. Considering the oxygen dependence, we presume that the exogenous mitochondria are a load for the cell.
In the presence of oxygen, the cell is able to handle this load and make it functional using the large amount of ATP produced by oxidative phosphorylation. However, in hypoxic conditions, host cells require additional energy to handle this load, which accelerates cell death. Also, according to our transcriptome data, the exogenous mitochondrial function is closely related to lipid metabolism, which may increase its oxygen dependence. In fact, there is little known about oxygen dependence, and this will be one of our future research directions.
The therapeutic effects of isolated mitochondria in our cell and animal models are similar to other studies [30,31,64,65], which showed that mitochondrial intervention attenuated I/R injury, improved neurological outcomes, and reduced cerebral infarct size. Our data again confirmed the potential clinical application of mitochondrial transplantation. The transcriptomic data suggest that the therapeutic effect of mitochondria may be related to altered metabolism, especially lipid metabolism, providing clues for future mechanistic studies. Few studies focused on the behavior of isolated mitochondria in vivo, especially whether it can cross the brain-blood barrier. Nakamura et al. [65] injected mitochondria intravenously and found exogenous mitochondria distributed in the brain under ischemic-reperfusion condition. Shi et al. [36] injected isolated mitochondria intravenously in mice and found that the exogenous mitochondria distributed in various tissues including the brain, liver, kidney, muscle, and heart. However, we did not find that mitochondria can pass the intact blood-brain barrier in our projects. More research is needed on the permeability of the blood-brain barrier to mitochondria.
The discovery of mitochondrial component separation phenomenon was based on different mitochondrial labeling techniques, and this gives us a new perspective to study the behavior of mitochondria. To our knowledge, most studies [25,30,31,33,36,64,65] labeled mitochondria with a single MitoTracker dye or a fluorescent fusion protein and got a conclusion based on that. However, we used both techniques to label the same mitochondria. Surprisingly, we found the different markers are separated. Regardless of whether it was technical or not, at least, it proved that a different conclusion may be made based on a different single-label method. Therefore, previous works need to be revisited. This is one of the important information provided in this article. MitoTracker dyes are roughly divided into voltagedependent and non-voltage-dependent. We chose the nonvoltage-dependent dye MitoTracker Green to label mitochondria, which are covalently bound to the free sulfhydryl group of cysteine in mitochondrial protein [68,69]. Therefore, the possibility of dye transfer between mitochondria is extremely low, and no studies have reported this dyetransfer phenomenon. Thus, MitoTracker Green represents the mitochondrial component that binds to it. Fluorescent fusion proteins (most fused with mitochondrial targeting sequence of cytochrome c oxidase subunit VIII) are another commonly used method of labeling mitochondria [33,36,37]. Due to the wide distribution in the mitochondria, both labeling methods can display mitochondria and are well overlapped (Figure 6(a)). When the extracted doublelabeled mitochondria enter the host cell, the two markers are separated, which represents that different component (not subgroups) of mitochondria has different fates. Considering the lysosomal labeling and mitochondria dynamic protein Western blot experiments, it is indicated that the isolated mitochondria may function by fusing its useful part to the host mitochondria rather than replacing it entirely. The unfused part will enter the lysosome for degradation. Furthermore, we used iPSC to study the effect of cell connections on the entry of mitochondria into host cells. The result implied that tight intercellular connections will greatly reduce the red component of isolated mitochondria from entering the host cell.
Few studies have covered the fate of isolated mitochondria after being internalized by host cells. Cowan et al. [70] reported that after being incorporated into host cells, isolated mitochondria are transported to endosomes and lysosomes, and then, most of these mitochondria escape from the compartments and fuse with the endogenous mitochondrial network. Their work described the fate of exogenous mitochondria, treating mitochondria as a whole, while our work found that the different part of mitochondria has different fate, which is consistent with the interaction of components between intracellular membrane systems. This result reminds us to understand the behavior of mitochondria from a more microscopic perspective and to pay more attention to its communication with other organelles.
Based on our data, we hypothesized that when the exogenous mitochondria entered into the host cells, in the presence of oxygen, mitochondrial component separation occurred, reducing ROS levels, apoptosis, and altering cellular metabolism, thus improving cell survival ( Figure 10). We made a reasonable assumption. Whether correct or  Figure 10: Hypothetical model diagram. When isolated exogenous enters host cells, the specific components of it undergo separation. Some fuses with the host mitochondria, while the other enters lysosome and undergoes lysosomal degradation. This process reduces ROS and apoptosis and alters metabolic profile, which in turn attenuates ischemia-reperfusion injury. 16 Oxidative Medicine and Cellular Longevity not, it is of great significance, because this phenomenon will allow us to reexamine previous studies and consider this factor in future research design. Still, there are limitations. The correctness of the conclusion is closely related to the mitochondrial label methods. When previous studies labeled mitochondria by a single method, they got incomplete information. The current label methods are mainly developed to display mitochondria in cells. Unpredictable events may happen when mitochondria are isolated. Therefore, a new fate tracking tool should be developed. Another limitation would be lacking proper controls in the fate experiments. Due to the superficial knowledge of the event, we had no idea where to intervene. Although we used a mitochondrial fission inhibitor, Mdivi-1, as a control ( Figure S6), it did not seem to show any difference. Another similar study did not set up controls too [70]. Therefore, our future direction is to develop new mitochondrial label tools and clarify the fate and transportation of isolated mitochondria through proper controls. In addition, the mechanism conclusions need to be verified in vivo.
In general, studies focusing on mitochondrial transplantation therapy are in their infancy, but existing data indicate a promising clinical application. When there is no effective treatment for ischemia-reperfusion injury, mitochondrial transplantation therapy provides a new idea. However, more data is needed to confirm its safety and efficacy, and more mechanism studies are needed to figure out how it works. We hope our study provides useful information in this area and has enlightenment for future studies.

Conclusions
Mitochondrial transplantation may attenuate cerebral I/R injury. The mechanism may be related to selective mitochondrial component separation, altering cellular metabolism, reducing ROS, and apoptosis in an oxygen-dependent manner. The way of isolated mitochondrial transfer into the cell may be related to intercellular connection.    Figure S5: the influence of the mitochondrial tag on cell and mitochondrial functionality. (A) Cell growth curve was detected by a CCK-8 assay between 293T (control) and 293T expressing COX8A gene Nterminal signal peptide-mCherry fusion protein (Mito-mCherry) (n = 6); (B) ROS levels of two cells detected by DCFH-DA probes through a fluorescence plate reader (n = 6), Rosup is a positive control drug; (C) ATP levels of two cells detected a luminescence plate reader (n = 6).  Table S1: related primer sequences for absolutely quantitative PCR analysis.