Chemical suppression of specific C-C chemokine signaling pathways enhances cardiac reprogramming

Reprogramming of fibroblasts into induced cardiomyocytes (iCMs) is a potentially promising strategy for regenerating a damaged heart. However, low fibroblast–cardiomyocyte conversion rates remain a major challenge in this reprogramming. To this end, here we conducted a chemical screen and identified four agents, insulin-like growth factor-1, Mll1 inhibitor MM589, transforming growth factor-β inhibitor A83–01, and Bmi1 inhibitor PTC-209, termed IMAP, which coordinately enhanced reprogramming efficiency. Using α–muscle heavy chain–GFP–tagged mouse embryo fibroblasts as a starting cell type, we observed that the IMAP treatment increases iCM formation 6-fold. IMAP stimulated higher cardiac troponin T and α-actinin expression and increased sarcomere formation, coinciding with up-regulated expression of many cardiac genes and down-regulated fibroblast gene expression. Furthermore, IMAP promoted higher spontaneous beating and calcium transient activities of iCMs derived from neonatal cardiac fibroblasts. Intriguingly, we also observed that the IMAP treatment repressed many genes involved in immune responses, particularly those in specific C-C chemokine signaling pathways. We therefore investigated the roles of C-C motif chemokine ligand 3 (CCL3), CCL6, and CCL17 in cardiac reprogramming and observed that they inhibited iCM formation, whereas inhibitors of C-C motif chemokine receptor 1 (CCR1), CCR4, and CCR5 had the opposite effect. These results indicated that the IMAP treatment directly suppresses specific C-C chemokine signaling pathways and thereby enhances cardiac reprogramming. In conclusion, a combination of four chemicals, named here IMAP, suppresses specific C-C chemokine signaling pathways and facilitates Mef2c/Gata4/Tbx5 (MGT)-induced cardiac reprogramming, providing a potential means for iCM formation in clinical applications.


Reprogramming of fibroblasts into induced cardiomyocytes
(iCMs) is a potentially promising strategy for regenerating a damaged heart. However, low fibroblast-cardiomyocyte conversion rates remain a major challenge in this reprogramming. To this end, here we conducted a chemical screen and identified four agents, insulin-like growth factor-1, Mll1 inhibitor MM589, transforming growth factor-␤ inhibitor A83-01, and Bmi1 inhibitor PTC-209, termed IMAP, which coordinately enhanced reprogramming efficiency. Using ␣-muscle heavy chain-GFP-tagged mouse embryo fibroblasts as a starting cell type, we observed that the IMAP treatment increases iCM formation 6-fold. IMAP stimulated higher cardiac troponin T and ␣-actinin expression and increased sarcomere formation, coinciding with up-regulated expression of many cardiac genes and down-regulated fibroblast gene expression. Furthermore, IMAP promoted higher spontaneous beating and calcium transient activities of iCMs derived from neonatal cardiac fibroblasts. Intriguingly, we also observed that the IMAP treatment repressed many genes involved in immune responses, particularly those in specific C-C chemokine signaling pathways. We therefore investigated the roles of C-C motif chemokine ligand 3 (CCL3), CCL6, and CCL17 in cardiac reprogramming and observed that they inhibited iCM formation, whereas inhibitors of C-C motif chemokine receptor 1 (CCR1), CCR4, and CCR5 had the opposite effect. These results indicated that the IMAP treatment directly suppresses specific C-C chemokine signaling pathways and thereby enhances cardiac reprogramming. In conclusion, a combination of four chemicals, named here IMAP, suppresses specific C-C chemokine signaling path-ways and facilitates Mef2c/Gata4/Tbx5 (MGT)-induced cardiac reprogramming, providing a potential means for iCM formation in clinical applications.
Cardiovascular disease (CVD) has remained the leading cause of death in the United States and the whole world in the last decades (1). Myocardial infarction (MI) is the most severe outcome caused by cardiovascular disease, which leads to impaired heart function because of massive cardiomyocyte (CM) 3 loss and fibrosis (2). Existing treatments for myocardial infarction are primarily pharmacological and device-based, and do not address the underlying problem of CM loss (3). Indeed, the prevalence of chronic cardiomyopathy is steadily increasing worldwide, escalating the urgency of developing novel therapies for this morbid disease.
Direct reprogramming of fibroblasts into CM-like cells by introducing three cardiac transcription factors Gata4, Mef2C, and Tbx5 (G, M, and T) has emerged as an attractive strategy to generate induced CMs (iCMs) (4 -6). One major advantage of this strategy is utilizing abundant cardiac fibroblasts (CFs) in the heart to generate functional CMs to replace the scar tissue. Other advantages include the immunologic compatibility of the reprogrammed cells and lower possibility of tumorigenicity. However, low conversion rate, poor purity, and lack of precise conversion of iCMs still present significant challenges.
Several strategies have been applied to address these challenges in both mouse and human cells (5,. It has been shown that by adding other core transcriptional factors or protein kinases such as Hand2, Akt1 could further improve cardiac reprogramming efficiency in several fibroblast lineage s (5,12). Growth factors, such as fibroblast growth factor (FGF) 2, FGF10, and vascular endothelial growth factor (VEGF), can fur-This work was supported by the National Institutes of Health Grant HL109054 and a Pilot Award from the Joint Institute of University of Michigan Health System and Peking University Health Science Center (to Z. W.). The authors declare that they have no conflicts of interest with the contents of this article. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. This article contains Figs. S1-S3, Table S1, and Movies S1-S8. 1 To whom correspondence may be addressed. cro ARTICLE ther promote cardiac reprogramming (17). Specific chemical compounds that influence different signaling pathways also enhance the efficiency and quality of cardiac reprogramming (16,26). Moreover, recent studies have identified epigenetic barriers in the reprogramming process. For example, targeting the polycomb Ring finger oncogene Bmi1 and Mll1 H3K4 methyltransferase can enhance cardiac conversion (28). Singlecell study of cardiac reprogramming indicates that RNA splicing also plays an important role in the cardiac reprogramming (27,29). Nevertheless, to bring this strategy closer to clinical studies, the iCM conversion and maturation need to be substantially improved.
Here we screened and identified four different chemical compounds, insulin-like growth factor-1 (IGF-1), Mll1 inhibitor MM589, transforming growth factor (TGF)-␤ inhibitor A83-01, and Bmi1 inhibitor PTC-209 (IMAP) that remarkably enhanced the conversion from fibroblasts into cardiomyocytes. We observed significant improvement in cardiac reprogramming including cardiac gene expression, sarcomere formation, calcium flux, and spontaneous beating. Importantly, RNA-Seq data showed that immune response-related genes, especially those involved in chemokine signaling pathways, were down-regulated. Therefore, we chose three chemokine axes, C-C motif chemokine ligand 6 (Ccl6)-C-C motif chemokine receptor 1 (Ccr1), Ccl17-Ccr4, and Ccl3-Ccr5, as the potential targets. We then examined chemokine signaling pathway ligands CCL3, CCL6, and CCL17 and observed decreased reprogramming efficiency. In contrast, corresponding chemokine receptor inhibitors significantly increased iCM formation, similar to IMAP treatment. Taken together, these findings identified specific immune responses as potential barriers for cardiac reprogramming and offered a promising strategy to resolve the low-efficiency challenge of iCM formation by providing a novel chemical combination that targets immune responses.

The combination of four chemicals IMAP achieved high reprogramming efficiency of MEFs
To explore the significance of each chemical in combination, we applied minus-one strategy to eliminate possible overlapping effects of these compounds. Using the same screening platform, we found that the combination of all four chemicals showed the highest efficiency and increased iCM formation 6-fold compared with MGTϩDMSO treatment (Fig. 2A). Eliminating every single compound in the IMAP caused a significant reduction of the reprogramming efficiency (Fig. 2B). qPCR results indicated that IMAP maximally enhanced the expression of cardiac genes and repressed fibroblast genes (Fig. 2C). After 2 weeks of reprogramming with IMAP treatment, immunofluorescence staining showed 3-to 4-fold increases in the expression of cardiac troponin T (cTnT) and ␣-actinin as well as significantly increased cTnT/GFP-double positive cells ( Fig.  2D and Fig. S1A). Cell counting in several high power fields (HPFs) showed similar increase in cTnT, ␣-actinin, and cTnT/ GFP-double positive cells ( Fig. 2E and Fig. S1B).

IMAP also enhanced cardiac reprogramming of neonatal cardiac fibroblasts (NCFs)
Considering cardiac fibroblasts are the major in vivo target cell type for iCM reprogramming, we next determined the effect of IMAP on conversion of NCFs to iCMs. NCFs were isolated from the same ␣-MHC-GFP mouse line from which we obtained MEFs (30). Flow cytometry results indicated that the conversion efficiency was elevated 4-to 5-fold by IMAP compared with MGTϩDMSO group (Fig. 3A). qPCR showed significant increase of cardiac genes and decrease of fibroblast lineage-related genes (Fig. 3B). Immunofluorescence staining revealed a more significant increase of cTnT, ␣-actinin expression, and cTnT/GFP-double positive cells in NCFs compared with MEFs ( Fig. 3, C and D and Fig. S1, C-E) and clearly visualized sarcomere formation in iCMs derived from NCFs (Fig.  S1C).
Next, we tested spontaneous beating and calcium transient of iCMs because these two features represent characteristic cell functions of CMs. After treating NCFs with IMAP for 2 weeks and maturation medium for another 2 weeks, 8-to 9-fold increase in spontaneous beating of iCMs was observed ( Fig. 3E and Movies S1 and S2). Additionally, spontaneous calcium transient was measured by Ca 2ϩ imaging with Rhod-3 staining (4). With Rhod-3 staining in HPFs, hundreds of cells with spontaneous calcium transients were observed in MGTϩIMAP group whereas only dozens were observed in MGTϩDMSO group ( Fig. 3F and Movies S3 and S4).
It is possible that potential CM contamination could contribute to the iCM population. To address this potential issue, MEFs or NCFs without GMT infection and IMAP treatment were examined for GFP-positive signals from initial fibroblast isolation to 4 weeks after iCM induction. With microscopic examination, flow cytometry or immunofluorescence staining (Figs. 2, A, B, D, and E and 3, A, C, and D and Fig. S1, A-D), no GFP signals were detected in those MEF or NCF control groups, indicating that CM contamination is highly unlikely in our experiments. Likewise, no cells in

Suppressing chemokines enhances cardiac reprogramming
MEF or NCF control groups were observed to exhibit any spontaneous calcium transient or other CM properties (Fig. 3, E and F).

Lineage-tracing further validated that IMAP significantly enhanced iCMs formation from NCFs
To determine unequivocally that iCMs were indeed reprogrammed from NCFs, we performed iCM experiments with two complementary lineage-tracing experiments. First, isolated NCFs were infected at day 1 with lentiviruses produced with two lentiviral plasmids LV-Thy1.2-Cre pLKO.1 and pMSCVloxp-dsRed-loxp-eGFP-Puro-WPRE (Fig. 4A). Because Cre is driven by fibroblast-specific Thy1.2 promoter, the Cre expression will convert the original dsRed cells with pMSCV-loxp-dsRed-loxp-eGFP-Puro-WPRE into GFP green cells. Indeed, flow cytometry data showed that ϳ99% of the NCFs were Red without LV-Thy1.2-Cre pLKO.1 co-infection and ϳ90% of NCFs showed strong GFP expression with LV-Thy1.2-Cre pLKO.1 co-infection (Fig. 4B). At day 2, NCFs were infected with retrovirus encoding Mef2c, Gata4, and Tbx5 to induce the cardiac reprogramming. Green cells with spontaneous beating were counted 2, 3, and 4 weeks after infection and iCMs derived from MGTϩIMAP groups showed much better spontaneous beating, one of the most representative CM characteristics, than those obtained from MGTϩDMSO groups ( Fig. 4C and Movies S5 and S6).
To characterize further the iCM lineage conversion and CM functionality, ␣MHC-Cre/Rosa26A-Flox-Stop-Flox-GCaMP3 NCFs were isolated from corresponding transgenic mice (7) and calcium flux was measured during reprogramming (Fig. 4, D and E). This system is able to trace lineage conversion from non-CM lineages to CM lineage. No cells exhibiting spontaneous calcium transient under fluorescence microscope were observed in control groups, indicating no CM contamination in our experiments (Fig.  4F). Importantly, IMAP promoted calcium flux as early as 2 weeks after treatment (Fig. 4F). Four-week treatment of IMAP resulted in 10-fold increase of calcium flux ( Fig. 4F and Movies S7 and S8),

Suppressing chemokines enhances cardiac reprogramming
indicating that iCM matured further under extended IMAP treatment. Altogether, these data indicated that iCMs were reprogrammed from originally isolated fibroblasts and showed much better CM functionality such as spontaneous beating and calcium transient in MGTϩIMAP groups than those obtained from MGTϩDMSO groups.

RNA-Seq data showed significant changes in gene expression profiles defining cell identity and immune responses
To identify the downstream targets of IMAP in iCM formation, we first extracted the total RNAs and performed RNA-Seq analyses using cells reprogrammed from ␣-MHC-GFP NCFs Suppressing chemokines enhances cardiac reprogramming after 2 weeks of GMT and IMAP treatment. Cells reprogrammed with IMAP showed a much more robust cardiomyocyte gene expression profile compared with MGTϩDMSO group (Fig. 5A). iCMs derived from MEFs showed similar results with 1442 differentially expressed genes when comparing MGTϩIMAP group with MGTϩDMSO group (Fig. 5B and Fig. S2A). 528 overlapping genes were identified in both datasets, and gene ontology (GO) enrichment analysis revealed that IMAP stimulated gene expression related to muscle structure formation and muscle contraction as well as decreased gene expression in extracellular matrix formation ( Fig. 5B and Fig.   S2, B and C). Intriguingly, GO analysis identified significant gene repression in immune system process and inflammation response by IMAP ( Fig. 5B and Fig. S2C). We next examined in more detail the effect of IMAP on immune response gene expression. Several immune response representative genes were tested in reprogramming. Il6, Tnf, Ccl2, and Ptgs2 expression decreased as early as day 3 after IMAP treatment (Fig. 5C). RNA-Seq analysis further identified many immune-related genes that were significantly down-regulated after IMAP treatment in both NCFs (Fig. 5D) and MEFs (Fig. S2D). GO molecular function analysis revealed that these down-regulated genes

Suppressing chemokines enhances cardiac reprogramming
shared similar biological function in C-C chemokine signaling pathways ( Fig. 5E and Fig. S2E). These data suggested that C-C chemokine signaling pathways could be potential novel targets in improving iCM formation and may provide mechanistic insights into IMAP-mediated cardiac reprogramming.

Specific C-C chemokine signaling pathways were likely important barriers in cardiac reprogramming that were overcome by IMAP treatment
Based on the above results, we examined C-C chemokine signaling pathways Ccl6-Ccr1, Ccl17-Ccr4 and Ccl3-Ccr5 in cardiac reprogramming, as they were significantly down-regulated in MGTϩIMAP-treated group. We first performed loss-of-function studies by testing the inhibitors of corresponding C-C chemokine receptors in cardiac reprogramming. We applied CCR1 inhibitor BX471, a CCR4 antagonist, and CCR5 inhibitor Maraviroc in ␣-MHC-GFP MEFs. In MGT-induced cardiac reprogramming, all three inhibitors enhanced reprogramming individually whereas the three inhibitors together, called 3i, achieved a combinatorial improvement (Fig. 6A). These chemokine receptor inhibitors promoted the expression of cardiac-related genes and repressed the expression of fibroblast-related genes, similar to the effect of IMAP (Fig. 6B). Similar results were observed in NCF reprogramming (Fig. 6, C and  D). Immunofluorescence staining showed much higher number of cTnT, ␣-actinin, and cTnT/GFP-double positive cells in

Suppressing chemokines enhances cardiac reprogramming
MGTϩ3i-treated group compared with MGTϩDMSO group (Fig. 6, E and F and Fig. S3, A and B).
To further validate our findings that blocking the activities of specific C-C chemokine signaling pathways increased iCM reprogramming, we also performed gain-of-function experiments to examine the effect of chemokine ligands in reprogramming. Addition of chemokine ligands Ccl3, 6, and 17 in MGT-mediated reprogramming significantly reduced repro-

Suppressing chemokines enhances cardiac reprogramming
gramming efficiency (Fig. 6G and supporting Fig. S3C). In contrast, the inhibitory effect of these ligands was abolished when their receptor expressions or activities were inhibited in the presence of IMAP and 3i (Fig. 6H and Fig. S3D), indicating that these ligands' inhibitory effect on reprogramming was specifically through C-C chemokine signaling pathways. Taken together, our research showed that IMAP enhanced cardiac reprogramming by eliminating iCM reprogramming barriers imposed by specific C-C chemokine pathways (Fig. 6I).

Discussion
We report here a chemical mixture that enhances cardiac reprogramming by suppressing specific C-C chemokine signaling pathways. Administration of chemicals IGF-1, MM589, A83-01, and PTC-209 (IMAP), significantly enhances the reprogramming efficiency of MEFs or NCFs to iCMs. IMAP promotes cardiac gene expression, sarcomere formation, spontaneous beating, and calcium transient in reprogramming, indicating a significant role of IMAP in promoting maturation and functionality of iCMs. Importantly, repression of Ccr1, Ccr4, and Ccr5 signaling pathways is likely to be responsible for cardiac reprogramming augmentation mediated by IMAP. Direct inhibition of these chemokine receptors has achieved similar positive effect on cardiac reprogramming whereas their corresponding ligands repress reprogramming. Therefore, our study provides a novel platform of identifying chemicals that repress specific C-C chemokine signaling pathways for cardiac reprogramming.
In this study, we have developed a new platform enhancing cardiac reprogramming by combining four chemicals named IMAP. Two chemicals IGF-1 (12) and A83-01 (13), targeting IGF growth factor and TGF-␤ signaling, enhance reprogramming, as reported previously. PTC-209 is an inhibitor of polycomb Ring finger oncogene Bmi1, which functions through chromatin remodeling as an essential epigenetic regulator (31) and has been identified as a barrier in cardiac reprogramming (28). Importantly, our previous work reveals that targeting WD repeat domain 5 (WDR5) and blocking the WDR5 mixed-lineage leukemia (MLL) protein-protein interaction enhance cardiac reprogramming (28). In this study, we have identified that MM589, a newly invented version of WDR5-MLL1 inhibitor, promotes reprogramming. IMAP combination shows significant enhancement of MGT-induced cardiac reprogramming, and elimination of each of them decreases reprogramming efficiency. Our data thus indicate that these four chemicals work together to convert fibroblasts into iCMs.
Our IMAP studies identify an unexpected connection between C-C chemokine signaling pathways and cardiac reprogramming. In particular, we have observed that immune response-related genes, especially those involved in chemokine signaling pathways, were down-regulated in IMAP-mediated cardiac reprogramming. Subsequently, we have identified three chemokine axes, Ccl6-Ccr1, Ccl17-Ccr4, and Ccl3-Ccr5, as important reprogramming barriers. Administration of these three chemokine signaling pathway ligands, CCL3, CCL6, and CCL17, significantly decreases reprogramming efficiency. In contrast, addition of corresponding chemokine receptor inhib-itors in our reprogramming assay significantly increased iCM formation, similar to IMAP treatment. Thus, our study reveals that immune responses may act as barriers for cardiac reprogramming. Mechanistically, the effect of chemokine signaling pathways in reprogramming could be linked to profibrotic signaling, which has been identified to inhibit cardiac reprogramming as shown by others (13,15) and our studies. The three C-C chemokine pathways we have identified are all reported to promote fibrosis process both in rodents and humans (32)(33)(34). In addition, the effect of C-C chemokine pathways could be through Wnt pathway regulation. Wnt pathway-related genes also show significant decrease in our MGTϩIMAP treatment group compared with MGTϩDMSO group, and Wnt pathway is required for chemokine signaling-mediated functions in cardiac reprogramming (26,35). Future studies will be required to further reveal the underlying mechanisms.
Even though the identification of chemokine signaling pathways in inhibiting cardiac reprogramming is surprising, studies of immune response and chemokine signaling pathways in cell fate change and development provide strong rationale that chemokine signaling pathways are indeed very likely a key barrier in cardiac reprogramming. Traditionally, C-C chemokine signaling pathways were initially considered merely as low-molecular-weight proteins that stimulate recruitment of leukocytes (36). However, recent studies show that specific C-C chemokine signaling pathways, such as stromal cell-derived factor-1-cxcr4 axis, can exert negative feedback on the Fgf pathway at the beginning of fin regeneration (37) and regulate myoD and myf5 expression and fast fiber formation in muscle differentiation of zebrafish (38). Importantly, C-C chemokine signaling pathways can significantly impact cardiac development. In particular, Cxcr4 is considered an important marker for cardiomyocyte progenitors and may play a functional role in their differentiation (39). Further studies also show that repressing Cxcl10 and Ccl5 by WT1 is also required during cardiac development whereas Cxcr4 is required for spontaneous beating of hiPSC-CMs (40,41). In addition, targeting C-C chemokine signaling pathways has been proposed in cardiac repair after injury (42). Finally, anti-inflammatory drugs also facilitate induced pluripotent stem cell reprogramming (43,44), and anti-inflammation treatment may benefit cardiac reprogramming (25). Therefore, further investigation of chemokine pathways in cardiac reprogramming may provide key insights into cell fate change.
Direct cardiac reprogramming provides a promising strategy for repairing injured heart after massive cardiomyocyte loss. Application of chemical approaches for reprogramming represents a highly desirable approach to improve the efficiency of iCM formation, as reported previously by us (28) and others (13,15,16,21,26,45,46). Importantly, our studies have identified specific C-C chemokine signaling pathways regulated by a combination of four chemicals IMAP in cardiac reprogramming. We expect that future studies toward combined chemical approaches in targeting chemokine signaling pathways will further improve the iCM formation and bring cardiac reprogramming closer to clinical applications.

Mouse lines
The ␣-MHC-GFP transgenic mice were used to derive MEFs and NCFs as described previously (30,47). All animal-related procedures were approved by the Institutional Animal Care and Use Committee of the University of Michigan and are consistent with the National Institutes of Health Guide for Use and Care of Animals.

Plasmids
The pMXs-based retroviral polycistronic vector encoding Mef2c, Gata4, Tbx5 were provided from Dr. Li Qian's lab (18). This polycistronic constructs vector was constructed by DNA fragment containing Mef2c, Gata4, and Tbx5 sequentially, which were separated by oligonucleotides encoding P2A and T2A peptides. This polycistronic vector also contains puromycin selection marker for cell purification.

Primary cell isolation
Preparation of MEFs (isolated at E13.5) was reported previously (47). Briefly, embryos were harvested from ␣-MHC-GFP transgenic mice at 13.5 days post coitum followed by decapitation and removal of internal organs. The tissue was minced and digested with 0.05% trypsin/EDTA (Gibco, Thermo Fisher Scientific). Cells were resuspended in MEF medium (DMEM containing 10% FBS, 1% penicillin/streptomycin, 10 l/ml Glu-taMAX, and 1 mM sodium pyruvate) and plated onto one 10-cm dish per embryo. Cells were passaged at the ratio of 1:3 (passage 1). Passage 3 MEFs were used for reprogramming.
NCFs were isolated from P2-P3 ␣-MHC-GFP transgenic or ␣MHC-Cre/Rosa26A-Flox-Stop-Flox-GCaMP3 mice as described previously (30). Briefly, heart tissue was isolated, minced, and digested with 0.05% trypsin-EDTA. Then NCFs were collected with type II collagenase (0.5 mg/ml) in Hanks' Balanced salt solution. After washing and resuspending in Dulbecco's PBS with 2.5 g BSA and 0.5 M EDTA, cells were incubated with CD90.2 microbeads (Miltenyi Biotec, MidiMACS Starting Kit) at 4°C for 30min. Positive cells were isolated by magnetic-activated cell sorting and plated onto a 10-cm dish with FB medium (Iscove's modified Dulbecco's medium with 20% FBS and 1% penicillin/streptomycin) for future use. Isolated fibroblasts were routinely examined under fluorescence microscope or with GFP immunostaining to determine potential CM contamination.

Retrovirus preparation
80% confluent 10-cm plates of Plat-E cells were transfected by using Lipofectamine 2000 (Thermo Fisher Scientific) with 10 g retrovirus vectors in additional 1.5 ml Opti-MEM (Thermo Fisher Scientific). 24 h later, medium was changed into 10 ml fresh MEFs medium. 48 and 72 h after transfection, viral medium was collected twice and filtered through a 0.45-mm cellulose filter. The virus containing medium was added 1/5 volume of 40% PEG8000 solution to make a final concentration of 8% PEG8000. The mixture was kept at 4°C overnight and spun at 3000 ϫ g, 4°C, 30 min to get concentrated. Virus was resuspended by fresh MEFs medium with 8 g/ml Polybrene (Sigma).

Direct reprogramming of fibroblasts to iCMs
The protocol of direct cardiac reprogramming was similar to previous studies with minor optimization (30). Fresh fibroblasts were seeded on tissue culture dishes at a density of 10,000 cells/cm 2 before virus infection. Fibroblasts were infected with freshly made viral mixture containing 8 g/ml Polybrene (Sigma) 24 h post seeding. 24 h later, the viral medium was then replaced with induction medium composed of DMEM/199 (4:1) (Gibco, Thermo Fisher Scientific) containing 10% FBS, 1% penicillin/streptomycin, and 10 l/ml GlutaMAX (Gibco, Thermo Fisher Scientific). Medium was changed every 2-3 days with or without indicated chemicals for 2 weeks before cells were examined. 1 g/ml puromycin (SIGP8833-25MG, Sigma) was added into the medium 3 days after infection to eliminate noninfected cells. For spontaneous beating and calcium transient assessment experiments, induction medium was replaced by mature medium containing StemPro-34 SF medium (Gibco, Thermo Fisher Scientific), GlutaMAX (10 l/ml, Gibco, Thermo Fisher Scientific), ascorbic acid (50 g/ml, Sigma), recombinant human VEGF165 (5 ng/ml, R&D Systems), recombinant human FGF basic146 aa (10 ng/ml, R&D Systems), and recombinant human FGF10 (25 ng/ml, R&D Systems). Then mature medium was changed every 2-3 days for another 2 weeks as described previously (17).

Immunocytochemistry
Cells were fixed with 4% formaldehyde for 15 min and permeabilized with 0.1% Triton X-100 in PBS for 15 min at room temperature. Cells were blocked with 4% horse serum in PBS for 1 h and then incubated with primary antibodies against cTnT (Thermo Fisher Scientific), ␣-actinin (Abcam), and GFP (Thermo Fisher Scientific) overnight at 4°C followed by incubation with appropriate Alexa fluorogenic secondary antibodies (Thermo Fisher Scientific) at room temperature for 1 h. cTnT, ␣-actinin, and cTnT/GFP-double positive cells were manually quantified by single-blind method from 10 randomly HPFs within each well.

Lineage-tracing experiments
Two lentiviral plasmids LV-Thy1.2-Cre pLKO.1 and pMSCV-loxp-dsRed-loxp-eGFP-Puro-WPRE were used in our cardiac fibroblast lineage-tracing experiments. The plasmids were transfected with psPAX2 and pMD2.G into HEK-293 cells. Lentivirus was harvested 2 days after transfection and purified using the method described above for retrovirus preparation. Isolated NCFs were infected with these two lentiviruses at day 1 of cardiac reprogramming. At day 2, NCFs were infected with retrovirus encoding Mef2c, Gata4, and Tbx5 to induce cardiac reprogramming and medium was changed every 2-3 days. Green cells with spontaneous beating were counted under fluorescence microscope after 2, 3, and 4 weeks.

Flow cytometry
Fluorescence flow cytometry data were from 10,000 singlecell events which were collected using a standard MoFloAstrios flow cytometer (Immunocytometry Systems; BD Biosciences). Data were analyzed using Summit (BD Biosciences).

Quantitative real time PCR (qPCR)
Total RNAs from all induced cells with 2-week GMT and IMAP treatment were extracted using TRIzol Reagent (Thermo Fisher Scientific) following the manufacturer's instructions. RNA integrity was determined using formaldehyde denaturalization agarose gel electrophoresis. RNA concentrations were measured with the NanoDrop spectrophotometer (Thermo Fisher Scientific). RNA was reverse transcribed using iScript cDNA Synthesis Kit (Bio-Rad). qPCR was performed using StepOne Real-Time PCR System (Thermo Fisher Scientific). Primer oligonucleotides were synthesized by Sigma and are listed in Table S1.

Spontaneous beating and calcium transient assessment
Spontaneous beating assessment was performed by light microscopy at room temperature after indicated treatment of NCFs at 4-week time point. Beating cell number was manually quantified by single-blind method after isoproterenol treatment in each well of 48-well plate.
Calcium transient was measured with Rhod-3 Calcium Imaging Kit (Thermo Fisher Scientific) according to the manufacturer's instructions. ␣MHC-Cre/Rosa26A-Flox-Stop-Flox-Gcamp3 NCFs were also used for calcium transient measurements as reported previously (12) which were performed by fluorescence microscopy at room temperature after the indicated treatment of retrovirus and chemicals. Three HPFs of view were randomly selected within each well, video recorded for 3 min, and manually quantified.

RNA-Seq
Total RNAs from all induced cells with 2-week GMT and IMAP treatment were isolated using TRIzol following the provider's instructions. RNA (RIN Ͼ8.5) was used for RNA-Seq library preparation using NEBNext® Ultra TM II Directional RNA Library Prep Kit for Illumina (E7760S). The libraries were sequenced using HiSeq 4000 by the University of Michigan Sequencing Core. The quantification of RNA expression was estimated by Kallisto (48). Differential gene expression analysis was done using the R package DESeq2. The abundance of genes was used to calculate -fold change and p values. Cutoff values of -fold change greater than 2 and p values less than 0.05 were then used to select for differentially expressed genes between sample group comparisons. Significant pathway enrichment analysis was performed using PANTHER Overrepresentation Test (release 20150430, http://geneontology.org) 4 (49,50).

Statistic analysis
Results were presented as mean Ϯ S.E. Statistically significant difference between groups was analyzed by one-way analysis of variance (ANOVA) followed by the Student-Newman-Keuls multiple comparisons tests. A p value Ͻ0.05 was regarded as significant. Each experiment was performed at least twice.