The atlas of ACE2 expression in fetal and adult human hearts reveals the potential mechanism of heart-injured patients infected with SARS-CoV-2.

Numerous studies have shown that severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) can infect host cells through binding to angiotensin I converting enzyme 2 (ACE2) expressing in various tissues and organs. In this study, we deeply analyzed the single-cell expression profiles of ACE2 in fetal and adult human hearts to explore the potential mechanism of SARS-CoV-2 harming the heart. The molecular docking software was used to simulate the binding of SARS-CoV-2 and its variant spike protein with ACE2. The genes closely related to ACE2 in renin-angiotensin system (RAS) were identified by constructing a protein-protein interaction network. Through the analysis of single-cell transcription profiles at different stages of human embryos, we found that the expression level of ACE2 in ventricular myocytes was increased with embryonic development. The results of single-cell sequencing analysis showed that the expression of ACE2 in ventricular myocytes was upregulated in heart failure induced by dilated cardiomyopathy compared with normal hearts. The upregulation of ACE2 increases the risk of infection with SARS-CoV-2 in fetal and adult human hearts. We also further confirmed the expression of ACE2 and ACE2-related genes in normal and SARS-CoV-2-infected human pluripotent stem cell-derived cardiomyocytes. In addition, the pathway analysis revealed that ACE2 may regulate the differently expressed genes in heart failure through calcium signaling pathway and Wnt signaling pathway.

INTRODUCTION

Severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) is a novel coronavirus, causing the coronavirus disease 2019 (COVID-19). COVID-19 is a new type of infectious pneumonia, and the ongoing outbreak of COVID-19 has caused a global pandemic. The patients suffering from COVID-19 generally showed fever, fatigue, dry cough, and dyspnea, and some patients eventually developed into respiratory failure to cause death (1). As the pandemic of COVID-19 continues, some variants have been discovered, and infections with these variants tend to increase transmissibility, show more severe clinical manifestations, and even resist existing vaccines (2). A clinical study showed that ∼50% of diagnosed patients have chronic underlying diseases, of which cardiovascular and cerebrovascular diseases are at the top (40%) (3). Furthermore, the elderly patients with coronary heart disease or hypertension were more susceptible to COVID-19 and had a much more aggressive illness causing earlier mortality (4). Patients with underlying heart diseases may be more susceptible to COVID-19 and more likely to develop severe diseases and even death. However, the detailed mechanisms leading to this situation need to be understood.

Previous study showed that SARS-CoV-2 and SARS-CoV shared 76.47% homologous sequence in S-protein (5), and SARS-CoV-2 could bind to the same host cellular receptor angiotensin-converting enzyme 2 (ACE2) like SARS-CoV (6). Previous studies showed that SARS-CoV-2 could bind to the same host cellular receptor angiotensin-converting enzyme 2 (ACE2) using its S-protein (7). ACE2 is an important negative regulator in renin-angiotensin system (RAS), and its ACE2-ANG-(1–7)-Mas receptor axis is a classic cardiac protection pathway (8). ACE2 has a broad distribution in various tissues such as small intestine, heart, kidney, thyroid, and testis (9). The distribution of ACE2 is one of the most important determinants in SARS-CoV-2 entering into the human body (10, 11). Chen et al. (11) reported that the pericytes of adult human hearts with highly expressed ACE2 showed an intrinsic susceptibility to SARS-CoV-2 infection, and the basic heart failure patients with increasing ACE2 expression might have high possibility of heart attack and develop into the severe situation after SARS-CoV-2 infection. However, the reasons of increasing ACE2 expression in heart failure patients caused by different reasons remains unclear. Moreover, recent studies have reported that fetuses also were seemed to be at risk for SARS-CoV-2 (12–14), and the expression level of ACE2 in different stages of heart development is still unknown. It is known that some cytokines are involved in the process of ACE2-mediating biological effects. Studies have reported that after SARS-CoV-2 throughout binds to ACE2, several host proteases including transmembrane protease serine protease-2 (TMPRSS2) (15), FURIN, and cathepsins, may play an important role in the host infection process by cleavage of S protein. Therefore, the expression of cytokines closely related to ACE2 in the heart may also affect SARS-CoV-2 infection and prognosis after infection, which needs to be clarified (7, 16).

In this study, we conducted a comparative study on the binding of SARS-CoV-2 and its variants to ACE2 by molecular docking. We analyzed ACE2 expressions in hearts of coronary atherosclerotic disease (CAD)- and dilated cardiomyopathy (DCM)-induced heart failure patients and fetuses using single-cell RNA (scRNA)-seq data. We also further analyzed the RNA-seq data of human pluripotent stem cells-derived (hPSC) cardiac muscle cells infected by SARS-CoV-2. Through comparing the different reasons induced in heart failure samples as well as monitoring the dynamic changes of critical proteins and factors related to ACE2 expression in adult, embryonic hearts, and human-induced pluripotent stem cell (hiPSC)-derived cardiomyocyte, we constructed the atlas of ACE2 expressions in fetal and adult human hearts, and concluded the potential mechanism of heart infection by SARS-CoV-2.

MATERIALS AND METHODS

Molecular Docking

Acquisition of target primary amino acid sequences.

We obtained the published sequences of SARS-CoV-2 spike with 1,281 amino acids (https://www.ncbi.nlm.nih.gov/protein/6VXX_A). SARS-CoV-2 spike protein and the variants were acquired from the PDB (Protein Data Bank, https://www1.rcsb.org/), including SARS-CoV-2 (PDB: 6M17), B.1.1.7 (PDB: 7LWU), B.1.351 (PDB: 7LYN), and B.1.1.28 (PDB: 7LWW). The structure of human ACE2 was also acquired from PDB database (PDB: 1R42).

Homology modeling.

We performed template mode on SWISS MODEL server (https://swissmodel.expasy.org/) to predict spatial structures of SARS-CoV-2 spike protein of B.1.617.1 and B.1.617.2 variants based on target-template alignment (17). We utilized the RBD domain of SARS-CoV-2 (PDB: 7NKT) as a template.

Reading the docking complex.

The fetched docking complex of RBD was modified using PyMOL software (18). The protein docking is done on GRAMM-X Protein-Protein Docking Web Server v.1.2.0 (http://vakser.compbio.ku.edu/resources/gramm/grammx). Then, we used PDBePISA (https://www.ebi.ac.uk/pdbe/pisa/), an interactive tool for the exploration of macromolecular interfaces, to analyze the docking results. Finally, we used PyMOL software to visualize the docking results.

Protein-Protein Interaction Network

The Search Tool for the Retrieval Interacting Genes (STRING, https://string-db.org/) was used to harvest protein-protein interaction (PPI) network (19).

Acquisition of scRNA-Seq Data

We downloaded and analyzed the published scRNA-seq datasets from heart tissues at different embryonic stages and disease stages, including normal human embryonic hearts, normal adult hearts, coronary atherosclerotic hearts, and dilated cardiomyopathy hearts (20, 21). The scRNA-seq data of human embryos were acquired from the Gene Expression Omnibus (GEO, http://www.ncbi.nlm.nih.gov/geo/) database (GSE106118). To further delineate the development time points, the normal human cardiac embryos groups were further categorized into three groups: the early stage (5–7 wk), the mid-stage (9–17 wk), and the late stage (20–25 wk). The normal adult human data were obtained from GSE109816. The coronary atherosclerotic disease (CAD) data and dilated cardiomyopathy (DCM) data were downloaded from GSE121893.

Analysis of scRNA-Seq Data

We used Seurat (v.2.3.4) (22) in R programming environment to conduct routine pipeline for single-cell data analysis.

Seurat object creating.

We used the following criteria to create Seurat objects for each individual cell: minimal expression of three gene per cell and minimal expression of 500 genes per cell.

Cell filtering.

We first filtered out droplets and low-quality cells based on quality control criteria with nFeature_RNA < 9,000 for normal human cardiac embryos, nFeature_RNA < 10,000 for normal adult human, and nFeature_RNA < 7,500 for the coronary atherosclerotic hearts and dilated cardiomyopathy hearts. Meanwhile, we set metric with percentage of mitochondrial genes <72% (means ± SD) to exclude the influence of the variation in mitochondrial genes (gene symbols starting with MT-). Then, we used the LogNormalize method with the scale factors of 10,000 to obtain normalized data.

Linear dimension reduction PCA.

Variable genes among cells were found using vst method. We converted the average expression value of each gene into 0, and the variance between cells to 1 was performed using ScaleData function. Then, the top 1,500 variable genes were selected to perform principal component analysis (PCA) (23). Here, JackStrawPlot function was used to compare the distribution of P values of principle component (PC) with parameters num.replicate = 1,000 for normal human cardiac embryos and 100 for the others (normal adult hearts, coronary atherosclerotic hearts, and dilated cardiomyopathy hearts). Moreover, we assessed the number of principle components by the elbow plot. We found that the first 20 PCs (18 PCs in CAD) could explain majority of correlations for single-cell sequencing data. Therefore they were used for subsequent cell clustering.

Single-cell clustering.

We performed cell clustering based on FindClusters function such that the perplexity was 100, the resolution of the normal human heart was set to 0.8 (normal human heart embryo was 0.3), and 1.0 for both CAD and DCM. Furthermore, we projected the results by two-dimensional t-distributed stochastic neighbor embedding (t-SNE) plot (24).

Cell annotation.

We used FindAllMarkers function of Seurat with the parameters min.pct = 0.2, min.diff.pct = 0.1, return.thresh = 0.05, and logFC.threshold = 1.0 to screen the genes that are significantly expressed in each cluster. The genes were filtered with logFC > 1 and P < 0.05. Next, cell types were annotated under the guidance of the classical cell-specific marker genes. Finally, t-SNE plots, scatter plot, and violin plots were generated using Seurat package in R. The pie chart was used to describe the proportion of cells that have been captured.

Differential expression analysis.

We further compared the expression of cardiomyocytes in different embryonic stages and different diseases with adults. We performed the expression matrix in terms of loge(TPM + 1) to analyze the differential expression in R using the limma package.

Analysis of RNA-Seq Data

We obtained the published expression profiling of human in vitro iPSCs derived cardiomyocyte with SARS-CoV-2 infection from GEO database with accession number GSE162113. Raw counts were normalized in terms of CPM (counts per million) with package edgeR (Version 3.28.0), and then transformed log2(CPM + 1). Differential expression analysis was conducted with limma package in R.

GO and KEGG Pathway Enrichment Analysis

The enrichment analysis of GO and KEGG pathways is implemented in the clusterProfiler package in R, with genes that were significantly differentially expressed (logFC > 5 and adjust P value <0.05) on the first day and the second day after SARS-CoV-2 infection.

Pathway Enrichment Analysis

We analyzed marker genes of CAD and DCM ventricular myocytes with logFC > 1 and adjust P value <0.05. David database (https://david.ncifcrf.gov), clusterProfiler package in R, and consulted literatures were used to screen and integrate the results of pathway enrichment analysis.

RESULTS

Molecular Docking and Protein-Protein Interaction Network between ACE2 and SARS-CoV-2

Based on the previous biochemical interaction studies and crystal structure analysis, SARS-CoV-2 spike protein has a strong binding affinity to human ACE2, and it directly binds to ACE2 located on the surface of host cell to promote virus entry and replication. To further understand the role of human ACE2 in the infection process of the host cells with SARS-CoV-2, especially in the heart cells, we constructed a three-dimensional (3-D) molecular model of the combination of SARS-CoV-2 spike protein and its variants bound to human ACE2 (Fig. 1, Supplemental Fig. S1; all Supplemental materials are available at https://doi.org/10.6084/m9.figshare.18469907). The results of molecular docking showed that the amino acids TYR-369, SER-373, and PHE-374 in the SARS-CoV-2 receptor binding domain bind to GLN-736 on ACE2 by hydrogen bond binding force. There is double hydrogen bonding between ASN-720 and ASP-428, and only one hydrogen bond is detected between ASN-735 and PHE-377 (Fig. 1). The ASN-343, SER-530, GLU-340, and VAL-362 in the variant 501Y.V1 binding domain are bound by hydrogen bonds with TYR-158, ARG-705, TYR-255, and THR-698 on ACE2, respectively (Fig. 1, Supplementary Fig. S1B). The amino acids LEU-518 and ALA-522 in variants 501Y.V2 and 501Y.V3 are both hydrogen bonded to SER-626 on ACE2 (Fig. 1, and D, Supplemental Fig. S1, C and D). The combination of variants B.1.167.1 and B.1.167.2 with ACE2 is dominated by hydrogen bonds, of which ASN-440, SER-373, and CYS-336 on B.1.167.1 are combined with ARG-768, ILE-759, VAL-755, and LEU-743 on ACE2 (Fig. 1, Supplementary Fig. S1E). CYS-336, ASN-439, ASN-437, and SER-373 located on B.1.167.2 are combined with THR730, SER-740, ASN-735, and PRO-733 on ACE2 (Fig. 1, Supplemental Fig. S1F). Thus, we confirmed that SARS-CoV-2 and its variants may recognize human ACE2 as a receptor to infect the host.

Figure 1.

Molecular docking and protein-protein interaction network. A–F: the molecular docking model of angiotensin I converting enzyme 2 (ACE2) and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and its variant spike protein. The blue domain represents ACE2 protein. The light red domain represents the spike protein of SARS-CoV-2 or its variants. The content in the box is a closeup of the binding sites of the two proteins, the part marked in dark red is the protein binding amino acid site, the solid chain represents the protein binding peptide bond, and the dashed line represents hydrogen bonding. G: protein-protein interaction network of ACE2 and related proteins. It shows both known and predicted interactions. Colored nodes are the first shell of interactors, white nodes are the second shell. The blue line interaction is determined from curated database, and the purple line connection represents the experimentally verified results.

We found several proteins closely interacted with ACE2 through the STRING: functional protein association networks and constructed a protein-protein interaction network (Fig. 1). In addition to ACE2, we also focused five important genes interacting with ACE2 that are closely related to cardiovascular disease, including angiotensin-converting enzyme (ACE), angiotensinogen (AGT), type-1 angiotensin II receptor (AGTR1), type-2 angiotensin II receptor (AGTR2), and (pro)renin receptor (ATP6AP2). Research shows AGT gene is related to heart failure, myocardial infarction, cardiomyopathy and atrial fibrillation, hypertension, and high risk of atherosclerosis (25). The studies found that increasing prorenin can induce cardiac remodeling with loss of cardiac function through angiotensin-independent pathways (26). These genes may play the crucial role in myocardial damage induced by SARS-CoV-2 infection in patients with underlying heart disease.

The ACE2 Expression in Different Cell Types of Human Heart during Three Embryonic Stages

To understand the effects of embryonic development on the expression of ACE2 in human heart tissues, we processed single-cell transcriptional profiles of human fetal heart tissues from 5 wk to 25 wk obtained from left atrium, right atrium, left ventricle, and right ventricle. Cells were divided into three stages for subsequent analysis, including the early stage (5–7 wk), the mid-stage (9–17 wk), and the late stage (20–25 wk). We conducted cell filtering and gene filtering with Seurat R package to ensure the quality of single-cell transcriptional profiles. After quality control and removal of mitochondrial genes, we obtained 1,656 cells with each single cell detecting 19,525 genes in wk 5–7. For wk 9–17 and wk 20–25, 1,266 cells with 19,379 genes in each cell, and 1,876 cells with 20,768 genes were retained, respectively.

We performed t-SNE clustering analysis of filtered and normalized cells in three stages, then we totally identified 11 clusters in wk 5–7, 9 clusters in wk 9–17, and 13 clusters in wk 20–25, respectively. Subsequently, we annotated each cluster through the known specific marker genes of each cell type. Thirty-three clusters from three stages were divided into main six cell types (Fig. 2). We identified the atrial cardiomyocyte (CM-A) with the high expression levels of MYH6, MYL4, and MYL7. The ventricular cardiomyocyte (CM-V) was distinguished by the highly expressed MYH7, MYL2, and MYL3. We found several clusters that specifically expressed PECAM1, IFI27, and VWF, and these genes indicated that these clusters were endothelial cells (EC). Extracellular matrix genes were found in several clusters such as DCN, FBLN1, COL1A1, and COL1A2 that are well-known markers of fibroblasts (FB). Another main cell type was smooth muscle cells (SMC) with specific expression of ACTA2, MYH11, TAGLN, and CNN1. The exclusively expressed CD163, CCL4, and CXCL8 are hallmarks of macrophages (MP). Other cell types including T/B cell and epicardial cell (EP) were identified by CD3D and UPK3B, respectively.

Figure 2.

Overview of human fetal heart single-cell ACE2 expression atlases. A: visualization of t-distributed stochastic neighbor embedding (t-SNE) clustering at different embryonic development stages (5–7 wk, 9–17 wk, and 20–25 wk). Each dot represents a single cell. Different colors cluster represents different cell types, which are obtained by annotation of known marker genes. B: the proportion of different cell types in each embryonic development stages. C: t-SNE visualization of ACE2 expression in different stages. The darker the dot, the higher the expression of ACE2. D: the violin diagram shows the expression level of ACE2 in various cell types and the distribution of single cells expressing ACE2 at different embryonic stages. Expression level represents log (TPM). Unless otherwise specified, all expression levels in the article refer to log (TPM). CM-V, ventricular cardiomyocytes; CM-A, atrial cardiomyocytes; EC, endothelial cells; EP, epicardial cells; FB, fibroblasts; MP, macrophages; SMC, smooth muscle cells.

Since the total number of cells was different in three stages, we calculated the proportion of different types of cells in each stage (Fig. 2). We found that with the development of the embryonic heart, the proportion of CM-V and CM-A decreased significantly. In contrast, the proportions of FB and immune-related cells (such as MP, T cell, and B cell) were obviously increased. Interestingly, SMC accounted for the highest percentage in wk 20–25 (the late stage) compared with other types of cells. These results indicated that noncardiomyocytes played a growing role during embryonic development.

We further examined the expression of ACE2 in different embryonic stages, and found that the expression level of ACE2 was relatively low in the early and middle stages, but it was significantly increased in the late period of embryonic development compared with the previous two stages (Fig. 2). In addition, although the expression of ACE2 during embryonic period is not apparently high, we can still notice that ACE2 is mainly expressed in CM-V and CM-A (Fig. 2). The expression level of ACE2 in fibroblasts and immune cells is upregulated in the late stage of embryonic development (20–25 wk). Moreover, with the development of the embryo, the expression of ACE2 gradually increased, and the most obvious increase emerged in wk 20–25. It is noticeable that the expression of ACE2 in CM-V was increased the most significantly, although noncardiomyocytes were becoming more and more important during embryonic development.

We also analyzed the expression levels of several important genes interacting with ACE2 in renin-angiotensin system (Fig. 3, and Fig. 4). The expression level of ATP6AP2 is visibly high in various cell types in different developmental stages, and the expression levels of several other genes are relatively lower, including ACE, AGT, AGTR1, and AGTR2 (Fig. 3, –). However, the expression of these genes changed distinctly in different cell types with embryonic development (Fig. 3, –). The expressions of AGT and AGTR2 in fibroblasts were increased significantly. ACE was highly expressed in endothelial cells and upregulated with embryonic development. AGTR1 was mainly expressed in fibroblasts, and its expression was increased in the middle and late stages of embryo development. In addition, the expression of these five genes were increased significantly in ventricular myocytes with embryonic development.

Figure 3.

The expression of interacting genes in the single-cell map of human embryos. A–C: t-distributed stochastic neighbor embedding (t-SNE) visualization of gene expressions in different stages. The darker the dot, the higher the expression of genes. D–F: the violin diagram shows the expression level of interactors in various cell types and the distribution of single cells at different embryonic stages. Expression level: log (TPM). G–I: gene expression in different cells at different stages of embryonic development.

Figure 4.

Differential expression of angiotensin I converting enzyme 2 (ACE2) and its related genes in human fetal heart and diseased adult human heart. *Significant difference in gene expression upregulation or downregulation (P value is less than 0.05, P < 0.05). The closer to red, the more obvious the upregulation of gene expression. The closer to blue, the more obvious the gene expression is downregulated. The numerical value represents the fold change of gene expression (Log2FC), the positive value means upregulation, and the negative value means downregulation.

We also found that TMPRSS2 is less expressed during embryonic development, but it is somewhat upregulated in fibroblasts at later stages of development. CTSL, FURIN, ADAM17, GJA5, and GJA1 are mainly expressed in CM-V, CM-A, endothelial cells, fibroblasts, and immune cells. As the embryo develops, the expression of GJA5 in CM-V gradually decreases, and the expression of GJA1 in CM-V gradually increases (Supplementary Fig. S2).

The ACE2 Expression Profiles in Different Cell Types of Normal and Diseased Adult Human Heart

To further understand the different expressions of ACE2 between normal and injured heart in adults, we analyzed the single-cell sequencing data of heart tissues from healthy donors (normal), patients with coronary atherosclerotic disease [CAD hear failure (HF)], and patients with dilated cardiomyopathy (DCM HF). Similar to the processing described previously, we performed strict quality control on the expression data of the single-cell sequencing, and removed the low-expression cells and genes to ensure the smooth progress of subsequent analysis. Therefore, we left 8,140 cells (43,861 genes) in the analysis of the normal adult heart group. In addition, in the single-cell transcriptome analysis of CAD HF and DCM HF groups, we retained 1,432 cells (21,406 genes) and 2,750 cells (24,304 genes), respectively. Through t-SNE clustering in Seurat R package and annotating each cluster with the known marker genes, the corresponding cell types were obtained. The genes used to judge the cell types are the same as the marker genes described previously in human fetal hearts. Therefore, we collected seven main cell groups named CM-V (ventricular cardiomyocyte), CM-A (atrial cardiomyocyte), EC (endothelial cell), FB (fibroblast), MP (macrophage), T cell, and SMC (smooth muscle cell) as shown in Fig. 5

Figure 5.

Single-cell ACE2 expression profiles of normal and diseased adult human heart. A: Visualization of t-distributed stochastic neighbor embedding (t-SNE) clustering. B: the proportion of different cell types in each group. C: t-SNE visualization of ACE2 expression in different groups. D: the violin diagram shows the expression level of ACE2 in various cell types and the distribution of single cells. Expression level: log (TPM). CM-V, ventricular cardiomyocytes; CM-A, atrial cardiomyocytes; EC, endothelial cells; FB, fibroblasts; MP, macrophages; SMC, smooth muscle cells.

We also calculated the proportion of each type of cell in different groups. Among the seven cell types, CM-V, CM-A, and EC accounted for the largest proportion. Among these three types of cells, the proportion of EC in the disease groups was significantly increased compared with the normal group, and the corresponding proportion of cardiomyocytes was reduced (Fig. 5). However, the decrease in the proportion of myocardial cells mainly reflected in the decrease in CM-A, and the proportion of CM-V was not changed significantly.

In normal heart, ACE2 is mainly expressed in CM-V, CM-A, SMC, FB, and EC. Similarly, in the heart tissues of CAD HF, ACE2 is highly expressed in CM-V, CM-A, EC, and SMC, and there is no significant difference in the expression level of ACE2 in cardiomyocytes (CM-V and CM-A) compared with normal hearts. However, in DCM HF tissues, we found that the expression of ACE2 was obviously increased in cardiomyocytes (Fig. 5). In addition, as shown on the violin chart (Fig. 5), it can be concluded that ACE2 expression is mainly concentrated in cardiomyocytes both in normal and diseased heart tissues. And the obtained results revealed that the expression of ACE2 in CM-V exceeds that in CM-A in diseased heart tissues.

Subsequently, we examined the expression levels of ACE, AGT, AGTR1, and ATP6AP2, which are closely related to ACE2 in cardiovascular disease (Fig. 4). The cluster (Fig. 6, –) and violin diagram (Fig. 6, –F) showed that the genes were highly expressed in each cell type, and the bubble diagram (Fig. 6, –) showed the differential expression genes in the disease group and normal group. ACE is highly expressed in EC, and the expression level in the disease group is higher than that in the normal heart group, especially in CAD HF. AGT is mainly expressed in CM-A, SMC, FB, and MP in normal heart tissue, and richly expressed in SMC and CM-A in CAD HF, however, it is abundantly expressed in various cell types (including CM-V, CM-A, EC, FB, MP, T cell, and SMC) in DCM HF. ATP6AP2 is abundantly expressed in CM-V, CM-A, EC, FB, and MP, and it can also be found that the expression of ATP6AP2 in the disease group is lower than that in the normal group. AGTR1 is mainly expressed in CM-A in normal tissues and CAD HF, while it is mainly expressed in CM-V in DCM HF.

Figure 6.

The expression signature of ACE2-associated genes in normal and diseased adult human hearts. A–C: t-distributed stochastic neighbor embedding (t-SNE) visualization of associated genes in different groups. D–F: the violin diagram shows the expression level of genes in various cell types and the distribution of single cells. Expression level: log (TPM). G–I: gene expression in different cells with different diseases. J: enrichment results of differently expressed genes in ventricular cardiomyocytes (CM-V) in coronary atherosclerotic disease (CAD) and dilated cardiomyopathy (DCM) HF.

We further found that TMPRSS2 is only expressed in normal heart tissues, but not in the disease group. ADAM17 is expressed in both normal and disease groups. CTSL and FURIN are mainly expressed in CM-A, CM-V, EC, FB, and SMC, and the expression level in the disease group is decreased in CM-A and CM-V, especially in CAD HF. Compared with the normal group, the expression of GJA5 in CAD HF decreased significantly in CM-A, EC, FB, and SMC. The GJA5 expression of CM-V increased in DCM HF. GJA1 was expressed in CM-A, CM-V, EC, FB, and SMC in both the normal group and the disease group, and decreased in CAD HF and DCM HF groups (Supplemental Fig. S3).

KEGG Pathway Analysis of Differentially Expressed Genes in Adult Human Diseased Hearts

We performed pathway enrichment analysis on differently expressed genes of CM-V in CAD HF and DCM HF using DAVID (http://david.abcc.ncifcrf.gov/) and clusterProfiler package (v.3.14.3) in R. The results showed that in addition to the DCM and hypertrophic cardiomyopathy (HCM) signaling pathway leading to heart disease, the calcium signaling pathway and cGMP-PKG signaling pathway were also significantly enriched (Fig. 6). We checked six genes (ACE2, ACE, AGT, AGTR1, AGTR2, and ATP6AP2) through WikiPathways (https://www.wikipathways.org/index.php/WikiPathways) to obtain these genes participating in pathways, and we integrated the results into a simple mechanism diagram (Fig. 7). In the activated RAS system, after ANG II binding to AGTR1 receptor, the calcium signal changes (27–31), which affects the process of myocardial fibrosis and myocardial hypertrophy. When AGTR2 receptor is activated by ANG II, the protective mechanism is initiated by inhibiting myocardial fibrosis (32). In addition, we also noticed that the upregulation of ANG II led to the downregulation of connexin 40 and connexin 43 (33) (encoded by GJA5 and GJA1, respectively), resulting in reduced cardiomyocyte coupling and calcium electrical signal blockade (34–36), which caused abnormal myocardial electrophysiological function. We used WikiPathways to query the related pathways, and the results showed that GJA1 and GJA5 were involved in the regulation of the Wnt signaling pathway (37) and affected the calcium signaling. Similarly, we noticed that ATP6AP2 is also involved in the regulation of Wnt signaling pathway (38).

Figure 7.

The diagram of the potential mechanism that angiotensin I converting enzyme 2 (ACE2) participates in the regulation of heart failure and the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) bind to ACE2.

According to the results of pathway analysis, we found that GJA1 and GJA5 are involved in the regulation of the heart disease through RAS system. Moreover, we speculate that the expression levels of GJA1 and GJA5 are regulated by ACE2. In single-cell sequencing profiles, we detected the expression levels of GJA1 and GJA5 in different cells and different embryonic development stages (Fig. 3, –). At the same time, we also compared the expression of GJA1 and GJA5 in the heart tissues of different diseases (Fig. 6, –). GJA1 was mainly expressed in CM-V, and the expression level was increased with embryonic development, especially in wk 20–25. The expressions of GJA1 in two disease groups were higher than that in normal group, and the expression in DCM was significantly upregulated. In contrast, GJA5 is mainly expressed in CM-A, and the expression is downregulated with embryo development. The expression of GJA5 in the disease group was lower than that in the normal group, and the expression in CAD HF was significantly downregulated.

The ACE2 Expression Profiles in Normal and SARS-CoV-2-Infected Human Pluripotent Stem Cell-Derived Cardiomyocytes

We used the limma package in R to analyze the RNA-seq data of normal cardiomyocytes induced by human pluripotent stem cells (iPSCs) in vitro and SARS-CoV-2-infected iPSCs derived cardiomyocytes within 48 h. The results showed (Fig. 8, Supplementary Fig. S4, Supplementary Table S1) that within 48 h, compared with the normal group, the expression of ACE2 in the infected group was significantly downregulated on the first day, and the second day was almost the same as the first day. The expression of AGT and AGTR1 were significantly upregulated on the second day compared with the first day both in the normal group and the infected group, and the expression of the infected group was also significantly upregulated compared with the normal group within 2 days, especially the expression on the second day. The expression of ATP6AP2 remained stable within 2 days, but the expression of the infected group was significantly downregulated compared with the normal group, but the amplitude was not large. The expression of ADAM17 was slightly downregulated within 48 h, and the infected group was significantly downregulated compared with normal cardiomyocytes in the first 24 h. The expression of GJA1 was upregulated within 48 h. Compared with the normal group, the expression of GJA1 was significantly upregulated. The increase was relatively small in the first 24 h, and the expression of the infected group was greatly upregulated in the next 24 h. The expression of GJA5 was significantly upregulated in the normal group within 2 days, whereas the expression in the infected group was significantly downregulated, but the infected group was significantly downregulated compared with the normal group for two consecutive days. The expressions of CTSL were downregulated within 2 days, but they were not obvious. Compared with the normal group, FURIN and TMPRSS2 were significantly upregulated in the infected group, and CTSL was significantly downregulated.

Figure 8.

The expression signature of angiotensin I converting enzyme 2 (ACE2) and its related genes in normal and severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2)-infected cardiomyocytes induced by human pluripotent stem cells. A: the box diagram shows the expression levels of ACE2 and ACE2-related genes in normal and SARS-CoV-2 infected cardiomyocytes induced by human pluripotent stem cells in vitro within 48 h. Expression level:log (TPM + 1). *Expression of genes in the infected group is significantly different compared with the normal group. B and C: GO and KEGG pathway enrichment analysis results of differently expressed genes.

We further performed GO and KEGG analysis on differently expressed genes obtained from RNA-seq data. The GO results showed (Fig. 8) that most of the differential genes were significantly related to the regulation of cAMP, organophosphorus, purine compounds, and phosphorylation and protein phosphorylation. The results of KEGG showed (Fig. 8) that osteoclast differentiation, receptor activation, viral infection, and MAPK signaling pathways are abundantly expressed and accounted for a large percentage of genes, especially osteoclast differentiation signaling pathways.

We believed that SARS-CoV-2 attaches to the ACE2 receptor through the spike protein to invade the cell, leading to the internalization of the virus particle (Fig. 7). ACE2 can exist in cellular (membrane-bound) or circulating (soluble) form (sACE2). Once SARS-CoV-2 enters the host cell, ACE2 may be downregulated. The downregulation of ACE2 is achieved by the shedding mediated by ADAM17, whereby the membrane-bound ACE2 shedding is sACE2. TMPRSS2 initiates the S protein for membrane fusion and cleavage of ACE2 receptors (7). The protein encoded by CTSL cleaves the S1 subunit of the SARS-CoV-2 spike protein, which is necessary for the virus to enter the cell. The spike protein of SARS-CoV-2 is also cleaved by FURIN-encoded protease. We speculated that the aforementioned enzymes may play a role in the binding of SARS-CoV-2 to CAD HF, DCM HF, and embryonic myocardial ACE2, affect the downstream signaling pathway mediated by ACE2, and play a role in heart failure or embryonic heart attack by SARS-CoV-2. Further experimental studies are needed to confirm these effects (16, 39).

DISCUSSION

At present, ACE2 as a human cell receptor has a strong binding affinity with SARS-CoV-2 spike protein, which plays an important role in SARS-CoV-2 invading host cells (5, 40). In our study, we performed a prediction about the combination of ACE2 and SARS-CoV-2 using protein molecular docking, and further validated the conclusion that SARS-CoV-2 spike protein can directly bind to ACE2 receptor (6, 41). Our results also suggest that mutations in the virus reduce the binding area to ACE2 and may make it more likely to bind. We constructed a protein-protein interaction network to identify key genes (ACE, AGT, AGTR1, AGTR2, and ATP6AP2) related to ACE2 activity, and it has been demonstrated that these genes are involved in renin-angiotensin system (RAS). ACE converts angiotensin I into angiotensin II by release of the terminal His-Leu, and results in an increase of the vasoconstrictor activity of angiotensin. AGT is an essential component of RAS, as a potent regulator of blood pressure, body fluid, and electrolyte homeostasis. AGTR1 as a receptor of angiotensin II mediates its action through G proteins that activate a phosphatidylinositol-calcium second messenger system. AGTR2 as a receptor of angiotensin II also cooperates with MTUS1 to inhibit ERK2 activation and cell proliferation. ATP6AP2 as a renin and prorenin cellular receptor, may mediate renin-dependent cellular responses by activating ERK1 and ERK2 as well as increasing the catalytic efficiency of renin in AGT/angiotensinogen conversion into angiotensin I. Meanwhile, we also paid attention to the expression of GJA5, GJA1, ADAM17, TMPRSS2, CTSL, and FURI (7, 42–45). The changes of these genes expression are also likely to be involved in the heart damage caused by SARS-CoV-2. The aforementioned results indicated that these molecules are closely related to ACE2, and they may play a key role in the biological processes in which ACE2 is involved.

It has been reported that ACE2 is expressed in the heart, and it is an essential regulator of heart function (46). The heart cells expressing ACE2 may act as target cells for SARS-CoV-2 invasion (40). The cardiomyocytes show diversity and specificity. Figuring out the expression profiles of ACE2 in different types of myocardial cells will provide abundant information for further explore the target of SARS-CoV-2 in the heart. Thus, we used scRNA-seq to identify the heart cell types including CM-V, CM-A, FB, EC, and MP in fetal human hearts, and further profiled the expression of ACE2 and related genes in different cell types. The results showed the expression of ACE2 was gradually increased with the fetal development among CM-V, CM-A, MP, FB, and EC, whereas the expression of ACE2 in T/B cells was increased with the gestation cycles. Although the proportions of CM-V and CM-A in embryonic development are gradually decreased, they are still the cell types with the highest proportion among all cell types. ACE2 expression levels in CM-V and CM-A were the highest in different embryonic stages (wk 5–7, wk 9–17, and wk 20–25). In addition, the expressions of ACE2 in MP and FB cells were also relatively high at wk 20–25. The expressions of AGT and AGTR2 in fibroblasts were increased significantly. ACE was highly expressed in endothelial cells and upregulated with the embryonic development. AGTR1 was mainly expressed in fibroblasts, and its expression was increased in the middle and late stages of the embryo development. In addition, the expressions of ACE2-related genes were increased significantly in CM-V with the embryonic development.

The research on how fetuses are infected by SARS-CoV-2 is still incomplete. Many researchers have reported that there is no evidence to indicate intrauterine infection from the pregnant with SARS-CoV-2 pneumonia (47, 48). However, Dong et al. (49) had recently confirmed that there were elevated IgM antibody levels in a newborn. Notably, under the stimulation of the virus, IgM can only be produced by the fetus itself, and cannot be transferred from the mother to the placenta, which explains the possibility of vertical transmission of SARS-CoV-2 (50, 51),. A study in Hong Kong showed that pregnancy patients with severe acute respiratory syndrome (SARS) infection are more likely to have spontaneous abortions and intrauterine growth retardation. Comparing with the healthy pregnant women, the pregnant women with SARS pneumonia have the increasing risk of low birth weight, premature delivery, and restricted fetal growth. Given high homology between SARS-CoV-2 and SARS-CoV, research on the harm of SARS-CoV-2 to maternal and neonatal is particularly urgent. Surprisingly, according to the most recent report, the fetus may be infected by SARS-CoV-2 due to vertical transmission from the mother in the last few days of pregnancy, although it is relatively rare (13). The latest research confirmed that SARS-CoV-2 is transmitted through the placenta in newborns (14). However, the currently reported pregnant women with COVID-19 are in the third trimester of pregnancy. Women in early pregnancy infected with SARS-CoV-2 are rare. In our study, the expressions of ACE2 in fetal heart were studied from the first 5 to 25 wk of pregnancy (from the first trimester to the second trimester). Based on our results, we speculated that if SARS-CoV-2 can infect the fetus through ACE2, with the fetal maturity, virus may tend to attack the CM-V and CM-A of the hearts. Furthermore, during the 20–25 wk of embryo, virus may also attack MP and FB. In other aspects, our results suggested that once the fetus hearts were infected with SARS-CoV-2, the expressions of ACE2-related genes (ACE, AGT, AGTR1, AGTR2, and ATP6AP2) would be changed with that of ACE2, which might lead to the dysfunction of the RAS, and eventually might lead to heart fibrosis and heart failure. The expression of ADAM17, TMPRSS2, CTSL, and FURIN increases with embryonic development, which will further promote ACE2 to bind to SARS-CoV-2 and enter the cell. And when ACE2 is consumed by combining with SARS-CoV-2, Ang II cannot be converted into Ang 1–7 more, cannot form a protective mechanism, and damage the heart RAS (Fig. 7). Meanwhile, we observed that GJA1 and GJA5 were highly expressed in the cardiomyocytes, and were gradually increased with the embryo development. Interestingly, the expression level of connexin43 at the ventricle is higher than that in atrium. On the contrary, the expression level of connexin40 in the atrium is higher. These results are consistent with the previous study (52). Our results indicated that if pregnant women are infected with SARS-CoV-2 and virus combines with ACE2 in the fetal heart, the protective effect of ACE2 on the heart is weakened and the level of Ang II is elevated, which activates RAS and may change the expressions of gap junction protein connexin40 and connexin43, thereby disturbing the Wnt signaling pathway, and eventually causes cardiac dysfunction and heart involvement.

We further studied the expressions of ACE2 and ACE2-related genes in injured heart caused by CAD HF and DCM HF. There are two pathways related to ACE2 in human, including the ACE/Angiotensin II/AT1 receptor (AGTR1) axis and the ACE2/Angiotensin 1–7/Mas receptor axis, which have a synergistic relationship to protect heart tissues (53). We found that ACE2 was mainly expressed in cardiomyocytes of normal and diseased heart tissues. The expression of ACE2 in CM-V is significantly increased in DCM HF compared with normal hearts. However, perhaps due to the limited number of cells, we did not find a significant difference in CAD HF. It has been proved that the SARS-CoV-2 virus enters the cell through binding to the ACE2 receptor on the cell membrane (54), and virus has a significant impact on the diseased heart tissue through ACE2. It has been reported that the binding of SARS-CoV-2 to ACE2 will result in the downregulation of ACE2 expression (41), following by upregulation and downregulation of ANG II and ANG1–7 expression, respectively. The upregulation of Ang II can induce mRNA expression of molecular markers of pathological hypertrophy such as brain natriuretic peptide (BNP) and α-skeletal actin (α-SA) and mRNA expression of fibrosis factor like procollagen type Iα1 and transforming growth factor-β 1 (TGF-β1) (55). In the present study, the ACE2 upregulation of CM-V in DCM HF were discovered, and it can make the SARS-CoV-2 tend to attack the CM-V of DCM HF. Once SARS-CoV-2 is combined with ACE2, the expression of ACE2 will be decreased and the expression of ANG II will be increased. According to our results, the Ang II upregulation may further worsen heart failure through affecting the calcium signaling pathway and Wnt signaling pathway.

In our study, ACE was highly expressed in EC, and was increased significantly in DCM HF, compared with the normal heart. AGT was abundantly expressed in various cell types in DCM HF, however, the expression in SMC was declined compared with normal heart and CAD HF. The ANG II receptor AGTR1 and AGT were highly expressed in CM-A in normal tissues and CAD HF, while they were plentifully expressed in CM-V in DCM HF. Moreover, the expressions of AGTR1 and AGT were increased in DCM HF, whereas the expression of AGTR2 was decreased (not detectable). All these results indicated that SARS-CoV-2 could aggravate the infection and mortality rate of patients with original heart disease due to the decrease in the expression of ACE2 and changing the protection mechanism of the ACE/ANG II/AGTR1 receptor axis and the ACE2/Angiotensin1–7/Max axis. In the investigation of confirmed cases with COVID-19, 12.5% patients had acute myocarditis, and the incidence of people with heart complications exceeded 50% (56). We speculated that, due to the basic heart failure caused by different reasons, the expression of ACE2 was increased, which elevated the infection rate of SARS-CoV-2. Furthermore, the downregulation of ACE2 caused by SARS-CoV-2 infection would induce more serious heart injury, which would aggravate the original complications and increase the mortality rate.

In addition, we analyzed the expressions of GJA1 and GJA5 mRNA in patients with heart failure caused by DCM and CAD. We found that both GJA1 and GJA5 were mainly located in CM-A and CM-V. In our study, the GJA5 was downregulated in CM-A in CAD HF, and Vozzi et al. (57) found that the connexin40 was expressed higher in atrium than that in ventricle in human heart. However, the expression level of GJA1 was enhanced in DCM HF, and some studies had reported that both decrease and increase of GJA1 could cause abnormal heart function (58, 59). Previous study also showed that the downregulation expressions of connexin40 and connexin43, which disturbed the Wnt signaling and caused the calcium signal dysregulation (36, 37, 60). It could be inferred from our results that once CAD HF was infected with SARS-CoV-2, the combination of SARS-CoV-2 and ACE2 might further cause changes in mRNA expressions of GJA1 or GJA5, aggravate the degree of heart damage, and make patients with COVID-19 develop into severe illness and even death.

In conclusion, we confirmed the expressions of ACE2 and ACE2-related genes in the embryonic heart tissues, and described the dynamic expression changes with the fetal development in detail. We proposed that SARS-CoV-2 could infect the fetus through ACE2 through attacking the CM-V and CM-A, and increase the risk of fetal mortality. We also compared the differential expressions of ACE2 and ACE2-related genes in normal and diseased adult hearts (CAD and DCM HF). Our study showed that diseased hearts were more susceptible to SARS-CoV-2 infection, especially in CM-V and CM-A of DCM HF. The regulation of RAS in the hearts infected with SARS-CoV-2 was disturbed, resulting in severe heart involvement, which increased the risk of death.

DATA AVAILABILITY

The scRNA-seq data are derived from the Gene Expression Omnibus (GEO) database under accession codes: GSE106118, GSE109816, and GSE121893. The RNA-seq data are derived from Gene Expression Omnibus (GEO) database under accession code: GSE162113.

Supplemental Figs. S1–S4 and Supplemental Table S1:https://doi.org/10.6084/m9.figshare.18469907.

GRANTS

This work was supported by the National Natural Science Foundation of China under Grant No. 31400981. This work was supported by Covid-19 Foundation of China Medical University Grant No. 202007.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

R.F. conceived and designed research; X.Z., R.Z., R.Z., X.H., W.Y., and F.W. performed experiments; X.S. and X.Z. analyzed data; X.S. and X.Z. interpreted results of experiments; X.S., X.Z., and R.Z. prepared figures; X.S., X.Z., and R.Z. drafted manuscript; X.S., L.H., and R.F. edited and revised manuscript; X.S., X.Z., R.Z., R.Z., X.H., W.Y., F.W., L.H., and R.F. approved final version of manuscript.