Introduction

The heart is the first functional organ to develop in the human embryo, and the organ most commonly affected by disease in infants and adults. Heart development is tightly controlled by an evolutionarily conserved network of transcription factors and disruption of this network can result in a variety of congenital heart malformations. MicroRNAs (miRNAs), short (19–22 base pair) RNA regulatory molecules, add another layer of regulatory precision to reinforce core cardiac transcriptional networks1. Furthermore, the cardiogenic regulatory framework is reinforced by the direct control of certain microRNAs by critical myogenic transcription factors, including serum response factor (SRF), myocyte enhancer factor-2 (MEF2c) and GATA42,3,4. In addition to their key roles in heart development, miRNAs are critical to maintaining cardiac tissue homeostasis and function1,5. For example, global perturbation of the cardiac miRNAome via deletion of the microRNA processing enzymes Dicer and DGCR8 leads to dilated cardiomyopathy6,7,8. In view of this, miRNAs emerged as a new class of functional regulators of cardiomyogenesis and heart disease1,9,10,11,12. Given their important role in both heart development and function, careful compellation of the human cardiomyocyte miRNAome will facilitate studies in a number of areas in cardiac biology. Human pluripotent stem cells offer an opportunity to study the expression profile of miRNAs in human heart muscle cells13,14,15,16,17.

Here, we have used a well-established NKX2-5 cardiac reporter line18 and loss-of-function model19 to determine whether the human cardiac miRNAome is directly dependent on NKX2-5, a transcriptional factor that is essential for mammalian heart development19,20,21. We show that the cardiomyogenic miRNA program is activated early during human cardiomyocyte differentiation in vitro. The expression of the myogenic miRNAs9, MIR1-1, MIR133A1, MIR208A and MIR499A, was established by day 10 of pluripotent stem cell differentiation into the cardiac lineage. However, NKX2-5 was dispensable for maintenance of the human cardiomyocyte miRNAome, with no differentially expressed miRNAs identified in either cardiac committed mesodermal progenitors or immature differentiated cardiomyocytes. These data suggest that, in the main, establishing the cardiomyogenic miRNA program occurs independently of the highly conserved core NKX2-5 gene regulatory network.

Results

Defining miRNA expression profiles in hESC derived human cardiac progenitors

We utilized the NKX2-5eGFP/w hESC reporter cell line, in which enhanced GFP (eGFP) is expressed from the NKX2–5 locus18, to isolate hESC-derived cardiac mesodermal progenitors and cardiomyocytes by flow cytometry. Using a monolayer differentiation protocol22,23 GFP+ myogenic-lineages emerged between days 6 to 14 of the differentiation (Fig. 1a). To obtain a profile of miRNAs during early cardiomyocyte differentiation, cells were harvested and sorted into three populations: NKX2-5 negative mesoderm (day 6 GFPneg cells), cardiac mesodermal progenitors (day 6 GFP+ cells) and early cardiomyocytes (day 10 GFP+ cells) for sequencing of small RNA species (Fig. 1b,c). A summary of small RNA sequencing data is shown in Supplementary Table 1. The average number of high-quality reads in the day 6 GFPneg, day 6 GFP+ and day 10 GFP+ populations were 2.01 ± 0.10 × 106 (n = 3), 1.71 ± 0.08 × 106 (n = 3), and (2.35 ± 0.08) x 106 (n = 3), respectively. Small nucleolar RNAs (snoRNAs) were also detected in day 6 GFP+ and GFPneg populations (1.18 ± 0.24 × 105 (n = 3) and 3.19 ± 1.7 × 105(n = 3),reads respectively) and day 10 cardiomyocytes (1.6 ± 0.17 × 105 (n = 3)). Thus, other small non-coding RNA populations such as snoRNAs are also present during human cardiac differentiation, but their functions are currently unclear.

Figure 1
figure 1

MicroRNA expression profiling of NKX2-5 positive cardiac progenitors and cardiomyocytes. (a) Epifluorescent images of differentiating NKX2-5eGFP/w cells (Scale bar, 100 μm). (b) Percentages of GFP positive and negative cells at day 6 and 10 of differentiation. Data shows mean ± SEM (n = 3) determined by flow cytometry. (c) Flow cytometric purification of cardiac progenitors and cardiomyocytes based on eGFP expression from the NKX2-5 locus. Percentage of cells are indicated on plots. (d) Heat map of unsupervised hierarchical clustering of microRNA sequence profile showing that samples cluster according to day of differentiation (log2 RPM values). (e) Dot plot representation of miRNA-seq absolute expression (log2 RPM values) for miRNAs at day 6 of differentiation. Dashed lines mark 2-fold differential expression level. (f) Dot plot representation of miRNA-seq absolute expression (log2 RPM values) comparing day 6 and day 10 GFP positive cells. Dashed lines mark 2-fold differential expression level. (g) Q-PCR profiling supports miRNA sequencing data and confirms upregulation of canonical myogenic miRNAs in GFP+ cells at day 6. (h) Q-PCR of myomiRs demonstrates they are more highly expressed in day 10 cardiomyocytes than day 6 GFP+ cells.

Full size image

Unsupervised hierarchical clustering indicated that the miRNA profiles of day 6 GFP+ (cardiac mesoderm progenitors) and GFPneg populations were more closely related to each other than to day 10 cardiomyocytes (Fig. 1d and Supplementary Table 2). The similarity of miRNA expression profiles between the NKX2-5 positive and negative populations at day 6 most likely reflects the recent emergence of the NKX2-5+ cells from the pool of mesodermal cells. Differential expression analysis between day 6 GFP+ and GFPneg cells identified seven miRNAs enriched in the GFP+ population (Fold change >2 with adjusted p-value < 0.05)(Fig. 1e; Supplementary Table 2). Among the seven overrepresented miRNAs, six have been reported to have important roles in murine cardiac development. For example, MIR133A1 and MIR133A2 play critical roles in promoting pre-cardiac mesoderm, while suppressing endodermal and neuroectodermal lineages24. We also observed higher levels of MIR125B1 in d6 cardiac progenitors, which plays a role in mesoderm development25 and regulates MEF2d in the HL-1 atrial myocyte cell line. Q-PCR of the cardiomyogenic miRNAs MIR1-1, MIR133A1 and MIR208A showed these were more highly expressed in day 6 cardiac mesoderm while MIR499 was not differentially expressed between GFPneg and GFP+ cells (Fig. 1g,h). In addition, four miRNAs (MIR122, MIR126, MIR1247 and MIR127) were expressed at higher levels in GFPneg mesodermal progenitors than GFP+ cells. (Fig. 1e; Supplementary Table 2). These miRNAs have been implicated in a wide range of non-cardiac processes such as liver homeostasis, hematopoiesis and self-renewal of cancer stem cells26,27,28.

To identify miRNAs that may be important drivers of cardiomyocyte differentiation, we compared the expression profile of d6 and d10 GFP positive cells (Fig. 1d,f). This analysis identified a total of 112 differentially expressed miRNAs (Supplementary Table 2). Studies in the mouse have identified a core miRNA network involved in cardiomyocyte function9. The known myomiRs (miRNA-1-1,-1-2,133b and -208)9 were all expressed at higher levels at day 10 compared to day 6. Moreover, MIR499, a miRNA located within the intron of myosin heavy chain 7b (MYH7b), which encodes a cardiomyocyte sarcomeric protein, showed the highest differential expression (5.62 fold) in day 10 cardiomyocytes29. Q-PCR profiling of miRNA expression in day 10 GFP+ cardiomyocytes confirmed these key miRNAs are more highly expressed in cells committed to the cardiomyocyte lineage (Fig. 1h). MIR1-1, MIR133A1, MIR208A and MIR145 expression levels are further upregulated in the day 10 NKX2-5eGFP/w GFP+ populations relative to day 6 populations. The data indicates that expression of canonical myogenic miRNAs is established early in cardiomyocyte differentiation.

A set of miRNAs that regulate DNA synthesis were more highly expressed in the mesoderm progenitor stage compared to day 10 cardiomyocytes (Supplementary Fig. 1). The Hippo-YAP pathway is an established driver of cardiomyocyte proliferation18,19. It is possible that later in cardiomyocyte development NKX2-5 is important for maintaining the appropriate mix of miRNAs in the heart. Nevertheless, our data suggest that in the early stages of human heart development the miRNAome does not require NKX2-5. Conversely, other key regulators of cardiac muscle development such as MEF2c, TBX5 and GATA42,3,4,45,46, which are not disrupted in NKX2-5 null cardiomyocytes19, may have a role in regulating cardiomyogenic miRNAs. Studies focussed on the regulatory nexus between the cardiac gene regulatory network19,47,48 and the miRNAome17 in the human context will be required to fully understand the transcriptional regulation of the cardiac microRNA network.

Methods

Ethical approvals

All experiments were approved by the Royal Children’s Hospital Research Ethics Committee (HREC 33001 A). Methods were carried out in accordance with the relevant guidelines and regulations provided by National Health and Medical Research Council (National Statement on Ethical Conduct in Human Research).

Cell culture and cardiac differentiation

All cell culture reagents were purchased from Thermo Fisher unless stated otherwise. HES3 NKX2-5eGFP/w and NKX2-5eGFP/eGFP cell lines were routinely passaged using TrypLE Select and maintained on tissue culture flasks, as previously described19. To induce differentiation, hESCs were dissociated into single-cell suspension using TrypLE Select and seeded onto Geltrex coated (1:100 dilution) culture plates at 2.5 × 105 cells/cm2 in basal differentiation media consisting of RPMI (Thermo 61870), B27 minus vitamin A (Thermo 12587) and 50 µg/ml ascorbic acid (Sigma A92902), further supplemented with 10 µM CHIR99021 (Tocris Bioscience 4423) and 80 ng/mL Activin A (Peprotech). Following 24 and 96 h of induction, media was replaced with basal media supplemented with 5 µM IWR-1 (Sigma I0161) and from day 5 onwards, differentiating cultures were maintained in basal media only until harvested for analysis.

FACS analysis

To isolate cardiac-cell lineages from non-cardiac cell lineages, live cell sorting based on eGFP expression of differentiating cultures at either day 6 or day 10 were performed. Differentiating hESCs were dissociated, filtered through a 40 µM cell strainer and resuspended in PBS containing 2% fetal calf serum and 1 µg/mL propidium iodide. Flow cytometric gates of GFP positive and negative cells were set using negative control HESCs cultured in basal media containing 100 ng/ml bFGF. Sorted GFP positive and negative cells on day 6 and day 10 of differentiation were collected and snap frozen prior to RNA extraction. Cell sorting was done using BD InfluxTM cell sorters (BD Biosciences) and flow cytometric data was analyzed using Flowlogic software (Inivai Scientific).

Next generation sequencing and bioinformatics analysis

Total RNA from GFP sorted populations on day 6 and day 10 differentiating cultures were prepared in triplicate using the miRNeasy Mini (Qiagen 217004). A minimum amount of 500 ng of total RNA was analyzed for RNA integrity and submitted for sequencing. Small RNA libraries were prepared using Truseq Small RNA Sample Prep Kit (Illumina) and small RNA sequencing was performed on the Illumina Hi-Seq platform using the Illumina CASAVA 1.8.2 software (Australian Genomic Research Facility). Short RNAs were sequenced as 50 base pair single end reads and all adapter and primer sequences were removed using cutadapt49. Bowtie aligner was used to map the 17-26 base pair single end reads to the human reference genome (hg19) allowing one mismatch. The uniquely mapped reads were summarized across microRNAs with featureCounts (Rsubread v1.20.6)50 using ensemble miRNA gene annotation. Lowly expressed miRNAs were filtered out (less than 10 counts per million in fewer than three samples). The data was Voom transformed with cyclic loess normalization, and differential expression assessed using empirical Bayes moderated t-tests from the R Bioconductor limma package (Version 3.20.9)51 through the statistical language R. Unbiased hierarchical clustering was performed using standard complete linkage and Euclidean distance. Data has been deposited on GEO (GSE134852).

Quantitative PCR

To analyze miRNA gene expression levels, quantitative real time PCR was performed on differentiating cultures at day 6 and day 10. 1 µg of RNA was reversed transcribed using SuperScript® III (Invitrogen) and miRNA expression levels were determined using Taqman Gene Expression Assays with Taqman Universal PCR Master Mix (ThermoFisher) on the ABI 7300 Real-Time PCR detection system (Applied Biosystems). Transcript expression levels were normalized to the averaged expression of the reference gene snoRNA RNU24, and gene relative quantification was calculated using the 2−∆∆Ct method.