Exosomes exert cardioprotection in dystrophin-deficient cardiomyocytes via ERK1/2-p38/MAPK signaling

As mediators of intercellular communication, exosomes containing molecular cargo are secreted by cells and taken up by recipient cells to influence cellular phenotype and function. Here we have investigated the effects of exosomes in dystrophin-deficient (Dys) induced pluripotent stem cell derived cardiomyocytes (iCMs). Our data demonstrate that exosomes secreted from either wild type (WT) or Dys-iCMs protect the Dys-iCM from stress-induced injury by decreasing reactive oxygen species and delaying mitochondrial permeability transition pore opening to maintain the mitochondrial membrane potential and decrease cell death. The protective effects of exosomes were dependent on the presence of exosomal surface proteins and activation of ERK1/2 and p38 MAPK signaling. Based on our findings, the acute effects of exosomes on recipient cells can be initiated from exosome membrane proteins and not necessarily their internal cargo.

cellular signaling between cardiomyocytes, we used two DMD patient derived iPSC lines and one gene-edited iPSC in which the DMD gene was targeted by CRISPR/Cas9. The Dys1-iPSC line derived iCMs contain an DMD exon 3-6 deletion have been previously described and characterized 23 . The patient-derived Dys3-iPSC line was generated by reprogramming urine progenitor cells from a DMD patient harboring a DMD exon 3-6 mutation (Supplemental Fig. 1). The WT2-iPSC line was used to create the DysC-iPSC line by CRISPR/Cas9 targeting of DMD exon 1 to create a 6 bp deletion leading to dystrophin deficiency (Supplemental Fig. 2). WT1 and WT2-iCMs were included as non-disease controls. All iPSC lines were able to differentiate into cardiomyocytes and displayed typical cardiogenic markers 23 (Supplemental Fig. 3).
Dys-iCMs secrete paracrine signals that are protective for Dys-iCMs but not WT-iCMs. We first used a mitochondrial permeability transition pore (mPTP) opening assay as a functional screen to determine whether conditioned media had paracrine effects on WT1 and Dys1-iCM function. We have previously shown that Dys1-iCMs open the mPTP earlier than WT1-iCMs 23 and confirmed this phenotype in Fig. 1. Conditioned media from both WT1-and Dys1-iCMs, when applied to Dys1-iCMs, significantly delayed mPTP opening time compared to the vehicle group. Conditioned media from Dys1-iCMs did not change mPTP opening time in WT1-iCMs and conditioned media from WT1-iCMs delayed mPTP opening time in WT1-iCMs. This suggested differential paracrine effects between the WT1 and Dys1-iCM conditioned media.
We then characterized exosomes isolated from the conditioned media of WT1 and Dys1-iCMs. NTA and transmission electron microscopy revealed that WT1-iCM and Dys1-iCM exosomes (WT-exos and Dys-exos) were as small as 50 nm in diameter (Fig. 3a,b), but averaged 193 and 148 nm, respectively (Fig. 3c). Flow cytometric analysis of exosome-coated latex beads confirmed the presence of conventional exosome membrane markers, particularly CD63 and CD81 (Fig. 3d). The unlabeled exosome-coated beads without the primary antibody +/− the secondary antibody did not fluorescence (Supplemental Fig. 4). Furthermore, a surface protein array confirmed the presence of FLOT1, ICAM, ALIX, EpCAM, ANXA5, TSG101 on the surface of both WT-exos and Dys-exos (Fig. 3e). Using confocal microscopy, we show that iCMs successfully took up PKH26-labeled exosomes Figure 1. Treating Dys1-iCM with conditioned media delays opening of mPTP. Conditioned media collected from cardiomyocytes and added 2 hours prior to inducing mPTP formation with laser scanning confocal microscopy. WT1-iCM media, but not Dys1-iCM media delayed mPTP in WT1-iCMs. Both WT1-and Dys1-iCM media delayed mPTP formation in Dys1-iCMs. *p < 0.05 vs. vehicle, n = 3/group. by 2 hours (Fig. 3f). Supplemental Video 1 shows Z-stack imaging of cardiomyocytes with labeled exosomes distributed throughout the cytoplasm and nucleus.
Cardiomyocyte-exosomes protect against stress-induced injury in Dys-iCMs. We then investigated whether WT-exos and Dys-exos were responsible for the functional effects of conditioned media. Previously, it has been shown that Dys-iCMs were especially vulnerable to cellular stress by increasing ROS levels to increase cell death 23,24 . Therefore, we used ROS levels and cell death as endpoints when conducting a dose concentration curve. Five µL containing approximately 2.25 × 10 7 exosomes displayed the optimal cardioprotective properties in Dys-iCMs (Supplemental Fig. 5). This dose was used for all further studies. Exosomes acutely decreased injury-induced ROS levels in all Dys-iCMs compared to vehicle control ( Fig. 4a,b). Exosomes isolated from the conditioned media of dermal fibroblasts (fibro-exos) were used as an inert exosome control 25 . However, fibro-exos were not consistently inert showing some ability to protect the iCMs against stress-induced ROS, but not as significantly as cardiomyocyte exosomes, suggesting that the cardioprotective paracrine signaling was enhanced in cardiomyocyte exosomes.

Cardiomyocyte exosomes protect against mitochondrial triggers of apoptosis in
Dys-iCMs. We next investigated whether WT-and Dys-exos could mitigate stress-induced mitochondrial apoptotic pathways. Bax translocation to the mitochondria, with the loss of the mitochondrial membrane potential, can trigger opening of the mitochondrial permeability transition pore followed by caspase activation and cell death 26 . WT-and Dys-exos inhibited Bax expression and mitochondrial translocation in Dys1-iCMs in contrast to vehicle treated Dys-iCMs subject to stress (Fig. 5a). Stress does not cause Bax translocation in WT1-iCMs (Supplemental Fig. 6). Stress induces a loss of mitochondrial membrane potential (Fig. 5b,c) and an earlier mPTP opening time (Fig. 5d,e) in Dys1, Dys3 and DysC-iCMs, all of which is improved following exosome exposure and in comparison, to WT1 and WT2-iCMs. Both WT-and Dys-exos decreased the levels of caspases 3/7 in Dys1-iCMs ( Fig. 5f) but had no effects on WT1-iCMs. Both WT-and Dys-exos decrease stress-induced cell death as detected by propidium iodide staining in all Dys-iCMs (Fig. 5g,h). Fibro-exos mildly decreased cell death in Dys1, Dys3, DysC and WT1-iCMs but not to the same extent seen with cardiomyocyte exos. Interestingly, fibro-exos exacerbated stress-induced cell death in WT2-iCMs (Fig. 5h).
In summary, these data reveal that both WT-and Dys-exosomes decrease translocation of Bax to the mitochondria, preserve the mitochondrial membrane potential, delay the mPTP opening time, and decrease caspase 3/7 activity along with cell death. Dys-iCMs are more susceptible to the protective effects of both WT-and Dys-exos, whereas Dys-exos do not completely protect WT-iCMs against stress. Cardioprotective effects of cardiomyocyte exosomes depend on the ERK1/2 and p38 MAPK signaling pathway. Exosomal transfer of microRNA has been shown to protect cardiomyocytes by controlling cell survival gene expression 27 . However, considering the iCMs were exposed to exosomes for only 2 hours, we did not expect gene expression changes, but rather surmised exosomes would exert protective effects through an exosome surface ligand to activate cardiomyocyte cell survival pathways. To investigate whether an exosomal surface protein was required for its cardioprotective properties, we cleaved the surface proteins from Dys-exos with trypsin. Protein array analysis confirmed the absence of exosome membrane proteins after trypsin digestion (Fig. 6a). Trypsin-treated exosomes labeled with PKH26 were intact and taken up into iCMs at 2 hr (Fig. 6b). When compared to the intact exosomes, the trypsin-treated exosomes did not decrease stress-induced ROS levels in Dys1 or Dys3-iCMs (Fig. 6c,d), indicating a role for exosomal surface proteins in triggering the protective effect of exosomes. The mitogen-activated protein kinases (MAPK) including ERK1/2 and p38 MAPK coordinate responses that dictate cell death or survival 28 . ERK1/2 phosphorylation was stimulated at 30 min after exposure to both WT and Dys-exos (Fig. 7a). Trypsinized exosomes failed to stimulate ERK1/2 phosphorylation (Fig. 7b). U0126, a specific inhibitor of MEK1, was used to examine the potential role of the ERK1/2 pathway in WT and Dys-exos cardioprotection of iCMs. ERK1/2 inhibition abolished the protective effects of WT and Dys-exos on stress-induced ROS increases in Dys1-iCMs (Fig. 7c), mitochondrial membrane potential (Fig. 7d), mPTP formation ( Fig. 7e) and cell death (Fig. 7f). ERK1/2 inhibition alone with U0126 exacerbated the stress-induced increase in cell death (Supplemental Fig. 7) implicating its anti-apoptotic role in Dys-iCMs.
U0126 is known to block exosome uptake by disrupting lipid raft-mediated endocytosis 29,30 , therefore, we examined whether U0126 was inhibiting cardioprotection by blocking cellular uptake of exosomes. Supplemental Fig. 7 and Supplemental Video 2 show that cells pretreated with U0126 prior to the addition of PKH26-labeled exosomes were readily taking up the exosomes by 2 hours. This suggests that U0126 at this specific concentration did not impede exosome uptake, and therefore the reversal of the cardioprotective effect observed was likely due to the absence of an exosome surface protein to initiate ERK1/2 pathway activation.
We next examined the involvement of p38 MAPK with respect to the cardioprotective effects of exosomes. The basal phosphorylation of p38 MAPK was increased in Dys-iCMs compared to WT-iCMs (Fig. 8a). WT and Dys-exos stimulated phosphorylation of p38 MAPK at 30 min post exposure (Fig. 8a). Inhibiting p38 MAPK with SB203580 not only reversed the protective effects WT-exos and Dys-exos on Dys-iCMs but also exacerbated the stress-induced increase of ROS (Fig. 8b). SB203580 reversed the protective effects of WT-and Dys-exos on stress-induced loss of the mitochondrial membrane potential (Fig. 8c), mPTP opening ( Fig. 8d) and cell death (Fig. 8e). Inhibition of p38 MAPK alone further increased stress-induced ROS levels and cell death when compared vehicle treated groups (Supplemental Fig. 8).

Discussion
The present study was conceptualized to determine the paracrine effects of exosomes secreted from dystrophin-deficient cardiomyocytes on dystrophin-deficient cardiomyocytes. We initially postulated that Dys-exos would be detrimental to the functional effects on Dys-cardiomyocytes by exacerbating stress-induced cell injury. In vivo, the concentration or density exosomes in the intercellular space is unknown. Therefore, we started with a concentration-response curve to assess the functional effects of exosomes on Dys-iCMs. Unexpectedly, we found both WT-and Dys-exos were protective against stress-induced injury in the Dys-iCM rather than being detrimental. Therefore, we continued to investigate the mechanism downstream of the cardioprotective properties of exosomes.
We show that exosomes secreted from either WT-or Dys-iCMs protect the Dys-iCMs against stress induced increases in ROS levels, decreases in the mitochondrial membrane potential, and cell death. We have confirmed these findings in two unrelated DMD patient-specific iCMs (Dys1 and Dys3) along with a DMD gene-edited iCM lines (DysC) suggesting this is a robust effect of cardiomyocyte-specific exosomes on Dys-iCMs independent of iPSC clonal or genetic variation between iPSC lines. In addition, we provide supporting evidence that iCM-derived exosomes decrease Bax translocation to the mitochondria and caspase activation. Cardiomyocyte-derived exosomal cardioprotective effects were not replicated using exosomes secreted from dermal fibroblasts suggesting that cardiomyocyte-derived exosomes are enriched in cardioprotective factors.
Dermal fibroblast exosomes were previously shown to be an inert control for cardioprotection due to differences in microRNA cargo 16,25 compared to exosomes secreted from cultured stem cells and progenitor cells in vitro which are cardioprotective in in vitro and in vivo models of myocardial ischemia 18,31,32 . However, these studies focus on the transfer cargo inside the cell, mainly microRNA delivered by the exosome to alter gene expression to change the phenotype of the cardiomyocyte. In our study, we provide evidence that endogenous exosomes from dystrophin-deficient cardiomyocytes depend on a surface protein to retain its cardioprotective properties on the dystrophin-deficient cardiomyocyte suggesting that an exosomal surface protein may be involved in its paracrine effects. PKH26-labeled labelled exosomes taken up by the iCMs within the 2-hour exposure does not eliminate the possibility that microRNAs are contributing to the cardioprotective phenotype. However, cleaving the exosome surface proteins with trypsin and inhibiting ERK1/2 and p38 MAPK signaling pathways abrogates the exosome-mediated cardioprotection, which strongly suggests that these effects are mediated by a ligand-receptor interaction. However, we do not know whether this interaction occurs at the sarcolemma of the cardiomyocyte or after internalization of the exosomes. Exosomes are known to induce the phosphorylation of several downstream targets including p38 MAPK and ERK1/2 29,33,34 . Both activation of ERK1/2 and p38 MAPK signaling pathways are known to be acutely cardioprotective against cardiac stress by through anti-apoptotic mechanisms 35 . Both ERK1/2 and p38 MAPK have been demonstrated to form signaling modules by inhibiting GSK-3β 36 at the level of the mitochondria 37 to inhibit mPTP opening 38 . Both ERK1/2 and p38 MAPK have the ability to activate heat shock protein 27 which has previously been shown to play a role in exosome mediated protection of myocardial ischemia 39 . It has been shown that ERK1/2 is activated during lipid-raft dependent exosome uptake 29 . Specifically, the ERK1/2 inhibitor U0126 has been shown decrease exosome uptake in a dose dependent manner. In our study, the concentration of U0126 used blocks ERK1/2 phosphorylation and mitigates the cardioprotective effect of exosomes but does not block exosome uptake in the iCMs. Therefore, we do not know whether this interaction occurs at the sarcolemma of the cardiomyocyte or after internalization of the exosomes. A surface protein on the exosomes appeared to be involved in triggering acute protective signaling, as cleaving off these proteins with trypsin negated the cytoprotection of Dys-iCMs. For the first time, in this study, we provide evidence that exosome treatment stimulates cardioprotective signaling pathways in an in vitro model of dystrophin-deficient cardiomyopathy. This is a novel approach to understanding the mechanisms involved in this unique cardiomyopathy, and offers a way to further that understanding, as well as to investigate future therapies. Identifying the exosomal surface protein involved in initiating the cardioprotective effects may provide a novel target for the treatment of dystrophin-deficient cardiomyopathy. Studies to identify this protein are underway.
In summary, this study demonstrates that acute exposure of endogenous cardiomyocyte-secreted exosomes has the potential to protect against cellular stress in dystrophin-deficient cardiomyocytes. The protective pathways that are stimulated in this process include ERK1/2 and p38 MAPK. These signaling pathways are triggered by a surface receptor that is present on myocyte-secreted exosomes. Our results indicate that these pathways may be dysregulated in dystrophin-deficient cardiomyopathy, and offer a therapeutic target of interest in acute protection of the dystrophin-deficient heart against stress-induced injury.

Cell Lines and Cellular Reprogramming.
For this study, we used previously characterized induced pluripotent stem cell (iPSC) lines: a Dys1-iPSC line which contains an out-of-frame dystrophin gene deletion of exons 3-6 resulting in a null mutation with the complete absence of the dystrophin protein (SC604A/B-MD, Systems Biosciences; Mountain View, CA), a non-dystrophic wild-type iPSC line (WT1-iPSC) which was a generous gift from Dr. April Pyle 40 . A second dystrophin-deficient line (Dys3-iPSC) with dystrophin null mutation resulting from deletion of exons 3-6 was reprogrammed from patient-derived urine cells following the reprogramming procedure as described previously 41,42 . Briefly, Dys3 urine cells were reprogrammed using the CytoTune iPS Reprogramming Kit (Life Technologies, Carlsbad, CA) containing Sendai virus (SeV) vectors with OSKM factors at the multiplicity of infection (MOI) of 1.5. After the expansion and thorough selection of reprogrammed iPSC, purified Dys3-iPSC clones were established for further characterizations and evaluations. A separate wild-type iPSC line, HB53, was generously gifted by Dr. Ivor Benjamin and has been previously characterized 43 . HB53 was subjected to CRISPR gene targeting to introduce a mutation in the dystrophin gene which generated a dystrophin-deficient isogenic control iPSC line (DysC-iPSC) and untargeted cells were sub-cloned and served as an WT isogenic control iPSC line (WT2-iPSC).

Embryoid body formation and three germ layer differentiation.
To determine the pluripotent potentials of Dys3-iPSCs through spontaneous in vitro differentiation into three germ layers; ectoderm, mesoderm and endoderm, Dys3-iPSCs were cultured in suspension to form embryoid bodies (EBs) by hanging-drop protocol with STEMdiff ™ APEL ™ 2-LI Medium (StemCell Technologies, Vancouver, BC Canada) followed by adherent cell culturing on gelatin coated plates for 7days each. Dys3-iPSC derived spontaneously differentiated EBs were assessed for germ layer gene expression analysis by qRT-PCR for Nestin (ectoderm marker), Brachyury (mesoderm marker) and GATA4 (endoderm marker) (Supplemental Fig. 1). Primer sequences are listed in Table 1.

Dystrophin Genotypic Analysis.
To evaluate DMD gene mutations, genomic DNA from WT and Dys3-iPSCs was extracted with the QIAamp DNA Mini Kit (Qiagen, Valencia, CA) and DMD exons 3, 4, 6, 12 and 50 were amplified through PCR to screen the exon specific deletions in iPSC samples (Supplemental Fig. 1). The DMD exon specific primers used for the genotyping PCR are listed in Table 1.
Cardiac differentiation. Wild-type and Dys-iPSCs were differentiated into cardiomyocytes (iCMs) as described previously 23  Characterization of iPSC-derived Cardiomyocytes. Cardiac gene expression analysis of iPSC-derived differentiated cardiomyocytes was performed as described previously 23 . Briefly, total RNA samples were extracted using the miRCURY RNA Isolation Kit -Cell and Plant (Exiqon, Denmark) and complementary DNA (cDNA) samples were synthesized using the iScript cDNA synthesis kit (Bio-Rad, Hercules, CA). The cDNA templates were amplified for the expression of cTnT, GATA4, Nkx2.5, MYH6, MYH7 and GAPDH genes through RT-PCR analysis. Gene expressions were detected through agarose gel electrophoresis (Supplemental Fig. 3 Enhanced Green Fluorescent Protein Tagging of iPSC-derived cardiomyocytes. iPSC-derived cardiomyocytes were marked for live cell imaging as previously described 45 . In brief, 5 days after dissociation, iCMs were transduced with an NCX1-eGFP lentiviral construct 46  Exosome uptake into induced pluripotent stem cell derived cardiomyocytes. To assess exosome uptake by iCMs, exosomes from iCMs were labeled with PKH26 per the manufacturer's protocol. Cells were incubated with PKH26 labeled exosomes (5 µL). After 2 hours, z-stack images of cells were taken using a Nikon A1-R confocal microscope (Nikon Instruments, Melville, NY).

Stress-induced injury and treatment protocols.
Cells on glass coverslips were exposed to 100 μM H 2 O 2 in 10 mM deoxyglucose in RPMI minus glucose (Thermo Fisher Scientific, Waltham, MA) for 1 hour. Afterwards, cells were recovered in Maintenance media for 4 hours. This protocol mimics a transient metabolic/oxidative stress with recovery. Five µL of resuspended exosomes were added to iCMs for 2 hours prior to stress induction.
Superoxide and mitochondrial membrane potential staining. Cells were treated for 20 min with dihydroethidium (DHE, 10 µM) or tetramethylrhodamine ethyl ester (TMRE; 50 nM) to measure ROS levels and mitochondrial membrane potential (ΔΨ m ) respectively. The fluorescent intensity of the ethidium derivative or TMRE was detected by a laser ex/em 518/605 nm or ex/em 540/595 nm, respectively. The changes in membrane potential were monitored by calculating relative TMRE fluorescence. Five ROIs were selected from the nucleus (DHE) or mitochondria (TMRE) of GFP-positive iCMs and measured for mean fluorescent intensity (ImageJ, Version 1.48 v, Java 1.6.0_65, National Institutes of Health, Bethesda, MD). Imaging conditions such as gain levels, frames per second and aperture size were held constant.
Measuring mitochondrial permeability membrane transition pore (mPTP) opening. As described previously, cells were loaded with TMRE for 25 min at room temperature 45 . On laser-illumination, TMRE generates ROS within the mitochondria leading to mPTP opening and visualized by loss of the TMRE fluorescence. Time required to induce mPTP opening was determined from ΔΨ m recordings. The peak signal value over the recorded region (50 µm 2 ) was normalized as 100% and the lowest value as 0%. After normalization, the time required for a 50% decrease in signal was calculated and denoted as Time (s).
Propidium iodide staining. Cell death was evaluated after a 24-hour recovery period, by labeling the cells with propidium iodide (PI, Thermo Fisher Scientific, Waltham, MA) per the manufacturer's protocol. PI staining was quantified as a proportion of PI positive nuclei versus total nuclei.