Potent Paracrine Effects of human induced Pluripotent Stem Cell-derived Mesenchymal Stem Cells Attenuate Doxorubicin-induced Cardiomyopathy

Transplantation of bone marrow mesenchymal stem cells (BM-MSCs) can protect cardiomyocytes against anthracycline-induced cardiomyopathy (AIC) through paracrine effects. Nonetheless the paracrine effects of human induced pluripotent stem cell-derived MSCs (iPSC-MSCs) on AIC are poorly understood. In vitro studies reveal that doxorubicin (Dox)-induced reactive oxidative stress (ROS) generation and cell apoptosis in neonatal rat cardiomyocytes (NRCMs) are significantly reduced when treated with conditioned medium harvested from BM-MSCs (BM-MSCs-CdM) or iPSC-MSCs (iPSC-MSCs-CdM). Compared with BM-MSCs-CdM, NRCMs treated with iPSC-MSCs-CdM exhibit significantly less ROS and cell apoptosis in a dose-dependent manner. Transplantation of BM-MSCs-CdM or iPSC-MSCs-CdM into mice with AIC remarkably attenuated left ventricular (LV) dysfunction and dilatation. Compared with BM-MSCs-CdM, iPSC-MSCs-CdM treatment showed better alleviation of heart failure, less cardiomyocyte apoptosis and fibrosis. Analysis of common and distinct cytokines revealed that macrophage migration inhibitory factor (MIF) and growth differentiation factor-15 (GDF-15) were uniquely overpresented in iPSC-MSC-CdM. Immunodepletion of MIF and GDF-15 in iPSC-MSCs-CdM dramatically decreased cardioprotection. Injection of GDF-15/MIF cytokines could partially reverse Dox-induced heart dysfunction. We suggest that the potent paracrine effects of iPSC-MSCs provide novel “cell-free” therapeutic cardioprotection against AIC, and that MIF and GDF-15 in iPSC-MSCs-CdM are critical for these enhanced cardioprotective effects.

Excessive ROS generation and cardiomyocyte apoptosis have been proposed as the major mechanisms of AIC. We further investigated whether the cardioprotective effects of MSCs-CdM are mediated through attenuation of Dox-induced oxidative stress and apoptosis in NRCMs. In living cell culture of NRCMs, 2′ , 7′ -Dichlorodihydrofluorescein dilacerate (H 2 DCFDA) and terminal deoxynucleotidal transferase mediated dUTP nick end-labeling (TUNEL) staining was used to measure ROS generation and apoptosis, respectively, in the presence or absence of MSCs-CdM under challenge of Dox for 24 hours (Fig. 1b,c). H 2 DCFDA staining revealed that the ROS score was significantly increased approximately five fold in the Dox group compared with the control group ( Fig. 1b- TUNEL staining revealed that apoptosis of NRCMs in the Dox group was significantly increased compared with the control group ( Fig. 1c-i,ii, 31.75 ± 3% vs. 1.75 ± 0.9%; P < 0.01). Administration of CdM derived from BM-MSCs or iPSC-MSCs significantly decreased apoptosis of NRCMs compared with the Dox group ( Fig. 1c-ii; P < 0.01). Notably, both the fluorescence intensity of H 2 DCFDA staining ( Fig. 1b-ii; P < 0.01) and the apoptotic rate of NRCMs ( Fig. 1c-ii, 7.25 ± 1.25% vs. 20.75 ± 1.7%; P < 0.01) were significantly lower following administration of iPSC-MSCs-CdM than following BM-MSCs-CdM. The concentration of MDA in the iPSC-MSCs-CdM group was also significantly lower than in the BM-MSC-CdM group (Fig. 1b-iii, 1.45 ± 0.04 μ mol/g protein vs. 2.69 ± 0.07 μ mol/g protein; P < 0.01). The cardioprotection afforded by iPSC-MSCs-CdM was dose-dependent (Fig. 1d). Concentration of MDA generation was decreased with increased concentration of iPSC-MSCs-CdM treatment in NRMCs These findings suggest that soluble factors secreted by MSCs in the CdM are likely to confer important protective effects against Dox-induced oxidative stress and apoptosis in NRCMs via a paracrine 3 μ m pore size transwell inserts and 1 × 10 4 NRCMs were seeded on the bottom of a 96-well culture plate with or without presence of 1 μ M Dox. 24 hours later, the transwell inserts were removed and MTT assay was used to measure NRCMs cell viability (i). The percentage of cell viability of each group was calculated from triple experiments (ii) (*p < 0.01 vs. control group; # p < 0.01 vs. Dox group; † p < 0.01 vs. BM-MSCs-CdM group). b,c) NRCMs were treated with or without MSCs-CdM (BM-MSCs-CdM or iPSC-MSCs-CdM) under challenge of 1 μ M Dox for 24 hours. In living cell culture conditions, H2DCFDA (b-i) and TUNEL (c-i) staining were used to measure ROS generation and cell apoptosis respectively. Quantitative measurement of ROS (b-ii) and apoptotic rate (c-ii) was counted from five viewing fields in each group and triple experiments were performed (*p < 0.01 vs. control group; # p < 0.01 vs. Dox group; † p < 0.01 vs. BM-MSCs-CdM group). The concentration of MDA (b-iii) among the different groups was also measured (*p < 0.01 vs. control group; # p < 0.01 vs. Dox group; † p < 0.01 vs. BM-MSCs-CdM group). d) The effect of iPSC-MSCs-CdM showed a dose-dependent manner in attenuation of ROS generation and apoptosis induced by Dox. The concentration of MDA (d-i) and apoptosis (d-ii) among the different groups were also measured among the different groups was measured (*p < 0.01 vs. control group; # p < 0.01 vs. Dox group; † p < 0.01 vs. iPSC-MSCs-CdM-10 μ l group; $ p < 0.01 vs. iPSC-MSCs-CdM-20 μ l group).

Administration of iPSC-MSCs-CdM attenuates Dox-induced cardiomyopathy in mice.
The animal experimental protocol is outlined in Fig. 2a. As shown in Fig. 2b, left ventricular ejection fraction (LVEF), fractional shortening (FS) and + dp/dt were significantly decreased in the Dox group at week 0 compared with the control group. This confirmed successful establishment of an animal model of Dox-induced cardiomyopathy ( Fig. 2b; P < 0.01). At 3-weeks, left ventricular systolic pressure (LVSP), + dp/dt and slope of end systolic pressure volume relationship (ESPVR) were significantly decreased in the Dox group compared with the control group (Fig. 2c,d-i-iii; P < 0.01). In contrast, LVSP, + dp/dt and slope of ESPVR were significantly improved in the BM-MSCs-CdM group and iPSC-MSCs-CdM group at 3-weeks compared with the Dox group (Fig. 2c,d-i-iii; P < 0.01). LVSP, + dp/dt and slope of ESPVR were also significantly higher in the iPSC-MSCs-CdM group than the BM-MSCs-CdM group (Fig. 2d-i-iii; P < 0.01), indicating that iPSC-MSCs-CdM better aids recovery of heart function in AIC. Notably, both BM-MSCs-CdM and iPSC-MSCs-CdM groups, but not the Dox group, showed significant improvement in + dp/dt at 3-weeks compared with week 0 ( Fig. 2e; P < 0.01).
Hematoxylin and eosin (HE) staining revealed a dilated LV chamber, the typical feature of AIC, in the Dox group ( Fig. 3a-ii) compared with the control group ( Fig. 3a-i). Nonetheless injection of both BM-MSCs-CdM and iPSC-MSCs-CdM significantly attenuated LV dilatation induced by Dox ( Fig. 3a-iii,iv).
Cardiac fibrosis was assessed by Sirius red staining ( Fig. 3a-v-viii). The ratio of cardiac fibrosis was greatly increased in the Dox group compared with the control group (Fig. 3b, 8.1 ± 0.4% vs. 0.3 ± 0.08%; P < 0.01). This ratio of cardiac fibrosis was significantly decreased by 22.2% in the BM-MSCs-CdM group and by 36.8% in the iPSC-MSCs-CdM group compared with the Dox group (P < 0.01). Notably, the ratio of cardiac fibrosis was also significantly lower in the iPSC-MSCs-CdM group than the BM-MSC-CdM group (5.1 ± 0.3% vs. 6.4 ± 0.3%; P < 0.01), indicating that iPSC-MSCs-CdM is superior to BM-MSCs-CdM in attenuation of Dox-induced cardiac fibrosis (Fig. 3b).
These results demonstrate that injection of BM-MSCs-CdM or iPSC-MSCs-CdM significantly improves LV function and decreases myocardial fibrosis in a mouse model of Dox-induced cardiomyopathy. In addition, iPSC-MSCs-CdM is more effective than BM-MSCs-CdM.
Attenuated Dox-induced cardiomyopathy of mice is linked to reduced oxidative stress and cardiomyocyte apoptosis in-vivo. The level of ROS in heart tissue among different groups was evaluated by MDA. As shown in Fig. 4a, compared with the control group, the concentration of MDA was dramatically increased in the Dox group (Fig. 4a, 2.1 ± 0.23 μ mol/g protein vs. 10.8 ± 0.78 μ mol/g protein; P < 0.01). Nonetheless this concentration was significantly decreased by 24% in the BM-MSCs-CdM group and by 40% in the iPSC-MSCs-CdM group compared with the Dox group (P < 0.01). The concentration of MDA was also significantly lower in the iPSC-MSCs-CdM group than in the BM-MSCs-CdM group (Fig. 4a, 6.4 ± 0.4 μ mol/g protein vs. 8.2 ± 0.25 μ mol/g protein; P < 0.01).
Next, cell apoptosis in different groups was determined by TUNEL staining. Compared with the control group, cell apoptosis was significantly increased over the myocardium in the Dox group (Fig. 4b,c, 25.4 ± 3.2% vs. 1.22 ± 0.3%; P < 0.01). Administration of BM-MSCs-CdM or iPSC-MSCs-CdM significantly decreased the cell apoptotic rate (Fig. 4b,c; P < 0.01) over the myocardium compared with the Dox group. The cell apoptotic rate was much lower in the iPSC-MSC-CdM group than in the BM-MSCs-CdM group (Fig. 4b,c, 9.2 ± 1.5% vs.16.4 ± 1.2%; P < 0.01). These findings indicate that iPSC-MSCs-CdM is more potent than BM-MSCs-CdM in reducing ROS and attenuating CMC apoptosis induced by Dox in-vivo.

Distinct enrichment of paracrine components between iPSC-MSCs and BM-MSCs.
To understand the differential paracrine capacity of BM-MSCs and iPSC-MSCs against Dox-induced cardiotoxicity, we performed a cytokine antibody array that covered 507 human cytokines/chemokines and growth factors in the serum free-CdM derived from BM-MSCs and iPSC-MSCs culture. Both BM-MSCs and iPSC-MSCs were able to release a wide spectrum of cytokines, chemokines, adipokines, growth factors, angiogenic factors and soluble receptors.
Pair-wise comparison of the protein profiling of CdM revealed many common characteristics but also some distinct differences. Overall, 214 of 507 factors were commonly detected in both BM-MSCs and iPSC-MSCs with <2.0-mean fold difference and P>0.1 from two tailed paired t-test (no significant difference). Classification of these factors into bioprocesses disclosed that these secreted factors were involved in metabolism, immunity, apoptosis, migration and differentiation as well as homeostasis. Nonetheless 13 factors were preferentially over-represented in BM-MSCs (mean fold difference >2.0 and P < 0.1); and 129 of 507 protein factors were preferentially over-represented in iPSC-MSCs (mean fold difference >2.0 and P < 0.1).
Our results revealed that most prominent distinctions preferentially found in iPSC-MSCs secretions were closely associated with regulation of apoptosis, response to exogenous and endogenous stimulus, cell migration, inflammatory response and cell activation ( Table 1). The top 20 protein factors over-represented in iPSC-MSCs-CdM were classified into biological processes using the NIH DAVID  Pathway Analysis tool (Fig. 5a). Although we do not know the functional role of each cytokine/growth factor or which play the most important role in myocardial repair, many of these factors have been individually shown to have positive effects on cardioprotection and to be involved in anti-inflammation (GDF-15 24 , MIF), anti-apoptosis (endothelial monocyte-activating polypeptide-II, Thrombospondin-1), anti-oxidation (MIF), and pro-proliferation and migration of cardiac progenitor cells and cardiomyocytes (platelet derived growth factor AA, erythoblastic leukemia viral oncogene homolog 3). Ingenuity pathway analysis software identified that the functional networks of the top 20 factors were involved in modulation of cell survival (e.g. nuclear factor kappa B signaling), stress response (e.g. mitogen-activated protein kinase 1/2 signaling), extracellular matrix metabolism (e.g. matrix metallopeptidase 1, collagens, fibronectin), immunomodulation (e.g. transforming growth factor belta signaling, immunoglobulin) and inflammation (e.g. interleukinI-1, interleukin-12 signaling) ( Fig. 5b-i). Additional analysis revealed that the cellular and molecular functions of MIF and GDF-15 factors were significantly associated with  Consistent with the cytokine array results, the concentration of GDF-15 and MIF in iPSC-MSCs-CdM and BM-MSCs-CdM was measured by Enzyme-linked Immunosorbent Assay (ELISA) assay. As shown in Fig. 5c, the concentration of GDF-15 ( Fig. 5c-i, 4113 ± 121 pg/mg protein vs. 559 ± 55 pg/mg protein; P < 0.01) and MIF ( Fig. 5c-ii, 12316 ± 1244 pg/mg protein vs. 1239 ± 151 pg/mg protein; P < 0.01) was significantly higher in iPSC-MSCs-CdM than in BM-MSCs-CdM.
In contrast, the functional enrichment profiles of the secreted factors over-represented in BM-MSCs-CdM (Table 2) were more dispersed. Interestingly, the enrichment score for the apoptosis group in BM-MSCs-CdM was 5-fold lower compared with iPSC-MSCs-CdM (enrichment score 0.24 versus 1.25), although the enrichment score for the cell migration group was similar (enrichment score 0.82 versus 0.65). These observations demonstrate a differential profile of secretions for BM-MSCs and iPSC-MSCs that might account for their different therapeutic effect in Dox-induced cardiomyopathy. Bioinformatics analysis has revealed that iPSC-MSCs secretions are preferentially involved in anti-apoptosis, anti-inflammation, regulation of cell mobilization and cell proliferation compared with BM-MSCs secreted factors.

Enriched GDF-15 and MIF in iPSC-MSCs secretome contributes to attenuation of Dox-induced ROS and apoptosis of cardiomyocytes.
To explore whether GDF-15 and MIF are responsible for the enhanced therapeutic effects of iPSC-MSCs-CdM, we investigated the effects of GDF-15-or MIF-depleted iPSC-MSCs-CdM on Dox-induced cardiotoxicity. Both GDF-15 and MIF were depleted from iPSC-MSCs-CdM by immunoprecipitation with antibodies specific for GDF-15 and MIF, respectively. After depletion, the concentration of GDF-15 was reduced from 4113 ± 121 pg/mg protein to 90 ± 10 pg/mg; while MIF was reduced from 12316 ± 1244 pg/mg protein to 120 ± 16 pg/mg; suggesting the immunodepletion of GDF-15 and MIF was successful. As shown in Fig. 6a,c, iPSC-MSCs-CdM-inhibited ROS generation was remarkably attenuated by depletion of MIF from iPSC-MSCs-CdM (Fig. 6a

Discussion
The findings of this study demonstrate that both human iPSC-MSCs-CdM and BM-MSCs-CdM provide cardioprotection against Dox-induced cardiomyopathy by reducing ROS and cardiomyocyte apoptosis. The therapeutic effects of iPSC-MSCs-CdM are more potent than those of BM-MSCs-CdM. In particular, iPSC-MSCs-CdM enriched with MIF and GDF-15 contributes to the reduced ROS generation and apoptosis of cardiomyocytes induced by Dox. Although iPSC-MSCs and BM-MSCs share many common properties, iPSC-MSCs have several unique advantages such as their higher proliferative capacity and a more robust differentiation potential 22 . The potential risk of tumor formation and poor cell engraftment with iPSC based transplantation nonetheless remains a concern. As the majority of the proposed benefits of MSC-based therapy is attributed to their paracrine effects, rather than direct trans-differentiation 21,25,26 , this prompted us to compare the potential therapeutic effects of CdM derived from human BM-MSCs and iPSC-MSCs in the treatment of Dox-induced cardiomyopathy.
In this study, we confirmed that cardiomyocyte apoptosis induced by increased ROS generation is one of the major mechanisms of Dox-induced cardiomyopathy [27][28][29] . Consistent with our prior studies 23,30 , we demonstrated that compared with BM-MSCs-CdM, iPSC-MSCs-CdM provides stronger anti-oxidative stress and anti-apoptotic actions that contribute to superior therapeutic efficacy in attenuation of Dox-induced cardiomyopathy. Our findings from antibody array showed that iPSC-MSCs and BM-MSCs not only release similar cytokines but also many distinct paracrine factors that may account for the superior therapeutic effects of iPSC-MSCs-CdM. Analysis of the highly expressed cytokines of    over-presented in iPSC-MSCs-CdM compared with BM-MSCs-CdM. GDF-15 has been recently identified as a critical cardioprotective factor with anti-apoptotic, anti-inflammatory and anti-cardiac remodeling properties in response to myocardial injury 33 . The therapeutic effects of iPSC-MSC-CdM are greatly abolished by neutralizing MIF and GDF-15. Insulin-like growth factor (IGF)-1, a well-known key factor that activates resident endogenous cardiac stem/progenitor cells 34 ; and Neuropilin-2, a key factor that promotes cardiac progenitor cells (CPC) and endothelial cell survival and migration 35 , were also greater than two fold higher in iPSC-MSCs-CdM than in BM-MSCs-CdM. As shown in the present study, a cytokine cocktail derived from MSC-derived CdM may provide a unique therapeutic tool for the treatment of AIC. Despite major advances in the prevention and treatment, such as the use of angiotensin converting enzyme inhibitors alone or with beta-blockers, AIC remains a major cause of morbidity and mortality in cancer survivors 36 . The use of iPSC-MSCs-CdM may be a novel therapeutic approach to prevent or treat AIC, especially in those in whom conventional medical therapy has failed. There is an emerging interest to apply MSC-derived secretome or exosome for organ repair and/or regeneration 21,37 . Recently, transplantation of stem cell-conditioned medium, instead of stem cells, has been shown to be effective for tissue/cell repair in a range of conditions such as ischemic heart disease [38][39][40] , retinal ischemia 41 , acute lung injury 42 , chemical-induced colitis, enteric neuropathy 43 and spinal cord injury 44 . Delivery may be by a single (immediate or delayed) administration 38,40,41,43 or continuous administration over a short period of time 39,44 . MSCs can release many beneficial cytokines that are important in cardiac protection, but they can also release potentially harmful factors 45 . Thus optimization of MSC paracrine factors is preferable in order to maximize their beneficial effects. In this study, we identified two crucial cytokines, GDF-15 and MIF, which are highly presented in the iPSC-MSCs-CdM for heart repair, that offer great potential for the future development of a therapeutic MSC-secreted cocktail. A similar approach may also be applied to optimize the paracrine profile of iPSC-MSCs-CdM for myocardial repair in different cardiovascular diseases.
There are many advantages of stem cell conditioned medium for tissue repair. Nonetheless in certain disease conditions, transplantation of stem cells may be superior as it enables stem cell differentiation and engraftment to replace damaged myocardial tissue 46 .
There are some limitations to this study. First, only the two most prominent cytokines in iPSC-MSCs-CdM secretion were analyzed. The functions of other factors that are significantly enriched in iPSC-MSCs-CdM need further investigation. Second, there is increasing evidence that paracrine factor profiling of MSCs is highly dependent on the microenvironment 30,47 . MSCs are sensitive to the inflammatory microenvironment and may polarize into two distinct phenotypes following specific toll like receptor stimulation, and result in different immune-modulatory effects and distinct secretions 48 . Polarization of at-rest MSCs into a pro-inflammatory phenotype (MSC1) or an immunosuppressive phenotype (MSC2) that responds to different stimulation provides an attractive model to understand the multiple facets of MSCs 49 . We harvested at-rest MSC secretion in vitro that may not truly reflect MSC-mediated myocardial repair in vivo. Third, the optimal dose of CdM for cardioprotection has not been determined. Fourth, the long-term impact of CdM on heart function and survival was not investigated in this study.
This study provides evidence that iPSC-MSCs-CdM is more effective than adult BM-MSCs-CdM for cardioprotection against Dox-induced cardiomyopathy. The distinct protective effect of iPSC-MSCs-CdM is associated with some enriched important cytokines such as MIF and GDF-15 that are involved in anti-apoptosis, anti-inflammation/oxidation, regulation of cell mobilization and reactivation. We have provided evidence that iPSC-MSCs-CdM may serve as a novel therapeutic approach for the future development of "cell-free" therapy for cardiovascular repair.
Preparation of CdM and cytokine assays. As described previously 23 , CdM of cultured MSCs was prepared. In brief, MSCs were trypsinized and plated on a 15-cm plate, a total of 5 × 10 6 cells. Twenty-four hours later, the regular culture medium was replaced with 15 ml of serum-and antibiotic-free DMEM (no EGF and bFGF). After a further 24 hours, the supernatant was aspirated gently, filtered through a 0.22 μ m filter, then transferred to ultrafiltration conical tubes (Amicon Ultra-15 with membranes selective for − 5 kDa), and finally centrifuged (4,000 g for 30 min at 4 °C) to concentrate the CdM. The final concentration was adjusted to 20 times that of the collected CdM.
Cytokine assay of MSC-CdM was performed using RayBio ® Cytokine antibody array as previously described 21 . Briefly, a total of 200 μ l of BM-MSCs-CdM or iPSC-MSCs-CdM was assayed for the presence of cytokines and other proteins according to the manufacturer's instructions (RayBiotech, Norcross, GA. Cat No: AAH-BLG-1-4). Quantitative human cytokines were measured using customized Bio-Plex cytokine assay (Bio-Rad Laboratories, Hercules, CA; Cat No. M50007VNJK and MF00038C9E) according to the manufacturer's instructions.

Comparison of protein expression in iPSC-MSCs-CdM and BM-MSCs-CdM. The proteins in
two iPSC-MSCs-CdMs and two BM-MSCs-CdMs were compared. According to the expression value, the proteins were be classified into three types: commonly detected in both BM-MSCs and iPSC-MSCs with < 2.0-mean fold difference and P> 0.1 from two tailed paired t-test, preferentially expressed in BM-MSCs (mean fold difference>2.0 and P < 0.1), or preferentially expressed in iPSC-MSCs (mean fold difference >2.0 and P < 0.1). In each group, gene-enrichment analysis and functional classifications were conducted by The Database for Annotation, Visualization and Integrated Discovery (DAVID) v6.7. Each protein group is associated with several annotation terms, and each term is coupled with a p-value, or so-called EASE score, which is a modified one-tail Fisher exact probability value. The protein groups' enrichment score stands for the geometric mean of negative denary logarithm of EASE scores of those terms involved in this group. Interactions and the functional network centered on certain factors were identified by INGENUITY pathway analysis (IPA).
NRCMs and culture. The NRCMs were isolated and cultured as described previously 50 . Briefly, hearts were quickly removed from neonatal Wistar rats (0-to 1-day-old) sacrificed by decapitation, rinsed four times with modified Hank's solution, and cut into small pieces on ice. The tissue fragments were warmed in a 50 ml tube in a water bath with a magnetic bar for 10 min at 37 °C. After discarding the supernatant, the minced myocardium was digested with fresh pre-warmed 0.2% trypsin for 5 min at 37 °C, and then the supernatant was collected gently and transferred to a 50 ml tube on ice containing 7 ml fetal bovine serum. These two steps were repeated to collect all the supernatant that was then centrifuged at 156.8 g for 5 min to collect the cells. Cells were re-suspended in NRCM culture medium to reduce fibroblast contamination. Finally, the supernatant was aspirated gently, and the cells plated in 24-well plates containing collagen-coated glass coverslips at a density of 2 × 10 5 cells/ml. Culture media was changed every day. NRCMs were assigned to four different culture conditions for 24 hours: 1) control group; 2) 1 μ

Detection of ROS and apoptosis.
To detect the generation of ROS induced by Dox, H 2 DCFDA staining was performed. Briefly, NRCMs were cultured in 24-well plates with collagen-coated glass coverslips, and treated as described above. Then NRCMs were incubated with H 2 DCFDA (10 μ M; Invitrogen) for 10 min at 37 °C in the dark. Subsequently, cells were washed with PBS twice and mounted with 4′ , 6-diamidino-2-phenylindole (DAPI). Fluorescence intensity for ROS signal (percent arbitrary fluorescence units, % AFU) was measured on 300 cells using Image J software. Finally, fluorescence intensity was calculated from five different view fields of each group in three independent experiments.
To demonstrate the protective effects of MSCs-CdM on Dox-induced apoptosis of NRCMs, TUNEL staining was performed. Cell samples were fixed with fixation solution, and incubated with blocking solution and then with permeabilisation solution. The heart sections were washed and incubated with 1 μ g/ml Proteinase K/10 mM Tris solution for 15 min at room temperature and washed twice in PBS. Finally, all samples were incubated with 50 μ l TUNEL reaction mixture in a humid chamber for 1 hour at room temperature. Sections were then washed 3 times in PBS, mounted with DAPI, observed under a fluorescent microscope and finally photographed.

ELISA.
According to the manufacturer's instructions, the concentration of GDF-15 and MIF in the CdM from iPSC-MSCs and BM-MSCs was measured using a human GDF-15 Elisa kit (Human GDF-15 DuoSet, DY957, R&D systems) and human MIF Elisa kit (Human MIF DuoSet, DY289, R&D systems) respectively.
Immunodepletion of GDF-15 and MIF from iPSC-MSCs-CdM. GDF-15 and MIF were depleted from iPSC-MSCs-CdM as described previously 51 . Briefly, 0.5 μ g of anti-GDF-15 (Clone 147627; R&D) and/or anti-MIF antibodies (Clone12302; R&D) or normal human IgG control antibody (1-001-A; R&D) were mixed with a suspension (50% slurry) of protein G-agarose beads in PBS at 4 °C for 1 hour with intermittent shaking. Following recovery by centrifugation, beads were washed three times and used for immunodepletion of GDF-15 and MIF. iPSC-MSC-CdM (1 ml) was incubated with protein G-agarose beads immobilized with anti-GDF-15 and/or anti-MIF antibodies or control human antibody for 1 hour at 4 °C. Immune complexes absorbed on protein G-agarose beads were precipitated by centrifugation. Finally, iPSC-MSCs-CdM was collected and centrifuged to 50 μ l and then used immediately.
Animal model. All animal experiments were performed in accordance with relevant guidelines and regulations by the University of Hong Kong and approved by the Committee on the Use of Live Animals in Teaching and Research (CULTAR) (Approval ID:3358-14). The animal model of AIC was induced in adult mice (6-8 weeks, ICR strain) by intraperitoneal injection of Dox (3 mg/kg, 3 times per week with a total cumulative dose of Dox 18 mg/kg) 52 . In the negative control group, mice were injected with an equal volume of saline. To confirm the successful creation of an animal model of AIC, echocardiography and invasive hemodynamic assessment were performed to measure heart function (n = 6) one week following completion of Dox injection. Since a single direct intramyocardial injection of 100 μ l of MSC-CdM into rat heart 53 , or 50 μ l of adipose-derived stem cell (ADSC)-CdM into mice heart had been reported to improve ischemic heart function 54 , we decided to inject 50 μ l of MSCs-CdM into mice heart in this study. A further group of mice with AIC were randomized to receive intramyocardial injection of 1) phosphate-buffered saline (Dox group, n = 16); 2) 50 μ l BM-MSCs-CdM (BM-MSCs-CdM group, n = 13); or 3) 50 μ l iPSC-MSCs-CdM (iPSC-MSCs-CdM group, n = 12) at 4 sites of the left ventricle (LV). Three weeks after CdM injection, invasive hemodynamic measurements were performed. All mice were then sacrificed and detailed histological examination was performed to assess myocardial fibrosis, apoptosis and oxidative stress.
In order to verify the role of GDF-15 and MIF in iPSC-MSCs-CdM for the treatment of AIC, additional groups of mice with AIC were randomized to receive intramyocardial injection of 1) phosphate-buffered saline (Dox group, n = 12); 2) 50 μ l iPSC-MSCs-CdM Cardiac function assessment. At 3 weeks after CdM injection, mice were anesthetized (intra-peri toneal injection of 100 mg/kg of ketamine and 20 mg/kg of xylazine) to undergo transthoracic echocardiography and invasive hemodynamic assessment. M-mode echocardiogram was performed to measure LVEF and FS as previously described 55 . A 1.4-Fr high-fidelity microtip catheter connected to a pressure transducer (Millar Instruments, Houston, TX, USA) was inserted into the LV cavity via the internal jugular artery to evaluate LV pressure and dP/dt using the PowerLab system (AD Instruments, Inc., Colorado Springs, CO, USA) 55 . In this study, the echocardiographic and hemodynamic analyses were performed by an experienced investigator who was blinded to the experimental information.
Histological analysis. All mice were sacrificed following hemodynamic study; hearts were harvested and fixed with 10% buffered formalin and embedded in paraffin, then sectioned to 5 μ m slides. HE staining and Sirius red staining were performed. The ratio of fibrotic area was quantified with 6 randomly chosen high power fields for each heart section, 6 mice for one group and analyzed using Image J with additional threshold color plug-in to process the file images.
MDA assay. The concentration of MDA in the heart tissues and treated NRCMs from different experimental groups was measured using the Thiobarbituric Acid Reactive Substance Assay Kit (Cayman Chemical), according to the manufacturer's instructions.

Statistical analysis.
Values are expressed as mean ± standard deviation. The significant differences between groups were analyzed with unpaired Student t test for two groups or one-way ANOVA, followed by Bonferroni test for more than 2 groups. A value of P < 0.05 was considered statistically significant.