miRNA expression analysis in the human heart: Undifferentiated progenitors vs. bioptic tissues—Implications for proliferation and ageing

Abstract In developed countries, cardiovascular diseases are currently the first cause of death. Cardiospheres (CSs) and cardiosphere‐derived cells (CDCs) have been found to have the ability to regenerate the myocardium after myocardial infarction (MI). In recent years, much effort has been made to gain insight into the human heart repair mechanisms, in which miRNAs have been shown to play an important role. In this regard, to elucidate the involvement of miRNAs, we evaluated the miRNA expression profile across human heart biopsy, CSs and CDCs using microarray and next‐generation sequencing (NGS) technologies. We identified several miRNAs more represented in the progenitors, where some of them can be responsible for the proliferation or the maintenance of an undifferentiated state, while others have been found to be downregulated in the undifferentiated progenitors compared with the biopsies. Moreover, we also found a correlation between downregulated miRNAs in CSs/CDCs and patient age (eg miR‐490) and an inverse correlation among miRNAs upregulated in CSs/CDCs (eg miR‐31).

thrombolytic drugs, or heart transplantation, which is restricted by organ availability and donor age. In this respect, miRNAs may represent a valid therapeutic option for cardiac disease (revised by Hudson and Porrello 4 ). miRNAs are small noncoding RNAs of about 22 nucleotides (mature miRNAs) that regulate the post-transcriptional expression of several genes. After the transcription, the primary miRNA transcripts are 5' capped and 3' polyadenylated in the nuclei and processed by Drosha-DGCR8 complex to pre-miRNAs hairpin loop structure (canonical pathway) or by the spliceosome to mirtrons through direct splicing of mRNA introns and refolded into a pre-miRNA (alternative pathway). The pre-miRNAs are exported in the cytoplasmic region by exportin 5 and cleaved by Dicer proteins, which remove the loop from the pre-miRNAs structure, generating the miRNA duplex.
Finally, this duplex is unwound, and the guide filament of the miRNA is assembled with the RNA-induced silencing complex (RISC) and is ready to carry out its function by binding the 3'-untranslated region (3'-UTR) of mRNA target, repressing the translation or the deadenylation, in the case of partial complementarity with the mRNA, or degradating the mRNA when complementarity occurs, leading to the silencing of the protein synthesis. 5 Each miRNA can control different genes, and each gene can be regulated by several miRNAs, generating a complex network.
More than one thousand miRNAs have been described in humans to regulate about 20%-30% of the translation processes. In the healthy adult human heart and in cardiovascular pathological conditions, miRNA expression has been identified by large sequencing approaches. 6 The role of various miRNAs has also been defined 7,8 during embryogenesis, in postnatal cardiac development in the adult healthy heart and in cardiac remodelling. 9 As demonstrated for many tissues, the human heart is characterized by a subpopulation of adult stem cells with the properties to differentiate into cardiomyocytes, to express stem cell markers with clonogenic features, and with the ability to maintain their functionality in vitro and in vivo. These cells are the cardiosphere-derived cells (CDCs) and cardiospheres (CSs), which represent the progenitor/ stem cell-enriched population. 10 CDCs have been tested in clinical trials in which autologous CDCs are isolated from patients, and intracoronarially infused after AMI. Two major studies have demonstrated that their use decreased scar size, improving myocardial function and regeneration. 11,12 Nevertheless, these cellular treatments showed limited advantages.
The cost for the manipulation and culture of the explanted cells in safe conditions calls for new approaches in AMI treatments. A characterization of the genes regulated by miRNAs could be useful in clarifying the mechanisms in the human heart, particularly in the CSs/CDCs counterpart, opening the possibility for their use in therapy. Recently, it was reported that specific miRNAs promote cardiomyocyte proliferation in rats. 13 Three different groups of genes for pluripotent, transitional and mature cardiomyocyte stage were identified from gene analysis of mRNA expression during the differentiation of induced pluripotent stem cells (hiPSCs) from foetal, adult and hypertensive human heart biopsies. Moreover, a miRNA pattern controlling the stem process into the developmental stage at the transcriptional level was identified. 14 The aim of our work was to characterize miRNA expression in human heart biopsies and in their CS-and CDC-derived cell populations through microarrays and next-generation sequencing (NGS) approaches. Initially, we performed the miRNA expression analysis using the Agilent microarray platform and found that a set of miRNA clusters was preferentially expressed in CSs and CDCs compared with the biopsies. To have a wider miRNA profile of the three cell populations, we used an NGS approach: while some groups of miRNAs preferentially expressed in CDCs and CSs found with microarrays were also validated with NGS analysis, other miRNAs were specifically identified as more expressed or downregulated in CDCs and CSs than the biopsies. We identified a positive miRNA expression correlation between the microarray and NGS platforms.
Moreover, we identified miRNAs preferentially expressed in CSs and in CDCs, while others were downregulated. Among them, we validated some by real-time polymerase chain reaction (PCR), and we analysed the correlation between expression and patient age.
Moreover, analysis of the NGS results with Diana miRPath software revealed that in CDCs and CSs, miRNAs were involved in the Hippo pathway, the stem cell pathway, the signalling that regulates stem cell pluripotency, and in the signalling involved in arrhythmogenic right ventricular cardiomyopathy.

| Biopsy specimen processing and cell culture
The isolation, culturing and expansion of cardiac progenitors were carried out on fresh heart biopsies from patients who had undergone extracorporeal circulation. Human surgical auricola biopsies were obtained as part of routine surgical intervention in the Cardiac Surgery and Heart Transplantation Unit at ISMETT, in accordance with institutional guidelines and ISMETT's Ethics Committee. AVRm, aortic valve regurgitation; AAA, ascending aortic aneurysm; BAV, bicuspid aortic valve) (mean age 65 years) ( Table 1).
As previously described, 10  The cells were also evaluated for specific markers by RT-PCR. 15

| RNA extraction and RT-PCR
Total RNA was purified by miRNAeasy (Qiagen, Germantown, MD, USA) and reverse-transcribed using TaqMan UNIVERSAL MMix II (Applied Biosystems, Waltham, MA, USA) for random priming or miRNA-specific assay reverse transcription. Semi-quantitative PCR was performed with TaqMan-validated assays (Applied Biosystems). As endogenous reference gene for cDNA, we chose U6 (#001973) for miRNA. All analyses were carried out in triplicate. Real-time data were collected using Microsoft Excel and analysed with the following formula: Expression level = 2-ΔΔCt method. 17 All experiments were done as independent triplicates and analysed using standard deviation (SD). The p value was obtained with Student's t test.

| NGS analysis
Total RNA was extracted using the miRNeasy Isolation Kit (Qiagen, Hilden, Germany). Sequencing libraries were prepared according to the Illumina Protocol for small RNA (Illumina, San Diego, CA, USA release Feb. 2014), as previously described. 18 One microgram of total RNA was processed using the small RNA library

| Cluster analysis correlation
Up-/downregulated miRNAs (+/− 2 fold change) were loaded onto DianaTools (mirPath v.3) using the option tarbase v7; we used 6 for the gene intersection analysis on the KEGG pathway. After analysis, we identified the more significant pathways (stem cell, arrhythmogenic and Hippo). The relationship between miRNAs and pathways was represented graphically with Cytoscape, while the relationship between miRNAs and genes was represented with R 24 (version 3.5.1).

| RE SULTS
To evaluate miRNAs involved in cardiac stemness/proliferation phenotypes, we analysed miRNA expression profiles in human heart biopsies and in their derived CSs/CDCs. Cardiac biopsies ( Figure 1A) were used to isolate stem/progenitor cells that originated by sprouting in appropriate culture conditions, and a part processed to extract RNA. 15 The expanded progenitors were used for RNA extraction after 3-6 passages of sequential growth as CSs and CDCs, where CS growth as clusters ( Figure 1B), and CDCs showed a fibroblastoid morphology ( Figure 1C).
As a first approach, we evaluated miRNA expression using the  (Figures 4,5). In particular, progenitor-overexpressed miRNAs were hierarchically clustered together, while in another analysis miRNAs were downregulated in undifferentiated cells (Figures 4,5). The most upregulated and downregulated miRNAs are summarized in the tables (Table 2: miR-NAs upregulated in progenitors; Table 3: miRNAs downregulated in progenitors).
We evaluated the reliability of our NGS results by choosing some miRNAs to check the differential expression through RT-PCR analysis ( Figure 6A). In particular, we observed that miR-1, miR-10b, miR-133a and miR-490 were downregulated in CDCs and CSs, while they were highly expressed in differentiated human heart tissues. The expression of miR-31 was higher in CDC and CS samples ( Figure 6A).
Notably, the expression of the miR-133 family, as assessed in microarray analysis, was remarkable in its role in human heart F I G U R E 3 Scatter plot of total miRNA identified by NGS. On the x-axis, the means of miRNA expression for biopsies are reported, and on the y-axis, the miRNAs expressed by CDC/CS cells. miRNAs more expressed in progenitors are on the upper left side, and miRNAs more expressed in differentiated heart biopsies are on the lower right. Blue dots represent differentially expressed miRNAs among different samples (fold change, fc=±2; p<0.05).

BIOPSIES Spheres/CDCs
development. 28 We also tested whether the differential expression was correlated with patient age. We evaluated miR-490, miR-31 and miR-133a expression in different biopsies. Our analysis revealed that miR-31, which found more expressed in undifferentiated progenitors, has an inverse correlation with patient age, while the miRNAs overexpressed in differentiated tissues, such as miR-133a and miR-490, increase with the patient's age ( Figure 6B).
To better understand the implications of miRNA expression in the repair processes and in stemness maintenance, we analysed differentially expressed miRNAs with Diana mirPath to find a possible correlation between miRNA and target genes associated with their regulation. In particular, we selected the miRNAs more than twofold pathway. In particular, we found Yap1 clearly differentially expressed in proliferating progenitors vs. biopsies ( Figure 8). As expected, other Hippo pathway members were found to be differentially expressed, such as Smad proteins (Figure 8). Likewise, dishevelled 3 (Dvl3), a scaffolding protein implicated also in Wnt and pathway regulation, 29 is clearly upregulated in CDCs and CSs (Figure 8). Hippo downstream signalling, such as cyclin D1, 30 which promotes proliferation appear upregulated in undifferentiated progenitors (Figure 8).

| DISCUSS ION
The ability of the myocardium to regenerate after AMI was demonstrated more than twenty years ago with the observation in mice models in which delivered bone marrow (BM) progenitors can ameliorate the outcome by generating myocardial cells. 31 Furthermore, mobilized BM in MI patients showed an improvement in function and survival. 32 The use of human cardiac progenitors in the CADUCEUS 11 and PERSEUS 12 trials showed limited benefit in using cardiac progenitors (CDCs) for MI treatment. These studies found scar mass reduction, with increased heart mass and contractive activity. However, the following clinical trials were not satisfying, 33 showing a low expansion efficiency of transplanted cells due to loss of cell anchorage and a short life of the transplanted cells. 34 To circumvent these limitations, scaffolds embedded with cells, or even the administration of vesicles containing proteins, DNA molecules and miRNAs released by cardiovascular progenitor cells were also applied, showing that other factors could contribute to the regeneration of the infarcted heart. 35 In this scenario, miRNAs may represent a promising therapeutic tool because they are easily F I G U R E 5 Heat map analysis of miRNAs more abundant in biopsies, showing the correlation between miRNA expressions obtained from NGS. miRNA expression was hierarchically clustered, showing the association among different cell type: CDC "C" Cardiospheres, "S" Spheres and "B" Biopsies (orange: downregulated, red: upregulated) B4 B1 B5 S5 C1 S1 C4 S4 synthesized and can be proficiently encapsulated in particles/exosomes because of their small size. For this reason, we investigated possible mechanisms from molecular signature results that could determine the proliferative ability of cardiac progenitors. In our study, through microarray and NGS analysis, we aimed to clarify the miRNA expression profile in the CDCs and CSs compared with the differentiated tissue from the same patient.
We found a different miRNA expression pattern in our CDCs/ CSs compared with the biopsies, identifying several miRNAs that can be responsible for the maintenance of an undifferentiated state.
Notably, miR-1 and miR-133 were strongly downregulated in our CDC/ CS cells. From previous studies, both miRNAs have been found to play a role in the differentiation process, as well as to have a specific function in remodelling events 36 ; moreover, they were upregulated during cardiomyocyte differentiation. 14 It has been reported that miR-133a-deficient hearts show increased and aberrant cardiomyocyte proliferation throughout the atria and ventricles; this latter result potentially explains the development of a lethal ventricular septal defect in miR-133 knockout mice, where miR-133a regulates several transcription factors involved in cell cycle control such as cyclin D1, which we found overexpressed in progenitor cells. 28 This cyclin is also regulated by miR-1, 37 and its overexpression promotes cardiomyocyte commitment in human cardiovascular progenitors through the suppression of WNT and FGF signalling pathways. 38 The involvement of miRNAs in the FGF pathway opens an important question on the balance between the proliferative activity in cardiac repair after MI and the fibrotic processes. The FGF pathway has been described as influenced by miR-133, which suppresses atrial remodelling. 39 As here observed, miR-133 is downregulated in undifferentiated proliferation progenitors, and its expression increases with patient age, most likely accounting for the reduced capacity to repair in aged hearts. 40

TA B L E 3 (Continued)
has been shown to have a protective effect in rats against myocardial infarction. 45 Moreover, cardiomyocyte-derived conditioned medium can exert an action, miR-21-dependent, in reducing infarct size in MI induced rats. 45 Likewise, miR-181 has been found to rescue deterioration of cardiac function in post-MI models by suppressing the Aldo-MR pathway. 46 This miRNA has been found to be secreted in exosomes in MSCs models. 47  . Pathway clustering analysis was also performed for the biological pathways using KEGG pathways as in the heat map. miRNAs in downregulated progenitors (B) and upregulated (C) were considered.
Our results underscore a regulation of Hippo pathway genes by various miRNAs identified and differentially expressed among progenitors (CDCs/CSs) vs. bioptic tissues. In particular, we found a specific involvement of the Hippo pathway in progenitors that can be responsible for a higher expression of specific proteins such as Yap1, Smad, Dvl3 and cyclin D1. It has been demonstrated that Yap1 is necessary for regeneration in neonatal injured heart mice. 49,50 Its interaction with β-catenin remarks the connection among Hippo and Wnt signalling in the regulation of cardiomyocyte proliferation and heart size controls. 51 Another association between the two pathways is represented by the interaction Yap/Dvl. 29 Dvl3, which we found upregulated in progenitors, is responsible for heart formation in mice models. 52 Furthermore, Robinow syndrome has been linked to a mutation on the Dvl3 gene 53

ACK N OWLED G EM ENTS
We thank G. Leone, S. Pasqua and D. Galvagno for technical help and for their support. The study was funded by IRCCS Ismett.