OUP user menu

★ Editor's choice ★

Accelerated direct reprogramming of fibroblasts into cardiomyocyte-like cells with the MyoD transactivation domain

Hiroyuki Hirai, Nobuko Katoku-Kikyo, Susan A. Keirstead, Nobuaki Kikyo
DOI: http://dx.doi.org/10.1093/cvr/cvt167 105-113 First published online: 21 June 2013

Abstract

Aims Fibroblasts can be directly reprogrammed to cardiomyocyte-like cells by introducing defined genes. However, the reprogramming efficiency remains low, delaying the clinical application of this strategy to regenerative cardiology. We previously showed that fusion of the MyoD transactivation domain to the pluripotency transcription factor Oct4 facilitated the transcriptional activity of Oct4, resulting in highly efficient production of induced pluripotent stem cells. We examined whether the same approach can be applied to cardiac transcription factors to facilitate cardiac reprogramming.

Methods and results We fused the MyoD domain to Mef2c, Gata4, Hand2, and Tbx5 and transduced these genes in various combinations into mouse non-cardiac fibroblasts. Transduction of the chimeric Mef2c with the wild-types of the other three genes produced much larger beating clusters of cardiomyocyte-like cells faster than the combination of the four wild-type genes, with an efficiency of 3.5%, >15-fold greater than the wild-type genes.

Conclusion Fusion of a powerful transactivation domain to heterologous factors can increase the efficiency of direct reprogramming of fibroblasts to cardiomyocytes.

  • Cardiomyocytes
  • Direct reprogramming
  • Transcription factors

1. Introduction

Ischaemic heart disease is a leading cause of death in adults worldwide, primarily because most cardiomyocytes stop dividing shortly after birth and cannot replace damaged tissues.1 This problem has been addressed by reprogramming other cells to become cardiomyocytes. Several efficient protocols have been established to obtain cardiomyocytes from embryonic stem cells and induced pluripotent stem cells (iPSCs).2 In addition, direct reprogramming of fibroblasts to create induced cardiomyocyte-like cells (iCMs) is emerging as a novel approach to circumvent the teratoma formation associated with using pluripotent cells. Direct reprogramming is also faster than using fibroblasts to create first iPSCs and then cardiomyocytes, as has been shown in other lineages.3 Ieda et al.4 reprogrammed mouse cardiac and tail tip fibroblasts into beating iCMs in vitro by introducing three cardiomyocyte-specific transcription factor genes: Gata4, Mef2c, and Tbx5. Although another group found this approach inefficient,5 Song et al.6 added another cardiac factor, Hand2, and succeeded in converting tail tip fibroblasts to beating iCMs with an efficiency of 0.004%. When injected into infarcted hearts, both the 3- and 4-gene combinations reprogram cardiac fibroblasts into cardiomyocytes, resulting in improved cardiac functions.68 These studies provide evidence that iCMs can be prepared from other differentiated cells without going through a pluripotent state; however, the low efficiency is a major obstacle to applying iCMs clinically.

We recently found that the fusion of the powerful transactivation domain (TAD) derived from MyoD (the M3 domain) to the pluripotency factor Oct4 radically improves the efficiency of making iPSCs by facilitating epigenetic remodelling of Oct4-target genes.911 This successful experience prompted us to test whether fusion of the M3 domain to cardiac transcription factors would promote iCM formation.

2. Methods

2.1 Preparation of mouse fibroblasts

All animal experiments were conducted in accordance with the animal experiment guidelines of the US National Institutes of Health (Guide for the Care and Use of Laboratory Animals) and University of Minnesota (1002A78174). To prepare neonatal tail fibroblasts, whole tails were cut off from 5-day-old mice of the B6;129S4 genetic background (Jackson Laboratory #008214) with surgical scissors. The tails were rinsed in ethanol, washed with phosphate-buffered saline (PBS), and dissociated into small fragments using scissors, followed by incubation with collagenase and dispase (Roche Diagnostics) for 10–15 min. After centrifugation at 300 g for 5 min, the dissociated cells and tail fragments were cultured in fibroblast medium containing Dulbecco's modified eagle medium (DMEM), 10% foetal bovine serum (Hyclone), penicillin, and streptomycin. The medium was changed every other day. After 2–3 days the fibroblasts were frozen to store when they became 70% confluent.

Mouse embryonic fibroblasts were prepared from non-cardiac tissues of E13.5 embryos. The embryos were separated into three parts—the head, upper body, and lower body—under a dissecting microscope (Leica; Figure 1A). The upper body was discarded to avoid contamination with cardiac tissues. The lower body included the hip and hind limbs, but did not contain internal organs. The head and lower body were separately minced into small pieces and incubated with collagenase and dispase. After centrifugation at 300 g, the dissociated cells were cultured in the fibroblast medium and frozen to store when they became 70% confluent.

Figure 1

Establishment of beating iCM clusters from embryonic fibroblasts with M3 domain-fusion genes. (A) Fibroblasts were separately prepared from the head and the lower body of mouse embryos at E13.5. (B) Fusion constructs of the M3 domain attached to the amino-terminus of the four cardiac genes used in the study. Amino acid numbers are shown at the top. (C) Fusion constructs of the M3 domain attached to the carboxy-terminus of the four cardiac genes. (D) Fusion constructs of the VP16 transactivation domain attached to the carboxy-terminus of the four cardiac genes. (E) The number of beating iCM clusters prepared from head fibroblasts with various fusion gene combinations on Day 14. Fibroblasts were seeded into 48-well plates. The total number of beating iCM clusters in each well was obtained from more than three experiments with two to three wells in each experiment.

2.2 Preparation of iCMs

The M3 domain of mouse MyoD (amino acids 1–62) was fused to full-length cDNAs encoding mouse Mef2c, Gata4, Hand2, and Tbx5 at the amino- or carboxy-terminus to form M3M, M3G, M3H, M3T, MM3, GM3, HM3, and TM3, respectively (Figure 1B and C). The second half of the TAD of VP16 (amino acids 446–490) was separately fused to the carboxy-terminus of the four genes to form MVP16, GVP16, HVP16, and TVP16 (Figure 1D). pMXs-IP vectors12 encoding the fusion genes and wild-type genes were separately transfected into Plat-E cells13 (Cell Biolabs) with Lipofectamine 2000 (Life Technologies). The supernatants containing retroviruses encoding these genes were harvested 48 and 72 h later (called Day –1 and day 0, respectively), filtered through a 0.45-μm syringe filter, and transduced into fibroblasts. On Day –2, 5.0 × 103 to 8.4 × 103 fibroblasts were seeded in each well of 48-well plates in fibroblast medium. Fresh virus supernatant was added to the fibroblasts on Day –1 and Day 0 with 10 μg/mL polybrene. On Day 1, the virus supernatant was changed to fibroblast medium, which was replaced every other day thereafter.

2.3 Immunofluorescence staining and video imaging

Cells were fixed with 4% formaldehyde in PBS for 10 min and permeabilized with 0.5% Triton X-100 in PBS. Cells were stained with the primary and secondary antibodies (see Supplementary material online, Table S1) for 1 h each at 25°C. DNA was counterstained with Hoechst 33 342. Fluorescence images were captured with an AxioCam camera attached to an Axiovert 200 M microscope equipped with a ×10 A-Plan Ph1 Var1 objective lens (all from Zeiss). Images were processed with Photoshop 7 and Illustrator 10 (Adobe Systems). Phase-contrast video of beating iCMs was recorded with a DP70 digital camera on an IX81 motorized inverted microscope using DP Controller and edited with DP Manager (all from Olympus).

2.4 Quantitative reverse transcription PCR (qRT-PCR)

Total RNA was purified from fibroblasts on Day 1 after transduction with M-GHT and MM3-GHT using a PureLink Micro Total RNA Purification System (Life Technologies) and treated with DNase I as directed by the manufacturer. cDNA was synthesized using a SuperScript III First-Strand Synthesis System (Life Technologies) and qRT-PCR was performed with GoTaq qPCR Master mix (Promega) on a Realplex 2S system (Eppendorf) using the primer pairs listed in see Supplementary material online, Table S2. The mRNA level was normalized with that of glyceraldehyde 3-phosphate dehydrogenase (GAPDH). The mean and standard deviation (SD) were calculated from the values obtained from three independent experiments.

2.5 Calcium imaging

Fibroblasts transduced with the retroviruses were subcultured into glass chamber slides (NUNC) on Day 20 to observe calcium transients. The cells were labelled with 10 μM Fluo-3 (Life Technologies) for 1 h at 37°C, washed, and incubated for an additional 20 min in fibroblast medium to allow for de-esterification of the dye. Fluorescence images of calcium transients were acquired with an intensified CCD camera (Paultek Imaging) with a ×40 objective lens on an Olympus BX-51W upright microscope. Images were acquired and analysed using MetaFluor (Molecular Devices).

3. Results

Retroviruses encoding the M3 fusion genes and the wild-type genes were introduced into mouse embryonic fibroblasts in various combinations to examine the effect of the M3 domain on iCM formation. The number of beating iCM clusters was used to assess the effect of each combination. The most effective combination was MM3 (the M3 domain fused to the carboxy-terminus of Mef2c) with wild-type Gata4, Hand2, and Tbx5 (MM3-GHT), which produced beating iCM clusters from head fibroblasts 100-fold more efficiently than the combination of the four wild-type genes (M-GHT) as of Day 14 (Figure 1E, Group I, and Supplementary material online, Figure S1). The M3 domain fused to the carboxy-terminus of Gata4, Hand2, or Tbx5 did not produce beating iCM clusters when combined with wild-type genes (Figure 1E, Group I); rather, these fusion genes were inhibitory when combined with MM3 (Figure 1E, Group II). The M3 domain was ineffective when fused to the amino-terminus of each of the four genes (Figure 1E, Group III). In summary, the M3 domain was effective only when fused to the carboxy-terminus of Mef2c. We and others previously demonstrated that the TAD of VP16 also facilitates iPSC formation, although less efficiently than the M3 domain.9,10,14,15 The VP16 TAD did facilitate the formation of beating iCM clusters when fused to Mef2c (MVP16-GHT), but the efficiency was <20% of that with MM3-GHT (Figure 1E, Group IV, and Supplementary material online, Figure S1). In addition, we could not detect a synergistic effect when a VP16-fused gene was combined with MM3. Based on these results, we focused on the comparison between MM3-GHT and M-GHT in the following studies.

MM3-GHT produced much larger clusters, containing up to several hundred cells as determined by counting nuclear numbers with a phase-contrast microscope (Figure 2A and Supplementary material online, Videos), than M-GHT, whose beating clusters contained >10 cells by Day 13. Beating cells also demonstrated calcium transients as monitored with the calcium-sensitive dye Fluo-3 (Supplementary material online, Figure S2A). Some large beating iCM clusters with irregular shapes (Figure 2A, middle panel) could have resulted from the fusion of several nearby clusters. Therefore, we could be underestimating the reprogramming efficiency of MM3-GHT. To address this potential problem, we seeded fibroblasts at lower densities (5000 and 6000 cells/well). We also doubled the amount of each of the four genes in turn to further optimize the reprogramming conditions. The most efficient reprogramming condition was obtained with our initial cell density (8400 cells/well, our standard condition to make iPSCs10) transduced with equal volumes of the four genes (Supplementary material online, Figures S1 and S2B). The increased efficiency with MM3-GHT was not due to the higher expression level of some of the genes (Supplementary material online, Figure S3A) or to more active proliferation of the cells (Supplementary material online, Figure S3B) than with M-GHT.

Figure 2

Long-term culture of beating iCM clusters produced from embryonic fibroblasts. (A) Phase-contrast images of Day 13 small (left) and large (middle) beating iCM clusters (encircled) prepared with Mef2cM3, Gata4, Hand2, and Tbx5 (MM3-GHT). The entire area is beating in the right panel. Bars, 100 µm. Videos of these panels are available in Supplementary material online, Video. (B) The number of beating iCM clusters produced from head fibroblasts with Mef2c, Gata4, Hand2, and Tbx5 (M-GHT) and MM3-GHT over 40 days. Fibroblasts were seeded at 8400 cells per well into 48-well plates. The number of beating iCM clusters in each well is shown as means ± SD of more than three experiments. Asterisk indicates an experimental pair where the difference between the compared values was statistically significant. *P < 0.05 (Student's t-test). (C) The mean + SD of beating iCM clusters produced with M-GHT and MM3-GHT on Days 13, 25, and 40. The reprogramming efficiency (%) is shown under each graph. Fibroblasts were seeded into 48-well plates. **P < 0.01 (Student's t-test).

We followed the numbers of beating iCM clusters for 40 days in the optimized condition. The reprogramming efficiency was calculated by dividing the number of beating iCM clusters by the number of seeded fibroblasts, assuming that each cluster was a clone. The number of beating iCM clusters exceeded 100 by Day 13 and peaked around Day 25 for both head and lower body fibroblasts transduced with MM3-GHT (Figure 2B and C). As of Day 25, the efficiency was 3.2% (head) and 3.5% (lower body) for MM3-GHT and only 0.2% (both parts) for M-GHT, up to an 18-fold difference. The size and number of the beating iCM clusters obtained with MM3-GHT fluctuated during the 40 days; however, the number remained between 85 and 300 during this period (Figure 2B).

We next used immunofluorescence staining of cytoskeletal proteins to compare the efficiency of MM3-GHT and M-GHT for activating cardiac marker genes. Cardiac troponin T (cTnT) was expressed in 10–20% of the head fibroblasts transduced with MM3-GHT between Day 7 and Day 30, which was up to 2.2-fold higher than the frequency observed with M-GHT (Figure 3A, 10.6 vs. 4.9% on Day 30). The cTnT staining revealed a wide distribution of sarcomere-like structure throughout the cells with both gene combinations on Day 7 and Day 30 (Figure 3B for MM3-GHT). Additional immunofluorescence staining demonstrated that co-expression of cTnT and myosin light chain 2v (MLC2v), and MLC2v and cardiac myosin heavy chain (cMHC) was up to 3.4-fold more efficient with MM3-GHT than with M-GHT on Day 20 (Figure 3C and D, 4.8 vs. 1.4% for cTnT + MLC2v).

Figure 3

Expression of cardiac cytoskeletal proteins in embryonic head fibroblasts after transduction with M-GHT or MM3-GHT. (A) Frequency of cells positive for cTnT on Days 7, 20, and 30. Frequency was calculated as the number of positive cells divided by the total number of cells stained with Hoechst 33 342 and is shown as means + SD of more than three experiments. Asterisks indicate experimental pairs where the difference between the compared values was statistically significant. *P < 0.05 and **P < 0.01 (Student's t-test). (B) Immunofluorescence staining of cardiac troponin T (cTnT) on Day 7 and 30 after transduction with MM3-GHT. DNA was counterstained with Hoechst 33 342 (blue). Bars, 50 µm. (C) Double immunofluorescence staining of cTnT with myosin light chain 2v (MLC2v), and cardiac myosin heavy chain (cMHC) with MLC2v on Day 20 after transduction with MM3-GHT. DNA was counterstained with Hoechst 33 342. The areas surrounded by the rectangles in the left panels were magnified in the right three panels in each row. Bars, 50 µm for all panels. (D) Frequency of cells double-positive for cTnT and MLC2v, and cMHC and MLC2v on Days 20. **P < 0.01 (Student's t-test).

We also examined the expression patterns of ion channel proteins as examples of functional proteins of cardiomyocytes. Scn1b (sodium channel, voltage-gated, type I, beta subunit), Nav1.5/Scn5a (sodium channel, voltage-gated, type V, alpha subunit), and Hcn4 (hyperpolarization-activated cation channel 4) were immunostained for this purpose. The first two are widely expressed in the atrium and ventricle16,17; Hcn4 is abundantly expressed in the sinoatrial node and Purkinje cells in addition to weak expression in the atrium and ventricle.16,18 Although co-expression of cTnT and Nav1.5/Scn5A, and cTnT and Hcn4 was highly infrequent, MM3-GHT was consistently more efficient than M-GHT in the activation of Nav1.5/Scn5a and Hcn4 (Figure 4A and B). For example, MM3-GHT induced co-expression of cTnT and Hcn4 12-fold more efficiently than M-GHT (Figure 4B, 1.2 vs. 0.1%). The frequencies of double-positive cells for cTnT and Scn1b were not compared in this study because of very low frequencies of the cells (<0.1%). The frequencies of immunostaining-positive cells cannot be directly compared with the efficiency of forming beating iCM clusters due to the difference of the calculation methods (the number of nuclei surrounded by immunostaining-positive cytoplasm/total number of nuclei vs. the number of beating iCM clusters/the number of seeded fibroblasts). In addition, the current study examined only a small fraction of the many ion channel proteins; nonetheless, our results indicate the difficulty of activating these and other ion channel genes as a potential reason for the low efficiency of making beating iCMs.

Figure 4

Expression of ion channel proteins in iCMs generated from embryonic head fibroblasts with M-GHT or MM3-GHT. (A) Immunofluorescence staining of cardiac troponin T (cTnT), Scn1b (sodium channel, voltage-gated, type I, beta subunit), Nav1.5/Scn5a (sodium channel, voltage-gated, type V, alpha subunit), and Hcn4 (hyperpolarization-activated cation channel 4) on Day 20 after transduction with MM3-GHT. DNA was counterstained with Hoechst 33 342. Bars, 100 µm. (B) Frequency of cells double-positive for cTnT and Nav1.5/Scn5a, and cTnT and Hcn4 on Day 20. The frequency of positive cells (%) was calculated by dividing the number of positive cells by the total number of cells stained with Hoechst and is shown as means + SD of more than three experiments. Asterisks indicate experimental pairs where the difference between the compared values was statistically significant. **P < 0.01 (Student's t-test).

Finally, we used neonatal tail fibroblasts to form iCMs and found no beating iCM formation with M-GHT up to 35 days (Figure 5A). Lack of a cardiomyocyte-conditioned medium may explain why we did not detect beating iCMs with M-GHT, unlike the work by Song et al.6 However, MM3-GHT did form beating iCM clusters, albeit much less efficiently than with embryonic fibroblasts (0.025%, Figure 5A). Because of this low efficiency, we revisited many different gene combinations, but MM3-M3H-GT was the only other combination that formed beating iCM clusters (0.021%, Figure 5A). These beating clusters were small, containing less than five cells. Thus, neonatal tail fibroblasts were much less efficient than embryonic fibroblasts in forming beating iCMs, similar to iPSC formation.19 Despite such low efficiency, the expression frequencies of cTnT, MLC2v, cMHC, Nav1.5/Scn5a, and Hcn4 in neonatal fibroblasts were similar to or even higher than those observed with embryonic fibroblasts (Figure 5B, C and E). Double-positive cells for cTnT and Scn1b were again very rare (<0.1%). The staining patterns of sarcomere-like structure with the cTnT antibody were also similar between neonatal and embryonic fibroblasts (Figures 3B and 5D). Furthermore, the difference of the frequencies between MM3-GHT and M-GHT for the expression of these genes was not strikingly different between neonatal and embryonic fibroblasts.

Figure 5

Formation of beating iCMs from neonatal tail fibroblasts. (A) Comparison at Day 20 of the highest number of beating iCM clusters obtained with various gene combinations. The total number of iCM clusters was counted in each well of a 48-well plate. Means ± SD of more than three experiments and per cent efficiency are shown. (B) Double immunofluorescence staining of cardiac troponin T (cTnT) and myosin light chain 2v (MLC2v), and cardiac myosin heavy chain (cMHC) and MLC2v on Day 20 after transduction with MM3-GHT. DNA was counterstained with Hoechst 33 342. Bars, 100 µm. (C) Frequency (%) of cells positive for cTnT alone, double-positive for cTnT and MLC2v, and cMHC and MLC2v on Day 20. Asterisks indicate experimental pairs where the difference between the compared values was statistically significant. *P < 0.05 and **P < 0.01 (Student's t-test). (D) Sarcomere-like structures observed with cTnT staining on Day 20 after transduction with MM3-GHT. DNA was counterstained with Hoechst 33 342. Bar, 50 µm. (E) Frequency of cells positive for cTnT and ion channel proteins on Day 30.

4. Discussion

This work represents our second example of the successful application of the M3 domain to cell reprogramming. Our increased efficiency of making iCMs may still be insufficient to prepare enough cardiomyocytes to patch infarcted areas. However, the current increase is important for three reasons. First, it will allow us to obtain enough iCMs to analyse the epigenetics underlying the reprogramming. Second, the same strategy can be applied to prepare proliferating cardiac progenitor cells from fibroblasts using two recently reported genes (Mesp1 and Ets).20 Third, because cardiac fibroblasts are far more easily reprogrammed to iCMs than other fibroblasts,4 injection of MM3-GHT into infarcted areas can potentially radically improve cardiac function. Thus, this TAD-based approach is a powerful tool for studying cardiac direct reprogramming.

We used the number of beating iCM clusters as a primary indicator to evaluate the efficiency of cardiac reprogramming. This measurement indicates that embryonic fibroblasts were >100-fold more efficient in making iCMs than neonatal tail fibroblasts (3.2–3.5 vs. 0.025%). Such an age- or differentiation stage-dependent decrease of reprogramming efficiency is well known in the iPSC field. For instance, iPSC formation is 15-fold more efficient with embryonic fibroblasts compared with neonatal tail fibroblasts.19 Haematopoietic cells in the early stage of differentiation are more readily reprogrammed to make iPSCs than differentiated cells.21 It has been interpreted that epigenetic status is not tightly locked in embryonic or relatively undifferentiated cells; however, exact molecular mechanisms remain elusive.

It is notable that the six cardiac marker genes we studied here were activated at similar frequencies between embryonic and neonatal fibroblasts regardless of the transduction of MM3-GHT or M-GHT despite the major difference in the efficiency of forming beating iCM clusters between the two cell types. This could be because of the presence of other critical cardiac genes that were highly resistant to activation by these gene combinations in neonatal fibroblasts than in embryonic fibroblasts. It is also possible that although the frequencies of immunostaining-positive cells were similar, embryonic and neonatal fibroblasts behaved differently in terms of simultaneously activating multiple genes in a single cell. Double-immunostaining cannot distinguish the two situations where a minor population of the cells expressed the whole set of cardiac genes, resulting in spontaneous beating, and many cells stochastically expressed only a fraction of the entire set of cardiac genes in various combinations, falling short of beating. Transcriptome analysis at a single cell level can be applied to distinguish these two situations as used very recently in the study of iPSC reprogramming.22

As demonstrated with our results, only a highly selective combination between the TAD and the host transcription factor, as well as the location of the TAD in relation to the host factor (amino- or carboxy-terminus) can amplify the host factor's transcription capability. This is similar to our previous results with the fusion of the M3 domain to the Oct4 protein.9 Although the M3 domain was effective to increase the efficiency of iPSC formation when fused either at the amino- or carboxy-terminus of Oct4, it was not effective when fused to the other pluripotency factors Sox2 or Klf4. The optimal combination of the M3 domain and the host factors is not readily predictable from the domain structures of the host factors, such as the location of the TAD of the host factor. For instance, a TAD is located at each terminus in the Gata4 protein,23,24 in the middle portion the Tbx5 protein,25 and at the carboxy-terminus in the Mef2c protein.26 However, the information available in the literature about the physical interactions between these proteins might help with the proper design of the fusion genes. For example, the TAD at the carboxy-terminus of the Gata4 protein interacts with the amino-terminus of the Mef2a protein, which shares 100% identity with the Mef2c protein between amino acids 1–86.27 The amino-terminus of the Mef2c protein also interacts with the region close to the amino-terminus of the Tbx5 protein.28 It is likely that the fusion of the M3 domain to such critical regions disrupts the intricate original interactions between these transcription factors.

Both the M3 domain and VP16 TAD belong to a large family of acidic TADs.11 This family is characterized by an abundance of acidic amino acids (aspartic acids and glutamic acids) and hydrophobic amino acids (alanine, cysteine, phenylalanine, glycine, isoleucine, leucine, methionine, and valine).29 The M3 domain contains 13 acidic and 21 hydrophobic amino acids within the total 62 amino acids. The VP16 TAD, which is a prototype of acidic TADs, contains 21 acidic and 39 hydrophobic amino acids within the domain of 81 amino acids. Acidic TADs do not take specific three-dimensional structures in their free forms, but they do accommodate specific conformation upon binding to their target proteins.29 The binding proteins of acidic TADs include many chromatin proteins that play fundamental roles in transcription. For example, the binding proteins of VP16 TAD include basal transcription factors (TFIIA, TFIIB, TFIID, TFIIF, and TFIIH), Mediator, histone acetyltransferases (SAGA, NuA4, p300, and PCAF), and the SWI/SNF ATPase complex.11 Although little is known about the binding proteins of the M3 domain apart from p300, some of the aforementioned proteins potentially interact with the M3 domain.

Although we limited our study to the four transcription factors, two other groups screened miRNAs to make cardiomyocyte-like cells from fibroblasts. Jayawardena et al.30 showed that the combination of miR-1, miR-133, miR-208, and miR-499 can reprogram adult mouse cardiac fibroblasts to cardiomyocyte-like cells. While the current manuscript was under revision, Nam et al.31 reported that the combination of six genes including miRNAs (Gata4, Hand2, Tbx5, Myocd, miR-1, and miR-133) is the most efficient cocktail to activate cardiac genes in adult human fibroblasts among many combinations of transcription factor genes and miRNAs. This combination induces cardiomyocyte-like cells that spontaneously contract when introduced into human adult cardiac fibroblasts. However, spontaneous contraction was not observed when non-cardiac fibroblasts, such as neonatal foreskin fibroblasts or adult dermal fibroblasts, were used as parent cells. Overall, the current work and previous reports4,31 are consistent in finding that non-cardiac fibroblasts are much less able to become cardiomyocyte-like cells than cardiac fibroblasts. This suggests that the epigenetic memory derived from parent cells strongly restricts potential fate, which is a commonly observed problem during iPSC formation.3234 Because heart-specific transcription factor genes and miRNAs have been extensively screened in previous studies,4,6,30 more promising targets for screening lie in the category of epigenetic modifiers, such as SWI/SNF chromatin remodelling ATPases, histone covalent modifying enzymes, and regulators of DNA methylation. Indeed, many histone modifying enzyme genes have been shown to be critical in iPSC formation.35,36 Because the mechanistic analysis of nuclear reprogramming is far more advanced in the iPSC field than in the cardiac reprogramming area, the application of successful strategies used to identify new reprogramming factors in iPSC formation warrants serious consideration in the study of direct cardiac reprogramming.

Funding

This work was supported by the Engdahl Funds, the Office of the Vice President for Research of the University of Minnesota, and the National Institutes of Health (R01 GM098294).

Acknowledgements

We thank Toshio Kitamura for pMXs-IP, Meri Firpo for mice, and Atsushi Asakura for the video recording equipment.

Conflict of interest: none declared.

References

View Abstract