PRIMITIVE HEARTS WORK
Primitive mammalian heart transforms from a single tube to a four-chambered muscular organ during a short developmental window. We found that knocking out global microRNA by deleting Dgcr8 microprocessor in Mesp1 cardiovascular progenitor cells lead to the formation of extremely dilated and enlarged heart due to defective cardiomyocyte (CM) differentiation. Transcriptome analysis revealed unusual upregulation of vascular gene expression in Dgcr8 cKO hearts. Single cell RNA sequencing study further confirmed the increase of angiogenesis genes in single Dgcr8 cKO CM. We also performed global microRNA profiling of E9.5 heart for the first time, and identified that miR-541 was transiently highly expressed in E9.5 hearts. Interestingly, introducing miR-541 back into microRNA-free CMs partially rescued their defects, downregulated angiogenesis genes and significantly upregulated cardiac genes. Moreover, miR-541 can target Ctgf and inhibit endothelial function. Our results suggest that microRNAs are required to suppress abnormal angiogenesis gene program to maintain CM differentiation.
Knocking-out key miRNA processing proteins such as DGCR8 has been used to study the functions of global miRNAs (Wang et al., 2007). The two double-stranded RNA binding domains (dsRBDs) of DGCR8 recognize primary miRNAs (pri-miRNAs) (Han et al., 2006), while the conserved C terminus interacts with Drosha to form the microprocessor. The pri-miRNAs were processed by microprocessor into short hairpins, named pre-miRNA, which subsequently exported into cytoplasm, and processed by Dicer into double-stranded mature miRNAs (Wang et al., 2007). Dgcr8 conditional knock-out (cKO) in muscle cells lead to dilated cardiomyopathy and postnatal lethality, indicating that global miRNAs were essential for normal CM function (Rao et al., 2009). We reason that deletion of Dgcr8 at the beginning of heart formation could reveal functions of global miRNAs during this important window of development, and provide a sensitive system to study the role of individual microRNA enriched in the early heart. Many microRNA loss-of-function studies conducted in embryo systems appeared to cause mild or even no phenotype, but careful study revealed increase in variation or reduced robustness of the biological process (Cassidy et al., 2016; Ebert and Sharp, 2012; Kasper et al., 2017). Recent advance in single cell RNA-sequencing technology makes it possible to measure global gene expression in every cell of an organ. This greatly facilitated the identification of the affected cell type by a gene mutation and the associated transcriptome changes (DeLaughter et al., 2016; Lescroart et al., 2018; Li et al., 2016; Liu et al., 2017; Zhou et al., 2016). In this study, we crossed mice carrying floxed Dgcr8 alleles with transgenic mice in which the CRE recombinase was driven by early cardiovascular progenitor cell marker gene Mesp1. Dgcr8 cKO embryos showed severe cardiac defect at E9.5. Global transcriptome and miRNA profiling revealed that without miRNAs, cardiac genes were downregulated but vascular genes were upregulated in the E9.5 hearts. Using single cell RNA-sequencing, we discovered significant upregulation of cell adhesion, glycolysis and angiogenesis genes that may explain the defect in cKO CMs. We identified that miR-541 was highly expressed in E9.5 hearts and was a strong repressor of angiogenesis. MiR-541 can also promote CM differentiation from pluripotent stem cells. These results provided new insights about the development of nascent myocardial cells in vivo and uncovered novel function of miRNA-541, that can potentially be useful to treat blood vessel hyperplasia diseases and pathological cardiac remodeling.
To determine the cellular basis of the heart defect, E9.5 hearts from control and Dgcr8 cKO embryos were dissected and dissociated into very small clusters, and cultured in a chemically defined medium, referred to as in vitro CMs culture medium (IVCC medium) supplemented with bFGF (4 ng/mL). Both control and Dgcr8 cKO CMs attached well to the culture plate and kept beating after 24 h. Then they were fixed and stained for Troponin T (cTnT). Compared to the control CMs, Dgcr8 cKO cells had poorly organized sarcomere structure (Fig. 1G). Transmission electron microscopy (TEM) images showed that Dgcr8 null CMs contained only nascent myofibrils, while no clear Z-line could be observed. In contrast, mature Z lines and I bands were evident in control CMs (Fig. 1H). We also treated CMs with a Ca2+-sensitive dye, Fluo-4 AM to study the calcium handling ability of cKO cells. The rising slope and time to peak of Fluo-4 AM fluorescence were significantly different between the control and Dgcr8 cKO CMs (Fig. 1I). Strong, rhythmic calcium influx was observed in control group (Fig. 1I and Movie S3). In contrast, in Dgcr8 cKO group, the calcium transient had irregular rhythm, slower excitation cycle and longer decay time, suggesting decreased Ca2+ clearance from the cytosol and contractility (Fig. 1I, 1J and Movie S4). Finally, we analyzed proliferation ability of Dgcr8 cKO CMs. Dissociated small cell clusters were cultured in IVCC medium with bFGF for 7 days. Cells were fixed and immunostained for cTnT on day 1, 3 and 7. CM numbers of each group were counted. By day 3 and day 7, the CM number in control group increased approximately 2.54 and 5.06 folds compared to day 1, respectively. While in Dgcr8 cKO group, the cell number stayed the same after 7 days (Fig. 1K).
In the control hearts, 6,207 genes were downregulated at least 1.5-fold from E8.5 to E9.5. Among them, 527 genes were upregulated more than 1.5-fold in E9.5 Dgcr8 cKO hearts. These genes are likely to be repressed by miRNAs (Fig. 2F). The top GO categories associated with these 527 genes including angiogenesis and erythrocyte development, which were related to vascular development (Fig. 2G and Table S2C). The expression of several well-known positive regulators of angiogenesis such as Egfl7, Sox17, Vegfα and Ctgf, were significantly upregulated in E9.5 Dgcr8 cKO hearts, indicating that they were highly likely to be targets of miRNAs at this stage (Fig. 2H). Sum together, the abnormal transcriptome of E9.5 Dgcr8 cKO hearts was in accordance with the phenotype of the severely dilated and dysfunctional mutant heart.
The upregulation of angiogenesis genes in E9.5 Dgcr8 cKO hearts may attribute to two reasons: the derepression of vascular related genes in Dgcr8 cKO CMs, or the increase of EC proportion in the cKO heart. To distinguish between these two possibilities, we performed single cell RNA-seq of cKO and control hearts. We used homozygous Dgcr8loxp/loxp; ROSA26mT/mG/+, without any Cre alleles as control. The tdTomato fluorescence of the control group could be a marker to estimate the single cell sequencing quality in the following step, and the phenotypes of the control mice were normal from embryonic stages to adult.
The ventricular part of control and cKO E9.5 hearts were dissected and digested into single cell suspension. 96 single cells of each genotype were manually picked, followed by RNA-seq library construction and high-throughput sequencing (Fig. 3A). We obtained 76 high-quality single cell libraries from control and cKO ventricles respectively and there was no batch effect between two sequencing experiments (Fig. S3A, S3C and Table S3A). To determine whether the cells were progeny of Mesp1 in the cKO group, we included GFP and tdTomato sequences in the reference genome for mapping and expression level quantification (Showell and Conlon, 2007). Among 76 cKO single cells, high GFP and low tdTomato cells were destined to be Mesp1 progeny cells, while 7 low GFP and high tdTomato cells were considered to be control cells that had not expressed Mesp1-Cre. In 76 control single cells, 2 high GFP and low tdTomato cells were considered not to be control cells. These 9 cells with abnormal reporter gene expression were excluded from the following comparative analysis (Fig. S3B and Table S3A).
To discover miRNAs might be responsible for the phenotype of Dgcr8 cKO hearts, we performed miRNA sequencing of E9.5 wildtype (WT) hearts. The top 20 highest expressed miRNAs were shown in the heatmap graph (Fig. 4A), and their expression counts were listed in Table S4. Our results showed that miR-1 was the most abundant miRNA (20.45%) in E9.5 hearts, although not quite as high as its content in the adult heart (nearly 40%) (Rao et al., 2009). Many of the top 20 miRNAs were also reported to be enriched in the adult heart, such as miR-378a, miR-26a, miR-133a and miR-30 family. MiR-126, a miRNA highly expressed by ECs (Fish et al., 2008; Wang et al., 2008a), was also detected, possibly due to the ECs in the embryonic heart. We predicted targets of the top 20 miRNAs using TargetScan (Lewis et al., 2003). Among the 10,435 putative targets, 235 genes decreased more than 1.5 folds from E8.5 to E9.5 in control hearts but upregulated at least 1.5 folds in E9.5 Dgcr8 cKO heart (Fig. 4B). GO analysis indicated that these genes were predominantly involved in angiogenesis and blood vessel morphogenesis (Fig. 4C and Table S2D). We speculate that the upregulation of this group of genes in E9.5 Dgcr8 cKO hearts was due to lack of miRNAs inhibiting endothelial program, and subsequently contributed to the extremely dilated heart phenotype.
Global miRNA profiling in E9.5 mouse embryonic heart. (A) Heatmap showing the top 20 most highly expressed miRNAs in E9.5 WT embryonic heart. Values represent log2 of counts for each sample. (B) Venn diagram depicting the number of putative target genes of the top 20 miRNAs predicted by TargetScan (blue), genes downregulated from E8.5 to E9.5 in the control hearts (red) and genes upregulated in E9.5 Dgcr8 cKO heart at E9.5 (green), and overlapped genes between these groups. (C) GO analysis of 235 overlapped genes among three groups in (B). (D) Q-PCR confirmation of miRNA-1 and miRNA-541 expression in embryonic heart at different stages and in mESCs. Data represent mean s.e.m. from three biological repeats. (E) Venn diagram depicting number of putative target genes of miRNA-541 predicted by TargetScan that overlapped with genes upregulated in E9.5 Dgcr8 cKO hearts. (F) GO analysis of 126 overlapped genes in (E). (G) Cytoscape network generated from 34 angiogenesis genes that upregulated in E9.5 Dgcr8 cKO heart, including 8 putative miR-541 targets. Node color represents two types of genes, yellow and grey represents miR-541 targets and non-miR-541 targets respectively. Edge is weighted by combined interaction scores, high values correspond to red color whereas low values to green color 041b061a72