Primary neonatal C57BL/6J mouse cardiac fibroblasts were purchased from PELOBIOTECH (cat. no. PB-C57-6049). For expansion, fibroblasts (5,000/cm2) were plated into gelatin-coated (0.2 mg/ml, Sigma-Aldrich) vessels (CELLSTAR, Greiner Bio-One) and cultured in high glucose Dulbecco's modified eagle medium supplemented with heat-inactivated fetal bovine serum (10%) and penicillin/streptomycin (1%; 100 U/ml, 100 µg/ml) (standard medium, all Gibco) at 37°C and 5% CO2 in a humified incubator. All experiments were performed with passage 2–5 cells.
Cardiac cell isolation and culture
Single-cell suspensions of primary neonatal cardiac cells (CMs, endothelial cells, smooth muscle cells, etc.) were isolated from healthy neonatal (age, 2–4 days) C57BL/6J wildtype and homozygous Myh6-mCherry–transgenic mice (strain, [B6;D2-Tg(Myh6*-mCherry)2Mik/J], The Jackson Laboratory, stock no. 021577, ) as described previously with minor modifications . Collagenase B (Roche) was used for tissue digestion. CMs were not purified. Erythrocytes were removed using the 10X red blood cell lysis solution from Miltenyi Biotec. Before molecular beacon transfection, cardiac cells were plated (100,000/cm2) into fibronectin-coated (0.005 mg/ml, Sigma-Aldrich) vessels and cultured for 24 h.
HEK293TN (System Biosciences, cat. no. LV900A-1) and HT1080 (ATCC, cat. no. CCL-121) cells were cultured in standard medium and used for lentivirus production or titration.
Lentiviral plasmid cloning, lentivirus production, and lentiviral transduction efficiency
Human-immunodeficiency-virus–based lentiviral gene transfer plasmids encoding the murine cardiac transcription factors Gata4, Mef2c, Tbx5, or Myocd in combination with an enhanced green fluorescent protein (eGFP) transduction reporter and the neo resistance gene were cloned as described previously . Control plasmids encoded only eGFP.
2nd-generation lentiviruses were produced in HEK293TN cells, concentrated, and biologically titrated in HT1080 cells as described previously . Titers ranged from 67 to 330 million transducing units per ml. Infection of adherent cardiac fibroblasts with individual lentiviruses at a multiplicity of infection of 5 resulted in 89 ± 3% eGFP-positive cells (mean ± SD; Additional file 1: Fig. S1).
Direct fibroblast reprogramming/ transdifferentiation
Cardiac fibroblasts were plated (5,000/cm2) in standard medium into gelatin-coated vessels. The next day, fibroblasts were infected with the GMTMy-lentivirus-cocktail at a multiplicity of infection of 5. After 24 h, standard medium was replaced with cardiac reprogramming medium, composed of low glucose Dulbecco's modified eagle medium and M199 (3:1) and supplemented with heat-inactivated foetal bovine serum (10%) and penicillin/streptomycin (1%). Control cells were infected with eGFP-lentiviruses at a multiplicity of infection of 20. On day 3 after lentivirus infection, cells were sub-cultured (5,000/cm2). From day 4 onwards, cells were subjected to geneticin (1 mg/ml; Fisher Scientific) selection. Thereafter, medium was changed every other day until day 14 (see reprogramming timeline in Fig. 1a).
Immunocytology of adherent cells and fluorescence microscopy
At indicated points in time, cells were fixed in paraformaldehyde (paraformaldehyde, 4%, Carl Roth) and incubated in permeabilization and blocking buffer (0.25% Triton X 100 and 10% secondary antibody-matched serum in Dulbecco's phosphate-buffered saline (DPBS)). Thereafter, cells were incubated with primary antibodies overnight at 4°C and with secondary antibodies for 2 h at room temperature in the dark (Additional file 1: Table S1). Cells were stained with DAPI (Invitrogen) before images were acquired using the Axio Observer Z1 fluorescence microscope equipped with AxioVision software (version 188.8.131.52) (both Carl Zeiss). Image-based quantification of CPC marker expression was performed in ImageJ (version 1.53a, NIH) using 20X images and the Cell Counter plugin.
Immunocytology of harvested cells and flow cytometry
At indicated points in time, cells were harvested, fixed in paraformaldehyde (4%), and incubated in permeabilization and blocking buffer. Thereafter, cells were incubated with primary antibodies for 1 h at room temperature and with secondary antibodies for 30 min at room temperature in the dark (Additional file 1: Table S1). Cells were resuspended in cell sorting buffer before analysis on a MACSQuant flow cytometer equipped with MACSQuantify software (both Miltenyi Biotech). Data were analyzed with FlowJo software (versions 7.6.5 or 10; BD Biosciences). A representative gating strategy is provided in Additional file 1: Fig. S3.
Molecular beacons and iCMP sorting
Molecular beacons are single-stranded oligonucleotides that form a stem-loop structure in the absence of their complementary target sequence. In the presence of their target sequence, the probes undergo conformational changes to hybridize to their target, resulting in opening of the stem, separation of fluorophore and quencher, and finally fluorescence after excitation. The following molecular beacons were used for analysis of Myh6/7 mRNA expression and iCMP sorting: (1) negative control (probe degradation control, random sequence without match in mouse genome, cyanine 5 (Cy5) and Black Hole Quencher 2 (BHQ2) label at 5’- and 3’-end, respectively); (2) delivery control (transfection control, random sequence, Cy5-Cy5 label); and (3) Myh6/7 beacons (specific target sequence, Cy5-BHQ2 label). Molecular beacon sequences have been published previously . The beacons were synthesized as DNA probes by Microsynth Seqlab. 100 µM stock solutions were prepared by resuspension in TE buffer (pH 7.0, Invitrogen) and stored at -20°C.
Optimal conditions for the molecular beacon transfection using the X-tremeGENE HP DNA Transfection Reagent (lipid, Roche) were determined in GMTMy-transduced cardiac fibroblasts using the delivery control beacon (Additional file 1: Fig. S4a). CM specificity was confirmed in Myh6-mCherry–transgenic cardiac cells using the Myh6/7 beacon (Additional file 1: Fig. S4b).
After GMTMy reprogramming of fibroblasts, live cells were sorted on day 14 after lentivirus infection according to the following protocol: Myh6/7 beacon stock solution (1 µl), transfection reagent (4 µl), and cardiac reprogramming medium (95 µl) were combined and preincubated for 20 min. The transfection mix was combined with additional medium (300 µl), added to adherent cells (400 µl/cm2), and incubated for 4 h at 37°C. Control cells were transfected with the negative control probe. Transfected cells were harvested and resuspended in cell sorting buffer (2% bovine serum albumin, 2 mM EDTA in DPBS; Carl Roth).
Myh6/7 mRNA expression analysis and iCMP sorting were performed by the BIH Cytometry Core Facility on a BD FACSAria II SORP, configured with 4 lasers (violet, blue, yellow-green, red), and equipped with FACSDiva software (versions 6.1.3 or 8.0.2) (both BD Biosciences). Representative gating strategies are provided in Additional file 1: Fig. S4a,c. Data were analyzed with FlowJo software. Sorted Myh6/7-Cy5–positive and Myh6/7-Cy5–negative cell samples were reanalyzed using the same sorter, the same gates, and the same settings as used during sorting. Enrichment of Myh6/7-Cy5–positive iCMPs was analyzed by immunocytological staining for CM markers. Purified iCMPs were plated (10,000/cm2) in cardiac reprogramming medium for subsequent analyses.
iCMP expansion and phenotype maintenance
For expansion, purified iCMPs were plated (5,000/cm2) into gelatin-coated vessels and maintained in cardiac reprogramming medium with supplements (see below) for 27 days. Every 3–4 days, at ~ 80% confluency, iCMPs were sub-cultured (5,000/cm2). Viable cell counts were determined using trypan blue (Sigma-Aldrich) and cumulative population doubling levels (PDLs) were calculated using the following formular: PDLharvested cells=3.32[log(count cell harvest)–log(cell number plating)] + PDLplated cells. During expansion, cell morphology and Myh6/7 expression were assessed regularly by microscopy and flow cytometry (see above), respectively.
iCMP phenotype maintenance was examined in basic cardiac reprogramming medium or in the presence of various supplements: (1) geneticin (0.5 mg/ml); (2) ascorbic acid (250 µg/ml, FUJIFILM Wako Chemicals Europe, cat. no. 13-19641); (3) FGF2 (2 ng/ml) and VEGFA (5 ng/ml) (both PeproTech); (4) FGF2 (2 ng/ml), VEGFA (5 ng/ml), and BMP4 (20 ng/ml; R & D Systems); and (5) FGF2 (2 ng/ml) and BMP4 (20 ng/ml) . Media were changed every other day.
RNA isolation, RNA sequencing, and bioinformatics analyses
Total RNA was isolated from duplicate samples of purified iCMPs, starting cardiac fibroblasts, and mouse adult left ventricular heart tissue using the Qiagen RNeasy Mini Kit (Qiagen GmbH). Genomic DNA was removed with DNAse I (Sigma-Aldrich). RNA quality and quantity were assessed on a 2100 Bioanalyzer System using the RNA 6000 Nano Kit (both Agilent Technologies). For RNA sequencing, poly-(A)-selection was performed with 500 ng total RNA using the NEBNext Poly(A) mRNA Magnetic Isolation Module (NEB). RNA libraries were prepared using the NEBNext Ultra RNA Library Prep Kit for Illumina (NEB). Library preparation success was confirmed by analyzing the fragment size distribution on a 2100 Bioanalyzer using the DNA 1000 Kit (Agilent Technologies). Library concentration was determined using Qubit dsDNA BR Assay Kit and Qubit 3.0 (Thermo Fisher Scientific). After equimolar pooling, RNA samples were sequenced using a HiSeq 1500 System with Rapid Mode chemistry v2 (50 cycles, single-read; Illumina).
For RNA sequencing data analysis, sequence reads were demultiplexed using bcl2fastq2 (version 2.18, Illumina) and fastq file quality was assessed using fastqc (version 0.11.7, Bioinformatics Group at the Babraham Institute). Residual adapter sequences and low-quality reads were trimmed using AdapterRemoval (2.2.2) . Reads were aligned to the mm10 (GRCm38.82) version of the mouse genome using tophat (version 2.1.0)  and bowtie2 (version 2.2.5) . Counts per gene were calculated as the sum of all mapped reads within a gene region. Raw counts of protein-coding genes were normalized and variance stabilizing transformed using DESeq2 (1.28.1)  in R (A language and environment for statistical computing. R Foundation for Statistical Computing, Vienna, Austria. URL https://www.R-project.org/; 4.0.2). Variances across all samples were determined for each gene and 1,000 genes with the highest variances were selected for unsupervised analysis. Principal component analysis of the 1,000 top-variably expressed genes was done with the prcomp() function in R. Differentially expressed genes were identified in pairwise comparisons between treatment groups using a negative binominal model provided by DESeq2. Bonferroni-corrected P < 0.05 and a minimal absolute log2 fold change of 1 were used to select significant results. K-means clustering of differentially expressed genes was based on Euclidean distances and performed using scaled data and the in-built kmeans() function in R with 100 randomly defined start sets of 6 clusters. Overrepresentation analysis of Differentially expressed genes in terms of the gene ontology system was done using the topGO package (Adrian Alexa and Jörg Rahnenführer (2016). topGO: Enrichment Analysis for Gene Ontology. R package version 2.32.0.) in R using classical Fisher tests based on hypergeometric distributions to asses significance of enrichment. Due to the high redundancy of the gene ontology system, raw P values were not adjusted for multiple testing. Heatmaps were generated using scaled data. The k-means heatmap was generated in R using the in-built heatmap function; the selected genes heatmaps were generated in GraphPad Prism (version 8 and 9, GraphPad Software). RNA sequencing data sets have been archived in the NCBI Gene Expression Omnibus data repository (GSE159315).
Reverse transcription-quantitative PCR
For reverse transcription-quantitative PCR, 100 ng RNA were random hexamer primed and reverse transcribed into cDNA using the SuperScript III First-Strand Synthesis System (Invitrogen). PCR reactions (8 µl each) were prepared by combining cDNA samples (final, 0.125 ng/µl reaction) with gene-specific primers (final, 0.5 µM), and the 2X Power SYBR Green PCR Master Mix (final, 1X; Applied Biosystems). Quantitative PCR was performed on the Quant Studio 6 Flex Real-Time PCR System (Applied Biosystems) with the following PCR program: initial denaturation (95°C, 10 min), annealing and extension (50 cycles: 95°C, 15 s; 60°C, 30 s; 72°C 30 s), melting curve program. Relative gene expression levels were quantified using the ∆ct-method, were ct is the threshold/quantification cycle and ∆ct = ct(geometric mean Rpl13a, B2m)-ct(gene of interest). B2m (beta-2-microglobulin) and Rpl13a (ribosomal protein L13a) served as reference genes. PCR amplification efficiencies were 1.9–2. PCR primers were designed using Primer-BLAST(NCBI)  and Primer3 (open software development project) , or obtained via PrimerBank  (B2m, ID 31981890a1; Tnnt2, ID 6755843a1) (Additional file 1: Table S2). Primers were synthesized by Eurofins Genomics.
Cardiac differentiation of iCMPs
To induce cardiac differentiation, iCMPs were stimulated with 5-azacytidine (Sigma-Aldrich), TGFB1 (PeproTech), and ascorbic acid, and subsequently cultured for 20 days in differentiation medium as described previously  (see differentiation timeline in Fig. 4a). Cardiac differentiation was initiated 2–3 days after sorting and plating (5,000/cm2) into gelatin-coated vessels. During differentiation, cell morphology, cardiovascular protein expression, and spontaneous contractility were regularly assessed by microscopy and immunocytology.
Mouse myocardial infarction model, cell transplantation
103 male C57BL/6J mice (Charles River), 10–12 weeks old and 20–30 g in weight, were used to evaluate the therapeutic effect of iCMP transplantation after myocardial infarction in 2 study arms: (1) serial measurement of heart function by echocardiography and determination of scar size after 6 weeks; and (2) iCMP location by immunohistology 2 weeks after transplantation. The week-6 group size (nSHAM=10, nmyocardial infarction groups=23) was calculated under consideration of mortality rates and cardiac parameters of previous experiments as well as the planned statistical analysis. The week-2 group sizes (nMI groups=8) was estimated owing to the explorative aim of iCMP location. Animals were randomly assigned to treatment groups.
Animals were anesthetized by intraperitoneal injection of narcosis mix (150 µl; fentanyl (0.08 mg/kg, Rotexmedica), midazolam (8 mg/kg, Ratiopharm), and medetomidine (0.4 mg/kg, cp-pharma)) and subsequently intubated. Permanent left anterior descending artery ligation was performed as described previously . Phosphate-buffered saline (DPBS, vehicle control, n = 31) or purified eGFP-transduced cardiac fibroblasts (CFseGFP, noninduced cell control, 5 x 105 cells, n = 31), or purified iCMPs (5 x 105 cells, n = 31) were injected at 2 sites of the lateral infarct border zone (5 µl each). Animals were awakened by intraperitoneal injection of antagonist mixture (100 µl; flumazenil (0.3 mg/kg, hameln), atipamezole (1.7 mg/kg; cp-pharma), and buprenorphine (0.1 mg/kg; Indivior UK Limited)), along with a subcutaneous injection of carprofen (100 µl, 5 mg/kg, Zoetis). All surgical procedures were carried out under aseptic conditions by a single investigator that was blinded for intervention. For sustained pain relief, animals were given metamizole (Ratiopharm) in drinking water (7 mg/ml in 5% glucose, B. Braun) daily for 5 days. Animals were monitored for 2–6 weeks and housed in individually ventilated cages (caging density 2–5) under specific pathogen-free conditions at the Charité Research Facility for Experimental Medicine. Animals were sacrificed by cardiac puncture. SHAM animals (n = 10) underwent the same procedures, except for left anterior descending artery ligation and intervention. Data were excluded when: (1) myocardial blanching was not observed during operation and absent infarction was corroborated by echocardiography and/or histological analysis; or (2) follow-up was not available. Overall mortality was 35%, including intra-, peri-, and postoperative deaths. Accordingly, group sizes in week 2/6 were: nSHAM=0/10, nDPBS=5/12, nCFeGFP=5/14, niCMP=6/13.
Transthoracic echocardiography was performed 2, 4, and 6 weeks after myocardial infarction using the Vevo 2100 Imaging System equipped with a MS550D transducer and Vevo LAB analysis software (version 03.01.00, all FUJIFILM VisualSonics). Anesthesia was induced with 3–4% isoflurane (AbbVie) and maintained with 1–2% isoflurane (O2 = 2 l/min). For all animals, static and cine loop images were acquired from the left ventricle in parasternal long axis and short axis views during electrocardiogram-gated kilohertz visualization, B-Mode, and M-Mode scanning. Images were analyzed using the “Left Ventricle Trace” tool of Vevo LAB. For each animal, measurements from 3 cardiac cycles were averaged for subsequent analysis. Myocardial-infarction–induced anatomical irregularities occasionally interfered with ultrasound imaging and analysis and resulted in missing values. The range of values for each parameter was nSHAM=7–10, nDPBS=8–11, nCFeGFP=10–14, niCMP=9–13. All analyses were performed by a single investigator that was blinded for intervention.
Histology and immunohistology
2 and 6 weeks after myocardial infarction and treatment, hearts were excised from sacrificed animals, dehydrated in sucrose solution (15%, Sigma-Aldrich) overnight at 4°C, and frozen in Tissue-Tek optimal cutting temperature compound embedding medium (A. Hartenstein). Transverse sections of 7 µm thickness were cut from the apex to the ligation site, with 280 µm intervals after every 80 sections, and mounted onto glass slides for histological analysis.
For scar size quantification, cryosections were fixed in paraformaldehyde (4%) and Masson trichrome staining (Sigma-Aldrich) was performed according to manufacturer’s instructions. Stained sections were imaged using the NanoZoomer-SQ slide scanner (Hamamatsu Photonics). The scarred and total left ventricular areas were quantified using ImageJ. Infarct scar size was calculated by dividing scarred tissue areas by total left ventricular areas and multiplying by 100 (nDPBS=7, nCFeGFP=7, niCMP=9).
For immunohistology, cryosections were fixed in paraformaldehyde (4%) and incubated in permeabilization and blocking buffer. Thereafter, cryosections were incubated with primary antibodies overnight at 4°C and with secondary antibodies for 2 h at room temperature in the dark (Additional file 1: Table S1). Sections were mounted in Fluoromount-G with DAPI (Invitrogen) before images were acquired using the Axio Observer Z1 fluorescence microscope equipped with AxioVision software or the TCS SP8 STED confocal microscope equipped with LAS X LS software (both Leica). Image acquisition and analysis was performed in a blinded manner.
Data were analyzed and graphs generated using GraphPad Prism. Data are presented as mean ± SD (in vitro data) or mean ± SEM (in vivo data). Repeated measures cardiac function data were analyzed by fitting a mixed-effects model to the data instead of performing a repeated measures ANOVA because it can handle randomly missing values. The GraphPad Prism model uses a compound symmetry covariance matrix and is fit using Restricted Maximum Likelihood (REML). Sphericity (equal variability of differences) was not assumed, instead the Greenhouse-Geisser correction was used. Multiple comparisons between group means were performed at individual measurement times and corrected for by statistical hypothesis testing using Holm-Sidak's post-hoc tests. Change data were analyzed using an ordinary one-way ANOVA after the D'Agostino & Pearson normality test was passed and the Bartlett’s test did not reveal significant different variances. Parametric test statistics are reported as [F(DFn, DFd) = F-value, P = P-value], where F corresponds to the F-statistic and DFn or DFd correspond to the degree of freedom nominator or degree of freedom denominator, respectively. Scar size data were analyzed using the Kruskal-Wallis test followed by Dunn’s post-hoc test to determine statistical significance of intergroup differences. The nonparametric test was chosen because the assumption of equal variances as determined using Bartlett’s test was violated, whereas the Shapiro-Wilk normality test was passed. P-values smaller than 0.05 were considered statistically significant.