Let-7 family of microRNA is required for maturation and adult-like metabolism in stem cell-derived cardiomyocytes

Significance

The adult human heart is incapable of significant regeneration after injury. Human embryonic stem cells (hESCs) have the capacity to generate an unlimited number of cardiomyocytes (CMs). However, hESC-derived CMs (hESC-CMs) are at a fetal state with respect to their functional and physiological characteristics, diminishing their utility for modeling adult-related heart disease and therapeutic screening. Thus, the potential for hESC-CMs may improve immensely in cardiac-related therapeutic applications if factors that drive their maturation are uncovered. In this study, we show that members of let-7 miRNA family control CM metabolism, cell size, and force contractility, making them one of the best factors identified to date in promoting maturity of stem cell derivatives.

Abstract

In metazoans, transition from fetal to adult heart is accompanied by a switch in energy metabolism-glycolysis to fatty acid oxidation. The molecular factors regulating this metabolic switch remain largely unexplored. We first demonstrate that the molecular signatures in 1-year (y) matured human embryonic stem cell-derived cardiomyocytes (hESC-CMs) are similar to those seen in in vivo-derived mature cardiac tissues, thus making them an excellent model to study human cardiac maturation. We further show that let-7 is the most highly up-regulated microRNA (miRNA) family during in vitro human cardiac maturation. Gain- and loss-of-function analyses of let-7g in hESC-CMs demonstrate it is both required and sufficient for maturation, but not for early differentiation of CMs. Overexpression of let-7 family members in hESC-CMs enhances cell size, sarcomere length, force of contraction, and respiratory capacity. Interestingly, large-scale expression data, target analysis, and metabolic flux assays suggest this let-7–driven CM maturation could be a result of down-regulation of the phosphoinositide 3 kinase (PI3K)/AKT protein kinase/insulin pathway and an up-regulation of fatty acid metabolism. These results indicate let-7 is an important mediator in augmenting metabolic energetics in maturing CMs. Promoting maturation of hESC-CMs with let-7 overexpression will be highly significant for basic and applied research.

Several coronary heart diseases (CHDs) are characterized by cardiac dysfunctions predominantly manifested during cardiac maturation (1, 2). Dramatic changes in energy metabolism occur during this postnatal cardiac maturation (3). At early embryonic development, glycolysis is a major source of energy for cardiomyocytes (CMs) (4, 5). However, as the cardiomyocytes mature, mitochondrial oxidative metabolism increases with fatty acid oxidation, providing 90% of the heart’s energy demands (6–8). This switch in cardiac metabolism has been shown to have important implications during in vivo cardiac maturation (9). In contrast to the relatively advanced knowledge of the genetic network that contributes to heart development during embryogenesis (10, 11), molecular factors that regulate peri- and postnatal cardiac maturation, particularly in relation to the metabolic switch, remain largely unclear. So far, studies to understand the transition of the glycolysis-dependent fetal heart to oxidative metabolism in the adult heart have been mostly related to the peroxisome proliferator-activated receptor (PPAR)/estrogen-related receptor/PPARγ coactivator-1α circuit (7, 8, 12). However, it is currently unknown what other factors act upstream or in synergy with this pathway in controlling cardiac energetics.

miRNAs have emerged as key factors in controlling the complex regulatory network in a developing heart (13). Genetic studies that enrich or deplete miRNAs in specific cardiac tissue types and large-scale gene expression studies have demonstrated that they achieve such complex control at the level of cardiac gene expression (14–16). We sought to determine whether these small noncoding RNAs have an important role during cardiac maturation, specifically in relation to cardiac energetics. The in vitro-generated human embryonic stem cell-derived CMs (hESC-CMs), despite displaying several functional and physiological similarities to the CMs in the developing heart, are in a fetal state with respect to their ion channel expression and electrophysiological activity, as well as their metabolic phenotype (17–21). In this study, we therefore have used hESC-CMs as a powerful platform to understand and elucidate cardiac maturation. Using large-scale transcriptome analysis, we first show that molecular signature patterns of hESC-CMs taken through 1 y of culturing for cardiac maturation reflect in vivo cardiac maturation. Furthermore, large-scale miRNA sequencing of in vitro-derived mature hESC-CMs reveals several key differentially regulated miRNAs and miRNA families. Target analysis using miRNA and mRNA datasets from mature CMs indicates that the let-7 family, which is one of the most highly up-regulated families, targets several key genes in the PI3K/AKT/insulin pathway during cardiac maturation. Because the let-7 family has been previously associated with energy metabolism (22, 23), it was chosen as a prime candidate for further analyses. Knock-down (KD) of let-7 results in a significant decrease in a number of maturation parameters such as CM size, area, sarcomere length, and expression of several cardiac maturation markers. Overexpression of selected members of the let-7 family for just 2 wk in hESC-CMs significantly increases cell size, sarcomere length, contractile force, and action potential duration. More importantly, the overexpression (OE) of let-7 in CMs exhibit higher respiratory capacity and increased efficiency in using palmitate as an energy source, thus strongly implying a metabolic transition in these cells. This switch is synchronized with a significant down-regulation of a number of let-7 target genes in the PI3K/AKT/insulin pathway and other key regulators such as the histone methyl transferase enhancer of zeste homolog 2 (EZH2), suggesting let-7 acts as a global regulator to bring about the metabolic and functional changes required during cardiac maturation. Finally, repression of insulin receptor substrate 2 (IRS2) and EZH2 in CMs mimics the effect of let-7 OE, suggesting these targets could be important components of a let-7-driven maturation pathway. Altogether, our results indicate the let-7 family as a novel endogenous regulator that can simultaneously accelerate maturation and adult-like metabolism in human cardiac tissue.

Results

In Vitro Cardiac Maturation Physiologically Simulates in Vivo Cardiac Maturation.

To examine whether hESC-CMs can be used as a model to study CM maturation, we adopted two different maturation protocols for hESC (H7)-CMs: 3D engineered heart tissue culture and prolonged 2D culture conditions (Fig. 1A). 3D engineered heart tissue was generated in gels of type I collagen and mechanically conditioned via static stress for 2 wk by fixing the ends of the constructs between two posts (termed cEHT here, for conditioned engineered heart tissue) (24). Previous reports have shown that prolonged culturing of hESC-CMs for up to 1 y can result in a tightly packed and parallel array of myofibrils with mature Z, A, H, I, and M bands (25). Thus, in the second protocol we adopted, standard 2D CMs were subjected to prolonged culturing (13.5 mo, termed 1y-CM). In the current investigation, CM populations with higher than 70% purity assessed by flow cytometry for cardiac troponin T positive (cTnT⁺) cells were used for all assays (SI Appendix, Fig. S1). Quantitative PCR analysis of known cardiac markers further validated the maturation process (Fig. 1 B and C). To further verify the extent of maturation of CMs generated by in vitro methods, we used large-scale sequencing, using an Illumina platform, to compare the mRNA expression profiles between day 20-CMs and 1 y-CMs in relation to human adult heart (HAH) samples (see more details on HAH in Materials and Methods and SI Appendix) and 3-mo-old human fetal ventricular (HFV) and atrial (HFA) samples (26) (Fig. 1E and Dataset S1). 2D principal component analysis (2D PCA) of all genes for all of the samples clearly separates 1y-CMs and HAH samples the farthest from day 20-CMs while placing the HFA and HFV samples in the middle in the principal component 1 (PC1) axis (Fig. 1D). Examination of the transcript levels of all significantly regulated genes [P ≤ 0.001 and fold change (FC) ≥ 2] in the abovementioned samples, using Ingenuity Pathway Analysis (IPA), revealed several interesting patterns and groups across the different samples. Cardiac maturation is known to improve Ca handling (27), fatty acid metabolism (9, 28), and sarcomere organization (29) and results in the down-regulation of glucose metabolism/insulin signaling (30), cell proliferation (31), and pluripotency. Twelve categories reflecting these parameters are presented as a heat map (Fig. 1E and Dataset S2). Most categories show the same trend of up- or down-regulation between 1y-CMs and HAH, suggesting that several pathways known to be critical during in vivo heart development are also coregulated during in vitro cardiac maturation (Fig. 1E). A more in-depth evaluation of the data using density plots revealed that pathways related to hypertrophic signaling, sarcomere organization (actin cytoskeleton), calcium, and cAMP-mediated signaling (27) and integrin signaling were significantly up-regulated (P ≤ 0.01) in both HAH and 1y-CM samples, suggesting in vitro maturation processes physiologically simulate the in vivo cardiac maturation (Fig. 1 E–I and SI Appendix, Fig. S2 A–H). Previous studies have shown that CMs rapidly proliferate during fetal life (31). However, a vast majority of postnatal human CMs do not proliferate, although they are capable of DNA synthesis without nuclear division or nuclear division without cytokinesis, thereby increasing in ploidy (8N) and size (hypertrophy) (31–33). Consistent with these data, a number of cell cycle-related genes were still up-regulated in our 3-mo-old HFV and HFA samples (Fig. 1E) and did not show a significant down-regulation (Fig. 1 H and I). In contrast, in both HAH and 1y-CMs, the cell cycle genes were significantly down-regulated. (Fig. 1 F and G).

Fig. 1.

The molecular signatures of in vitro cardiac maturation reflect in vivo cardiac maturation. (A) Schematic representation of large-scale mRNA and miRNA sequencing using Illumina platform from day 20-CMs and in vitro-matured CMs derived from hESC (H7). (B and C) qPCR analysis of maturation markers in day 20-CMs and in vitro-matured CMs. Means ± SEM are shown. **P ≤ 0.05 (Student's t test). (D) 2D principal component analysis using genomewide expression data for day 20-CMs, 1y-CMs, HAH, HFA, and HFV samples attained using R. (E) Heat map depicting changes in gene expression of 12 different pathways between day 20-CMs, 1y-CMs, HAH, HFA, and HFV samples attained using R. The rows reflect read counts and are standardized individually and colored according to the Z score. Yellow and blue represent up- and down-regulation, respectively. (F–I) Density plots using R generated with fold change expression of genes from four representative categories for HAH (F), 1y-CM (G), HFA (H), and HFV (I) relative to gene expression of day 20-CMs. X axis indicates log₂ fold change in gene expression. Black line indicates expression of all genes. Colored lines toward the left and right side of the black line indicate down-regulation and up-regulation of pathways, respectively. All experiments were repeated at least three times.

In animal models, CMs are known to shift their metabolism from glycolysis to fatty acid oxidation during postnatal cardiac maturation. This is well documented in in vivo studies using murine and rabbit models (3, 34, 35). Furthermore, accumulating molecular and clinical data in humans support a similar transition from glycolysis to fatty acid metabolism as the CMs undergo postnatal maturation (36, 37). Consistent with this, although the HFA and HFV samples do not show an increase in fatty acid metabolism (Fig. 1 E, H, and I), several genes in the fatty acid metabolism pathway are up-regulated in both 1y-CMs and HAH samples (Fig. 1 E, F, and G and Dataset S2). Interestingly, in parallel to increased fatty acid metabolism, a down-regulation of several genes in the PI3/AKT/insulin pathway was observed in the 1y-CMs and HAH (Fig. 1 E–G and Dataset S2), suggesting a reduced use of glucose for their metabolic needs. These profiling data together indicate that in vitro maturation of hESC-CMs results in CMs that possess molecular signatures similar to those seen in postnatal CMs, and thus can be used as an excellent model to elucidate novel regulators during cardiac maturation. The effect of long-term culturing on cardiac maturation was also analyzed in the IMR90-induced pluripotent stem cell line and the overall gene expression of the IMR90 iPSC line was very similar to that derived from the H7 line (SI Appendix, Fig. S3). It was intriguing to see that despite the heterogeneity in the composition of human fetal and adult heart samples in comparison with the in vitro-matured CMs, the overall behavior in the trends of the various pathways was still consistent with what is known during cardiac maturation.

Let-7 Family of miRNAs Is Highly Expressed in hESC-CMs Matured in the Dish.

Because miRNA patterns in cEHTs and 1y-CMs should reflect the miRNA pattern changes observed during cardiac maturation, we used an Illumina high-throughput miRNA sequencing platform to elucidate miRNAs that are highly enriched in these samples compared with day 20-CMs (Fig. 2A and Datasets S3 and S4). Approximately 600 miRNAs were identified with deducible read counts (Fig. 2A) from each of the two datasets. Of these, ∼250 miRNAs were significantly regulated (FC ≥ 2 and P ≤ 0.001) in each dataset. To derive a robust list of miRNA candidates that are regulated during maturation, we only chose those miRNAs that were significantly regulated in both 1y-CM and cEHTs. This resulted in a list of 77 miRNAs (Dataset S5). Myogenic miRNAs (myomiRs) such as miR-1, miR-208, and miR-133 were significantly changed in only one of the two datasets (SI Appendix, Fig. S4A). A heat map analysis of the 77 miRNAs in the two datasets revealed four groups of miRNAs (Fig. 2B and SI Appendix, Fig. S4B): Some representative candidates of miRNAs up-regulated in both datasets (group 1) were members of the let-7 and mir-378 families and mir-30. Similarly, candidate members that were down-regulated in the both datasets were mir-502 and mir-129 (group 2). To delineate the pathways most significantly regulated by miRNAs during CM maturation, we analyzed two groups of miRNA–mRNA interactions from 1y-CMs using IPA and an miRNA-mRNA target filter algorithm: the overlap between the targets of down-regulated miRNAs and up-regulated mRNA in mature CMs, and the overlap between the targets of up-regulated miRNAs and down-regulated mRNAs during maturation. The three miRNAs showing the highest number of targets in our mRNA dataset were let-7, mir-378, and mir-129 (Fig. 2 A–C), and they thus were chosen for further pathway analysis using Genemania (www.genemania.org) and/or previous literature. Interestingly, a pathway analysis algorithm of Genemania revealed that a considerable number of the let-7 targets that were down-regulated in CM maturation belonged to PI3/AKT/insulin signaling (Dataset S6). Because the let-7 family is one of the most highly up-regulated miRNAs in both cEHTs and 1y-CMs, and because this miRNA family has the largest number of down-regulated targets in the mature CMs and a large subset of its targets belonged to the insulin signaling pathway, we chose to examine let-7 in more detail in relation to CM maturation.

Fig. 2.

Genome-wide sequencing of in vitro-matured CMs reveals let-7 as the most highly expressed miRNA family. (A) Plot depicting expression of all miRNAs with deducible read counts. The x axis indicates ranks of miRNAs based on relative fold change expression (y axis). Colored points highlight members of various miRNA families, including let-7d, let-7g, let-7f, let-7b, and let-7i; mir-378f, mir-378g, mir-378e, mir-378b, mir-378a, mir-378i, and mir-378c; mir-30b; mir-129–5p; and mir-502–5p. (B) Heat map generated using multiexpression viewer (mev.tm4.org) includes fold changes of all significantly regulated miRNAs (FC ≥ 2 and P ≤ 0.001) in common between 1y-CMs and cEHTs relative to day 20-CMs. Yellow and blue indicate up- and down-regulation, respectively. Numbers: 1 and 2 indicate significantly up- or down-regulated miRNAs, respectively. (C) miRNA-mRNA target analysis using IPA with 1y-CM expression datasets: three miRNAs with the highest number of targets in 1y-CMs. P values reflect a one-sided Fisher’s exact test calculated using the total number of targets for each miRNA and the number of targets present in the dataset.

Let-7 Family Required and Sufficient for Maturation of hESC-CM.

To first test whether let-7 is required for maturation of hESC-CM, we targeted to KD all members of the let-7 family by constitutively OE Lin28a, a negative regulator of let-7, for up to 2 wk in Rockefeller University embryonic stem 2 (RUES2)-CMs. To do this, we used a lentiviral-based cloning vector, pLVX, carrying a Zs-Green reporter, and all analyses of let-7 KD were carried out when the CMs were roughly at day 30. The transduction efficiency attained by counting the number of Zs-Green-positive cells was up to 70 ± 10%. qPCR validated the lin28a expression to be 40-fold higher in Lin28a OE CMs compared with the empty vector (EV) control (Fig. 3A). Furthermore, we selected a member of the let-7 family, let-7g, for further qPCR validation and found that let-7g showed a significant down-regulation in Lin28a OE CMs (Fig. 3B). Interestingly, down-regulation of let-7 correlated with the repression of several known maturation markers in Lin28 OE CMs (Fig. 3C). This further encouraged us to characterize multiple parameters that have been shown to be modulated during cardiac developmental maturation (21). For these studies, we performed α-actinin (Z-disk protein) staining to visualize the EV control and Lin28a OE CMs (n = 3; >50 cells each) (Fig. 3D). We found a significant decrease in cell perimeter (Lin28a OE, 25 ± 3 μm vs. EV, 108 ± 13 μm; P < 0.001), cell area (Lin28a OE, 30 ± 17.5 μm² vs. EV, 400 ± 30 μm²; P < 0.001), and sarcomeric length (Lin28a OE, 1.1 ± 0.09 μm vs. 1.65 ± 0.13 μm; P < 0.001) (Fig. 3 E–G). Conversely, circularity index [4π area/(perimeter)²] increased from 0.44 ± 0.03 in EV to 0.60 ± 0.04 in Lin28a OE CMs (Fig. 3H). To determine whether the Lin28 OE phenotype is dependent on let-7 function, we overexpressed let-7g, using let-7g mimics in Lin28 OE CMs. Using multiple parameters, we found that let-7g OE was able to partially rescue the Lin28 OE phenotype (Fig. A–H). In addition, we also knocked down let-7g, using let-7g antagomir (Fig. 3I). Interestingly, KD of let-7g resulted in a phenotype similar to that seen in lin28OE CMs (Fig. 3 J–N and SI Appendix, Fig. S6A), suggesting a normal level of let-7 is required for maturation in hESC-CMs.

Fig. 3.

Let-7 is required for hESC-CM maturation. (A–K) All analyses done in EV control, Lin28a OE, and Lin28a OE+ let-7g OE CMs. (A–C) qPCR analysis to (A) examine Lin28a expression, (B) demonstrate that let-7g is down-regulated in Lin28a OE CMs but its expression is rescued in response to let-7g OE using let-7g mimics, and (C) evaluate the expression of maturation markers. (D) α-Actinin (green) and DAPI (blue) staining of representative CMs from the three treatments. (Scale bar = 25 µm.) (E–H) Compared with EV control, Lin28a OE CMs showed significant decrease in (E) cell perimeter, (F) cell area, and (G) sarcomere length and (H) an increase in circularity index. The phenotype was partially rescued in Lin28a OE CMs+let-7g OE. (I–N) All analyses done in SCRAMBLE (SCM) control and let-7g antagomir-treated CMs. (I) qPCR analysis to examine let-7g expression. (J) α-Actinin (green) and DAPI (blue) staining of representative CMs from the two treatments. (Scale bar = 25 µm.) (K–N) Compared with SCM control, let-7g KD CMs showed significant decrease in (K) cell perimeter, (L) cell area, and (M) sarcomere length and an increase in (N) circularity index. n = 50 cells per condition, three biological replicates. Means ± SEM are shown. **P ≤ 0.05 (Student’s t test). All experiments were repeated at least three times, and representative results are shown for D and J.

To further examine whether let-7 is sufficient to induce CM maturation, we selected two members of the let-7 family, let-7g and let-7i, according to their fold-change, as well as P values. These were further validated for their up-regulation using qPCR in cEHTs, 1y-CMs, and HAH samples in comparison with day 20-CMs (Fig. 4A). For further functional analyses, we used a lentiviral-based pLKO cloning system to independently overexpress these candidates for up to 2 wk in RUES2-CMs. The overall transduction efficiency of the lentivirus in the RUES2-CMs was assessed to be ∼60%, using a Ds-Red-encoding virus (SI Appendix, Fig. S5; n > 25 cells from three biological replicates). qPCR analysis validated let-7i and let-7g overexpression in CMs that were transduced with let-7 OE lentiviruses (Fig. 4B). In comparison with the EV control, let-7 OE CMs also exhibited a significant increase in all of the cardiac maturation markers that were previously found to be up-regulated in the cEHTs and 1y-CMs (Figs. 1B and 4C). However, let-7 OE did not change the expression of myomiRs. Similar results were obtained when let-7g OE was carried out using let-7g mimics. In this case, transient transfections were carried out in RUES2-CMs at day 15 and day 22, and end-point assays were done at day 30 (SI Appendix, Fig. S6 B and C). This provided the first indication that overexpression of let-7 could accelerate the maturation process. Applying the same parameters used for Lin28a OE CMs, we further characterized let-7 OE CMs. In contrast to what we observed with the Lin28a OE CMs, α-actinin (Z-disk protein) staining demonstrated a significant increase in cell perimeter (let-7i OE, 300 ± 7.4 μm; let-7g OE, 302 ± 3 μm vs. 108 ± 15 μm; P < 0.001), cell area (let-7i OE, 1,110 ± 101 μm²; let-7g OE, 980 ± 95 μm² vs. 380 ± 70 μm²; P < 0.001) (Fig. 4 D–F and SI Appendix, Fig. S6D) in let-7 OE CMs. Circularity index decreased in CMs that were overexpressing let-7i and let-7g vs. EV control (let-7i OE, 0.15 ± 0.04; let-7g OE, 0.12 ± 0.02 vs. 0.41 ± 0.02) (Fig. 4 D and G). We also found that the sarcomere length increased from 1.65 ± 0.02 μm in EV control cells to 1.70 ± 0.01 μm and 1.69 ± 0.01 (P < 0.001) in let-7i and let-7g OE samples, respectively (Fig. 4 D and H). An increase in sarcomeric length generally corresponds to an increase in the force of contraction.

Fig. 4.

Let-7 is sufficient for hESC-CM maturation. (A–C) qPCR analysis to (A) validate let-7i and let-7g expression derived from miRNA sequencing analysis from day 20-CM, in vitro-matured CMs, and HAH, and (B) demonstrate that let-7i and let-7g OE in RUES2-CMs results in increased expression of the two members. EV indicates empty vector control in RUES2-CMs (three biological replicates were analyzed for let-7 OE and EV samples). (C) Examine the expression of maturation markers in H7 day 20-CMs, EV, and RUES2-CMs. Gene expression is shown normalized first to GAPDH and then normalized to EV control. (D) α-Actinin (green), α-actin (red), and DAPI (blue) staining of representative EV control, let-7i OE, and let-7g OE CMs. (Scale bar = 50 µm.) Compared with EV control, let-7 OE CMs showed significant changes in (E) cell perimeter, (F) cell area, (G) circularity index, and (H) sarcomere length. n = 50 cells per condition, three biological replicates. (I) Representative force traces in EV control and let-7 OE CMs. (J) Significant increase in twitch force in let-7 OE CMs. n = 25 for EV control, n = 32 for let-7i OE, and n = 29 for let-7g OE from a total of three biological replicates. (K) Frequency of beating CMs. Compared with EV control, let-7 OE CMs show an increase in (L–N) APD, APD90, and (O) APD50/APD90. EV, let-7i OE, and let-7g OE CMs are collected at day 30 and hence are 10 d older than day 20 samples. Means ± SEM are shown. **P ≤ 0.05 and ***P ≤ 0.001 (Student’s t test). All experiments were repeated at least three times and representative images are shown for D.

To characterize force production on a per cell basis, we used arrays of microposts to measure their contractile forces (Fig. 4I) (38). EV control CMs exhibited a twitch force of 7.77 ± 0.7 nN/cell. Let-7i and let-7g OE CMs exhibited a significantly higher average twitch force of 11.32 ± 0.86 and 9.28 ± 0.7 nN per cell (P < 0.001), respectively (Fig. 4 I and J). In addition, let-7 OE CMs (let-7i OE, 1.05 ± 0.1 hz; let-7g OE, 0.92 ± 0.094 hz) exhibited lower beat frequency compared with EV control (1.57 ± 0.1h z). This decrease in frequency corresponds well with what is seen in in vivo human heart development (i.e., as CMs mature, they begin to exhibit reduced beating frequency) (39). To examine whether let-7 supports CM maturation at an electrophysiological level, we overexpressed let-7g and let7i in transgenic RUES2-CMs stably expressing a voltage sensor protein called Arclight (40, 41). Using the Arclight sensor, we found that induction of let-7i and let-7g prolonged the action potential duration at 90% (APD90) repolarization time at room temperature (500 ± 22 ms; control, 900 ± 90 ms; P < 0.01) (Fig. 4 L–N). Moreover, let-7i OE and let-7g OE CMs displayed an increase in the ratio of action potential duration (APD50/APD90) (Fig. 4O), suggesting let-7 overexpression drives the CMs toward more ventricular-type CMs. Consistently, we also saw an increase in the expression of CACNA1C, an L-type Ca channel protein, suggesting there is an increase in inward depolarizing current (Fig. 4C) in let-7 OE CMs. The increase in APD90 and APD50/90, as well as increased expression of CACNA1C, has been shown to occur during cardiac maturation (42, 43). These data together demonstrate not only that let-7 OE results in morphological and molecular changes indicative of maturation but also that functionally relevant parameters, such as APD, contraction, and beat frequency, are appropriately regulated.

To further understand the effects of let-7 OE during CM maturation at a molecular level, we carried out whole-genome transcriptome profiling of let-7g OE CMs and corresponding EV control CMs using an Illumina RNA sequencing platform. Consistent with our qPCR data, several known maturation markers such as ryanodine receptor 2 (RYR2), myosin heavy chain 7 (MyH7), and inward rectifier potassium channel protein KCNJ2, showed increased expression in the let-7g OE CMs compared with EV control (Fig. 5A and SI Appendix, Fig. S7A). Using expression values for the genes that belonged to the 12 pathways (Fig. 1E and Dataset S2), we carried out a 2D-PCA comparing let-7g OE CMs and EV control CMs with H7-CMs at day 20 and 1y, IMR90 iPSC CMs at 1y, HAH, and 3-mo-old HFA and HFV samples. This analysis clearly separated the day 20-CMs from 1y-CMs derived from H7 and IMR90iPSCs and HAH in dimension 1 (41% variance), suggesting dimension 1 portrays the effect of maturation (Fig. 5B and SI Appendix, Fig. S7B). Significantly, let-7g OE was closer to 1 y than the EV and day 20 CMs in the first dimension, suggesting overexpression of let-7g does indeed accelerate maturation. Further evidence of let-7g–directed maturation was observed from known isoform changes accompanying CM maturation, such as a decrease in ratio of myosin heavy chain 6/myosin heavy chain 7 (SI Appendix, Fig. S8) (44, 45). Further, a new differential splicing analysis tool (Materials and Methods) identified 80 isoforms that show a consistent differential splicing pattern across all of the sequenced samples, selected excluding let-7 OE CMs (Datasets S7 and S8). When comparing H7-CM day 20 and the EV control with H7-CM 1 y fetal and adult samples, all but three of these isoforms were found to change monotonically, either increasing or decreasing in relative expression with maturity (Fig. 5C), indicating that despite the variety of tissues sequenced, cell maturation is the strongest determinant in the splicing changes we observe. We then used these isoforms as a benchmark of splicing maturity, evaluating splicing rates in let-7 OE CMs. Hierarchical clustering groups let-7 OE CMs with the 1 y and fetal samples (Fig. 5C), and a similar pattern is seen when principal component analysis is applied (Fig. 5D). In short, concerted and dynamic changes in splicing during maturation are observed in in vitro-matured and let-7 OE CM samples. Interestingly, among the 80 genes, several have been shown to be involved in cardiogenesis, including troponin T2 (TNNT2) (46) (SI Appendix, Fig. S9). The fact that both differential splicing and differential expression analyses cluster let-7 OE CMs with H7 1y-CM and fetal samples (Fig. 5 B and D) clearly strengthens the finding that let-7 is critical for maturation. These results demonstrate that let-7 is not only required but also sufficient for maturation of hESC-CMs.

Fig. 5.

Let-7 is critical for cardiac maturation (A, B, E, and F are analyses done with gene expression analyses and C and D are analyses based on splice variant signatures). (A) Scatter plot of let-7g OE (y axis) vs. EV control (x axis) from the mRNA sequencing dataset. Red dots indicate maturation marker genes in the dataset. A few are labeled in the plot: troponin I type 3 (TNNI3); gap junction protein alpha 1 (GJA1); actin alpha cardiac muscle 1 (ACTC1); myosin heavy chain 7 (MYH7); ryanodine receptor 2 (RYR2); potassium channel, subfamily J2 (KCNJ2); sodium channel protein 5 alpha (SCN5A); sarco endoplasmic reticulum Ca²⁺ATPase 2 (SERCA2); troponin T type 2 (TNNT2); calcium channel, voltage dependent, alpha 1C (CACNA1C). (B) 2D-PCA using mRNA signatures from 12 pathways (indicated in Fig. 1E) across the analyzed samples, as indicated in the figure. (C) Heat map showing the proportion of each of the 80 isoforms identified as differentially spliced across each condition. Each value is the estimated proportion of that isoform among all expressed isoforms of the same gene in that condition. (D) 2D-PCA based on the proportions of the 80 identified differentially spliced transcripts, applied to all replicates from these eight conditions. (E) Heat map demonstrating changes in gene expression of 12 different pathways between EV control and let-7g OE CMs. Left to right, columns 1–2 and 3–5 represent biological replicates of EV and let-7g OE CMs, respectively. The rows reflect read counts of various genes in the different categories. Rows are standardized individually and colored according to the Z score. Yellow and blue represent up- and down-regulation, respectively. (F) Density plots using R generated with fold change expression (x axis indicates log₂-fold change) of genes from four categories, indicative of cardiac function for let-7 OE/EV CMs. Black curve indicates expression of all genes. Curves toward the left and right side of the black curve indicate down-regulation and up-regulation of pathways, respectively.

Let-7 Promotes hESC-CM Maturation by Acting as a Metabolic Switch.

To understand the molecular signaling components of the maturation program that are modulated in let-7g OE CMs, we further probed the transcript profiling data from let-7g OE CMs for each of the 12 pathways previously identified (Fig. 5E). Pathways related to Ca signaling, G protein-coupled receptor signaling, cAMP-mediated signaling, and cardiac beta adrenergic signaling and hypertrophic signaling were significantly up-regulated in let-7 OE CMs, similar to that seen in 1y-CMs and fetal heart tissue samples (Fig. 5 E and F and SI Appendix, Fig. S10A). Importantly, fatty acid metabolism was significantly up-regulated, whereas PI3/AKT/insulin signaling was significantly down-regulated in the let-7g OE CMs in comparison with EV control (Fig. 6A). Programs related to cell cycle, actin-cytoskeleton, and integrin signaling also showed the correct trends (SI Appendix, Fig. S10 A and B). The inverse relationship between fatty acid metabolism and PI3/AKT/insulin signaling in let-7g OE CMs was similar to that observed in the 1-y-old CMs and consistent with the metabolic switch seen in maturing CMs in in vivo studies. Using let-7 OE CMs, we validated by qPCR the down-regulation of candidate let-7 targets such as EZH2 and those in the insulin pathway, as well as the up-regulation of genes in fatty acid metabolism (Fig. 6B).

Fig. 6.

Let-7 OE accelerates CM maturation. (A–F) Let-7 OE results in down-regulation of the PI3/AKT/Insulin pathway and up-regulation of fatty acid metabolism. Comparisons were done between let-7 OE and EV control for all assays. (A) Density plots using R generated with fold change expression (let-7g OE/EV) of genes for fatty acid metabolism and PI3/AKT/insulin signaling. (B) qPCR analysis of candidate let-7 targets and genes from the fatty acid metabolism. (C) Representative OCR profile in response to ATP synthase inhibitor oligomycin, uncoupler of electron transport and oxidative phosphorylation, FCCP, and electron transport chain blockers rotenone and antimycin during mito-stress assay. (D) Quantification of maximal respiration capacity; that is, changes in response to FCCP treatment after inhibition of ATP synthase by oligomycin. n = 24 from three biological replicates. (E) Representative OCR trace of let-7g OE CMs for fatty acid stress measuring Etomoxir (ETO)-responsive OCR changes after the second dose of palmitate addition. (F) Quantification of changes in OCR in response to ETO. n = 32 from three biological replicates. Means ± SEM are shown. **P ≤ 0.05 (Student’s t test). (G–K) Knock-down of IRS2 and EZH2 results in up-regulation of fatty acid metabolism and improved expression of cardiac maturation markers. (G and H) Representative OCR profile and quantification of maximal respiratory capacity in siIRS2-CM and siEZH2-CM. (I and J) Representative OCR trace for fatty acid stress using palmitate and quantification of OCR change (E). (K) qPCR analysis of cardiac maturation markers, fatty acid metabolism genes.

To test the functional relevance for these gene expression changes, we carried out metabolic analysis of let-7 OE CMs vs. EV control, using the Sea Horse metabolic flux assay. First, we analyzed mitochondrial maximal respiration capacity by measuring the oxygen consumption rate (OCR), a metabolic parameter representing mitochondrial respiration levels. To record the maximum activity of the electron transport chain uncoupled from ATP synthesis, mitochondrial ATP synthase was inhibited with oligomycin, and then maximum mitochondrial respiration was measured after addition of the proton gradient discharger, carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP). The OCR changes were significantly greater after FCCP treatment in let-7i OE and let-7g OE CMs compared with in EV control samples (Fig. 6 C and D). Increased mitochondrial respiration could be a result of increased mitochondrial copy number or increased mitochondrial activity; for example, as a result of higher efficiency of glucose or fatty acid use as energy substrates. We determined mitochondrial genome copy number by qPCR and found that let-7 OE had no effect on this parameter (SI Appendix, Fig. S11A). Interestingly, fatty acid stress test using palmitate revealed that the let-7g OE and let-7i OE CMs have greater OCR increase in response to palmitate than EV control (Fig. 6 E and F). To investigate whether the let-7 OE CMs also use glucose more efficiently, the extracellular acidification rate was determined. No significant difference was observed in the maximum extracellular acidification rate changes in the glucose stress assay for let-7 OE compared with for EV control (SI Appendix, Fig. S11 B and C). Together, these data indicate that the let-7 family of miRNA likely induces hESC-CM maturation in part by promoting a higher use of fatty acids to meet the CM’s increased energy demands. To further dissect the mechanism of let-7 function in this process, we chose to KD two validated targets of let-7: IRS2, a member of an insulin signaling pathway, and EZH2, a histone methyl transferase known to regulate gene expression (47). These two candidates were found to be down-regulated in let-7g OE CMs (Fig. 6B). Interestingly, siRNA KD of IRS2 and EZH2 in RUES2 CMs (SI Appendix, Fig. S11D) resulted in an increase in OCR (Fig. 6 G and H), as well as a higher efficiency in the use of palmitate (Fig. 6 I and J), compared with their siLuciferase control. Moreover, KD of IRS2 and EZH2 resulted in an increased expression of cardiac maturation genes as well as genes from fatty acid metabolism (Fig. 6K). However, E3 ubiquitin-protein ligase TRIM71, a target of let-7, did not show any change in expression in both siIRS2 and siEZH2 CMs (Fig. 6K). These data suggest that let-7 exerts its function on cardiac maturation in part by simultaneously acting on two of its targets, IRS2 and EZH2.

Let-7g Exclusively Promotes in Vitro Cardiac Maturation, and Not Early Cardiac Commitment.

Previous reports have demonstrated that induction of let-7 silences ESC self-renewal, or inhibition of let-7 promotes dedifferentiation of fibroblasts to a pluripotent state, showing that let-7 inhibits a stem cell state and suggests let-7 could have a role in early lineage commitment (48). Similarly, using in vivo mouse studies, Colas et al. (49) have shown that let-7 can promote mesodermal commitment during embryonic development. However, none of these studies have specifically demonstrated the role of let-7 in early cardiac commitment. To address this, we first carried out an elaborate time course expression analysis of let-7g, a candidate member of the let-7 family (SI Appendix, Fig. S12 A and B). We found that the expression of let-7g is very minimal at stages of mesodermal commitment (days 2–3), as well as commitment to cardiac lineages (day 5). However, a dramatic increase is observed during later points in our in vitro cardiac differentiation system (day 20–1 y). To gain more insights on this, we overexpressed let-7g using let-7g mimics at day 5 of the differentiation, where previous report has shown that cardiac commitment occurs (50). After 3 d of early let-7g OE, we monitored the expression of three known early cardiac markers [myosin light chain 2 (MLC2), basic helix-loop-helix family of transcription factor HAND2, zinc-finger family of transcription factor GATA4], whose expression was found to increase at early points (51). Interestingly, overexpression of let-7g, even as high as 1,000-fold (SI Appendix, Fig. S12C), and down-regulation of some of its targets (SI Appendix, Fig. S12D) did not change the expression of these cardiac markers after 3 d (SI Appendix, Fig. S12E). These data altogether demonstrate that the function of let-7g is rather exclusive during cardiac maturation.

Discussion

In this study, for the first time to our knowledge, we demonstrate that the let-7 family of miRNAs is required and sufficient for maturation of hESC-CMs. A wide range of functional, physiological, electrophysiological, and molecular parameters indicates that induction of candidate members of the let-7 family is sufficient to enhance a number of functional properties relevant to CM maturation. In contrast, KD of let-7 either by Lin28 OE or using antagomir for let-7g results in attenuating the process. Interestingly, introduction of let-7g mimics partially rescues the Lin28 OE phenotype in CMs, suggesting let-7 acts independently of Lin28 in the cardiac maturation pathway. Previous genetic studies have shown that the shift in cardiac fuel preference from glucose to fatty acids taking place during fetal to postnatal transition (3) is largely mediated by the PPARα-PPARγ coactivator-1α axis (52). However, our study has identified a previously unsuspected regulator that could promote this metabolic transition in sync with cardiac maturation. Overexpression of let-7g and let-7i specifically accelerates the CM’s capacity to use fatty acid as a major energy source without affecting mitochondrial copy number or improving the efficiency of glycolysis. Because these metabolic and functional changes mimic changes during postnatal maturation, we posit that induction of let-7 in young CMs results in changes that are equivalent to physiological hypertrophy (3). Thus, in addition to studies that have proposed using let-7 as a therapeutic tool for attenuating myocardial infarction (53), we propose that this miRNA can be potentially used for improving cardiac function during maturation.

Maturation of CMs is complex and bound to be regulated by multiple pathways acting simultaneously. In this context, let-7 is a great candidate, as it regulates a multitude of target genes that could potentially promote maturation. A direct inhibition of two such let-7 targets, IRS2 and EZH2, phenocopied let-7 OE on cardiac maturation, suggesting let-7 imparts its metabolic effect on CM maturation by acting on at least two of its targets. EZH2 is a histone methyltransferase and component of polycomb repressive complex, which methylates H3K27, resulting in the transcriptional repression of affected targeted genes (47). One attractive hypothesis is that EZH2 either directly or indirectly represses genes involved in fatty acid oxidation, and let-7-driven down-regulation of EZH2 releases transcriptional repression of fatty acid oxidation genes. Consistent with this hypothesis, repressing EZH2 resulted in an increased expression of candidate fatty acid genes. However, further experiments are needed to elucidate the interplay of IRS2, EZH2, and other key let-7 targets in cardiac maturation in greater detail. The fact that TRIM71, another candidate let-7 target, remains unchanged in siIRS2-CM and siEZH2-CM, but is significantly repressed in let-7 OE CMs, clearly suggests the multipronged function of let-7 as a developmental switch (SI Appendix, Fig. S13). Because let-7 is found to be highly expressed in other cell types such as late retinal progenitors and glial cells (54, 55), further studies to address the mechanism by which let-7 regulates cardiac maturation would also shed light on maturation of other cells types in vitro. In conclusion, our discovery of a small RNA essential to promote cardiac maturation and metabolic transition provides a unique focus for the rational development of strategies for generating mature tissue types useful for regenerative medicine.

Experimental Procedures

Cell Culture.

Undifferentiated hESC lines H7 (NIHhESCC-10-0061) and RUES2 (NIHhESC-09-0013) were expanded using mouse embryonic fibroblast-conditioned medium and subjected to directed differentiation under serum- and insulin-free conditions.

Immunocytochemistry and Morphological Analysis.

Cells were fixed in 4% (vol/vol) paraformaldehyde, blocked for an hour with 1.5% (vol/vol) normal goat serum, and incubated overnight with mouse alpha actinin (sigma clone EA-53) primary antibody, followed by secondary antibody [(Alexa fluor 488 (Goat anti mouse)] staining. Measurements of CM area, perimeter, and sarcomere length were performed using NIH Image J 1.44 software (rsb.info.nih.gov/nih-image/).

mRNA and miRNA Sequencing Analysis.

RNA and small RNA libraries were prepared independently, using Truseq library preparation kits (Illumina), following the manufacturer’s protocols. Sequence data for miRNAs were analyzed using miRDeep2 software. Both principle component analysis and hierarchical clustering were performed using log-transformed Transcript per million gene expression estimates, using R 3.0.2. Heat maps for the RNA and miRNA sequencing data were generated using the edgeR version 3.2.4 and multiexperiment viewer (www.tm4.org), respectively.

Splice Variant Analysis.

To define a set of differentially spliced genes, we looked for changes in isoform proportions, defined as an isoform's expression divided by the sum of the expression of all isoforms with the same transcription start site. To call differential splicing, we looked for isoforms that showed a specific and consistent ordering of the immature (day 20 and EV CM) versus the mature (fetal, 1 y, and adult) samples with posterior probability >0.75. Of all but three of the 80 isoforms called, this manifested as a consistent increase or decrease between immature and mature. To increase the chances of detecting changes in splicing dynamics, the AceView gene annotations were used, which more liberally include alternate isoforms (56).

Contractile Force Measurement.

Muscle twitches from individual hESCs-CMs were recorded with high-speed video microscopy within a live cell chamber at 37 °C, as described by Rodriguez et al. (38).

Detailed information on cell culture, lentiviral transduction, immunohistochemistry, flow cytometry, Arclight reporter analysis, qPCR and sequencing analysis, force contraction assay, and mitochondrial functional assay are provided in the SI Appendix.