Single-Cell RNA-seq Identifies Cell Subsets in Human Placenta That Highly Expresses Factors to Drive Pathogenesis of SARS-CoV-2

Infection by the Severe Acute Respiratory Syndrome-Coronavirus-2 (SARS-CoV-2) results in the novel coronavirus disease COVID-19, which has posed a serious threat globally. Infection of SARS-CoV-2 during pregnancy is associated with complications like preterm labor and premature rupture of membranes; a proportion of neonates born to the infected mothers are also positive for the virus. During pregnancy, the placental barrier protects the fetus from pathogens and ensures healthy development. However, whether or not SARS-CoV-2 can infect the placenta is unknown. Herein, utilizing single-cell RNA-seq data, we report that the SARS-CoV-2 binding receptor ACE2 and the S protein priming protease TMPRSS2 are co-expressed by a subset of syncytiotrophoblasts (STB) in the first trimester and extra villous trophoblasts (EVT) in the second trimester human placenta. The ACE2- and TMPRSS2-positive (ACE2+TMPRSS2+) placental subsets express mRNA for proteins involved in viral budding and replication. These cells also express mRNA for proteins that interact with SARS-CoV-2 structural and non-structural proteins in the host cells. We also discovered unique signatures of genes in ACE2+TMPRSS2+ STBs and EVTs. The ACE2+TMPRSS2+ STBs are highly differentiated cells and express genes involved mitochondrial metabolism and glucose transport. The second trimester ACE2+TMPRSS2+ EVTs are enriched for markers of endovascular trophoblasts. Further, both these subtypes abundantly expressed genes in Toll like receptor pathway, the second trimester EVTs (but not first trimester STBs) are also enriched for component of the JAK-STAT pathway that drive inflammation. To conclude, herein we uncovered the cellular targets for SARS-CoV-2 entry and show that these cells can potentially drive viremia in the developing human placenta. Our results provide a basic framework towards understanding the paraphernalia involved in SARS-CoV-2 infections in pregnancy.

Introduction 2020). The endosomal protease cathepsin L (CTSL) can also enhance viral entry (Hoffmann et al., 2020). Thus, the presence of such receptors and S protein primer proteases in host cells is a key determinant of SARS-CoV-2 infection. Indeed, expression of ACE2 and TMPRSS2 have been detected in lung airway cells and the upper respiratory epithelium, the primary site of SARS-CoV-2 action (Sungnak et al., 2020;Ziegler et al., 2020). Beyond respiratory distress, some patients with SARS-CoV-2 viremia develop multiple organ injuries, and cells of these tissues also express ACE2 and TMPRSS2 (Qi et al., 2020;Seow et al., 2020;Zou et al., 2020) The binding of enveloped viruses like SARS-CoV-2 to its receptors results in events related to membrane fusion and/or endocytosis and establishment of the primary infection. Following its entry and uncoating, the coronavirus replicative cycle is initiated by translation of its non-structural proteins including the replicases that allow viral RNA synthesis and capping. This course requires a network of host factors to create an optimal environment for facilitating viral entry, gene expression and, RNA synthesis or virus release (de Wilde et al., 2018). Further, most enveloped viruses bud at the plasma membrane by recruiting the host endosomal sorting complex required for transport (ESCRT) machinery (Ahmed et al., 2019;Gatta and Carlton, 2019). While the precise host proteins in SARS-CoV-2 entry and replication are not yet understood, the host SARS-CoV-2 interactome has been characterized and shown to overlap with host proteins involved in endocytosis and replication of SARS-CoV-2 (Gordon et al., 2020). Thus, elucidating tissue and cell-type-specific host machinery that not only mediate viral entry but also its replication and eventually budding from the host cell is essential to understand the pathogenesis of SARS-CoV-2 infection.
Single-cell RNA-sequencing (scRNA-seq) of different tissues has transformed our ability to map the types, subsets and states of cells in healthy and diseased conditions in an unprecedented manner (Sharma et al., 2018;Szabo et al., 2019;Iyer et al., 2020). Recently, scRNA-seq has been applied to expand our understanding of the cellular landscape during viral infection including that of SARS-CoV-2 (Russell et al., 2018;Galinato et al., 2019;Liao et al., 2020), and also to identify various tissues and cells that are potential targets of SARS-CoV-2 infection (Colaco et al., 2020;Lukassen et al., 2020;Qi et al., 2020;Seow et al., 2020;Sungnak et al., 2020;Zhang et al., 2020). These studies have immensely contributed to the identification of host tissues and cellular targets for SARS-CoV-2 infection, expanding our understanding of the molecular characteristics of the host cells that are targets of viral infection.
To gain an insight into the pathogenesis of SARS-CoV-2 infection during pregnancy, it is essential to identify and characterize the placental cell types that express the host receptors ACE2 and protease enzyme TMPRESS2. Previous studies have shown that BSG/CD147 is expressed in first trimester human trophoblasts and is essential for trophoblast invasion (Lee et al., 2013). Recent studies have reported ACE2 positive cells in early embryonic trophoblasts as well as first trimester human placenta (Colaco et al., 2020;Li et al., 2020a). Expression of ACE2 protein is observed in term human placenta (https://www.proteinatlas.org/ENSG00000130234-ACE2/tissue/placenta) and also in mouse placenta on day 18 of gestation (Bharadwaj et al., 2011), However, the detailed characteristics of the ACE2 and TMPRSS2 positive cells in the placenta are unknown. Herein, we surveyed the single-cell RNA-seq data of human placenta for the expression of ACE2, TMPRSS2, alternate receptor BSG/CD147 and the endosomal protease CTSL. The present study aimed to characterize the ACE2 and TMPRSS2 positive placental cells for their ability in viral endocytosis, replication, SARS-COV-2 interactions and viral budding. The results reveal that the syncytiotrophoblasts (STB) of the first trimester and the extra villous trophoblasts (EVT) of the second trimester placenta express the receptors and S protein processing enzyme and also possess viral endocytosis, replication and viral budding machinery indicating that these cells are the potential targets for SARS-CoV-2 infection. These results provide a basic framework towards the understanding of the fundamental mechanisms of SARS-CoV-2 infections in pregnancy.

Methods
To identify the population of human placental cells that express ACE2, BSG, TMPRSS2 and CTSL at single cell resolution, we analysed scRNA-seq data of first and second trimester human placenta (Liu et al., 2018) [Accession number GSE89497]. The available data was deconvoluted for extravillous trophoblast (EVT), cytotrophoblast (CTB), syncytiotrophoblast (STB) and villous stromal cells (STR) for the first trimester placenta; only EVT data was available for second trimester human placenta. Expression was determined in terms of Transcripts Per Million (TPM) per cell. In addition, pseudo-bulk scRNA-seq data was analyzed to understand the distribution of ACE2, BSG, TMPRSS2 and CTSL in different maternal cell populations in the feto-maternal interface (Suryawanshi et al., 2018).
We profiled the mRNA levels of 27 host proteins involved in human ESCRT for viruses (Ahmed et al., 2019) and mRNA levels of 30 proteins involved in viral replication (de Wilde et al., 2018) in placental cells that co-express ACE2 and TMPRSS2 (Supplementary Table 1). We also analyzed the transcript profiles of 332 human proteins that physically interact with SARS-CoV-2 in placental cells that co-express ACE2 and TMPRSS2 (Gordon et al., 2020). To guard against viral infection, the host cells express a plethora of genes to sense the presence of the virus on the cell surface, cytosol and the endosomes. This in turn, activates the host defense program to limit or eliminate virus infection and restore tissue homeostasis. We profiled the expression of 487 host viral response genes in placental cells that co-express ACE2 and TMPRSS2 (Supplementary Table 1).
To characterize the trophoblast cells that co-express ACE2 and TMPRSS2 from their counterparts that do not express both of these genes, we carried out pseudo-bulk analysis of ACE2-and TMPRSS2-positive (ACE2+TMPRSS2+) cells and ACE2-and TMPRSS2-negative (ACE2-TMPRSS2-) trophoblast cells. Single-cell data for ACE+TMPRSS2+ and ACE2-TMPRSS2-cells was independently aggregated and the mean TPM values were computed. The data was filtered for genes whose mean values were >0.1 TPM and the ratio of the mean value in ACE2+ TMPRSS2+ cells over ACE2-TMPRSS2-cells was calculated. The genes that had a ratio of >1.5 or < 0.5 and p value of <0.05 were filtered and the data was deconvoluted for single cells. Gene ontology (GO) analysis was performed using the DAVID Bioinformatics Resources 6.8 (https://david.ncifcrf.gov/home.jsp) and PANTHER database and overrepresentation test was performed using reference genes of PANTHER pathways (http://www.pantherdb.org/). All the data was processed using R Studio version 3.6.2. All heatmaps were plotted using heatmap.2 function from gplots R package. Statistical analysis was done using the Welch's ttest and the graphs plotted in GraphPad prism version 8.0.

Results:
Trophoblast cells express mRNA for SARS-CoV-2 receptors and its spike protein processing enzymes The human placenta is characterized by four distinct cell lineages: extravillous trophoblast cells (EVT), cytotrophoblast (CTB), syncytiotrophoblast (STB) and villous stromal cells (STR). To understand the distribution of the SARS-CoV-2 receptors in these cell types, we analysed single-cell transcriptome data from publicly available datasets of human placenta to comprehensively resolve the status of placental cell susceptibility to SARS-CoV-2 infection during pregnancy. The result revealed that ACE2, BSG, TMPRSS2 and CTSL were expressed in all the cell types of first trimester placenta; however, not every cell of each type expressed these genes ( Figure 1A). As evident, the highest proportion of STBs in first trimester placenta (39%) expressed ACE2, whereas only 2% of EVTs had ACE2 expression. BSG was abundantly expressed in almost all the cells (98-100%) of each cell type of the first trimester (Supplementary Table 2). Very few cells of the human placenta expressed TMPRSS2. The highest TMPRSS2 expression was detected in STBs of first trimester placenta (23%), whereas only 1% of CTBs had TMPRSS2 expression (Supplementary Table 2). CTSL was expressed in nearly all STBs, CTBs, EVTs and STRs of first trimester placenta ( Figure 1A). The numbers of cells that express these genes individually are given in Supplementary Table 2.
We then compared the expression of these genes in EVTs of the first trimester and second trimester human placenta ( Figure 1B). In the first trimester, 2% of EVTs expressed ACE2, while 62% of second trimester EVT cells expressed ACE2. Similarly, the numbers of EVTs expressing TMPRSS2 also increased in the second trimester as compared to first trimester (2% vs 19%) (Supplementary Table 2). In both cases, the increase was statistically significant (p-value<0.001). Similar to the first trimester, nearly 100% of EVT cells of the second trimester expressed BSG (195/200) and CTSL (199/200) (Supplementary Table 2). BSG expression was significantly reduced in second trimester EVTs as compared to first trimester (p-value<0.001); the expression of CTSL was significantly (p-value<0.001) higher in second trimester EVTs as compared to first trimester EVTs ( Figure 1B).
We also studied the expression of other SARS-CoV receptors DPP4 and ANPEP, which are utilized by Middle East Respiratory Syndrome coronavirus (MERS-CoV) and CoV-229E respectively. ANPEP was detected in the EVTs, CTBs and STRs but not the STBs of the first trimester placenta. DPP4 on the other hand was expressed by all the CTBs, STBs, EVTs and STRs (Supplementary Figure 1A). As compared to first trimester EVTs, the levels of ANPEP significantly decreased (p-value<0.001), while that of DPP4 significantly increased (p-value<0.001) in the second trimester EVTs (Supplementary Figure 1B).
To determine whether or not endometrial cells express these genes, we analysed pseudo-bulk data of the first trimester feto-maternal interface (Supplementary Figure 2). Expression of ACE2 was detected in smooth muscle cells, and a low abundance of ACE2 transcripts was also detected in decidual stromal cells and fibroblasts, vascular endothelial cells and NK cells.

Co-expression of mRNA of SARS-CoV-2 receptors and spike protein processing enzymes in human placental cells
SARS-CoV-2 infection in host cells requires coordinated expression of the entry receptor and S protein primer. We aimed to identify cells co-expressing ACE2, TMPRSS2, BSG and CTSL in different combinations (Figure 2). The results revealed that only a subset of STBs (14%) in the first trimester placenta co-expressed ACE2 and TMPRSS2 (Supplementary Table 3). No other cell types in first trimester placenta expressed this receptor and S protein primer protease pair; although, there were a number of cells expressing ACE2 (Figure 2). However, 15% of EVTs in the second trimester placenta co-expressed ACE2 and TMPRSS2 (Supplementary Table 3).
In the first trimester placenta, all the ACE2-positive trophoblast subtypes co-expressed BSG and CTSL, and all the BSG-positive cells co-expressed CTSL. All the ACE2-positive second trimester EVTs co-expressed BSG and CTSL, and all the BSG-positive cells co-expressed CTSL ( Figure 2). The absolute numbers and the percentage of the co-expressing cells are given in Supplementary Table 3

ACE2+TMPRSS2+ first trimester syncytiotrophoblast cells are highly differentiated and express the machinery for viral endocytosis, replication and budding
Since only STBs co-expressed ACE2 and TMPRSS2 in the first trimester, we carried out an indepth characterization of these cells. 14% of the total STB population of the first trimester placenta expressed both ACE2 and TMPRSS2, 52% did not express either, and the rest of the cells expressed either ACE2 or TMPRSS2 (Supplementary Table 3). We compared the expression profiles of classical STB genes between the ACE2 and TMPRSS2 co-expressing cells and ACE2-and TMPRSS2-negative STBs. We observed that both cell types abundantly expressed the transcripts for human chorionic gonadotropin beta 5 (CGB5) and somatomammotropin [placental lactogen (CSH1)], as well as steroid hormone biosynthesis enzymes (HSD17B1 and CYP19A1). Further, both these subsets of STBs also abundantly expressed the other putative SARS-CoV-2 S protein primers FURIN and Cathepsin B (CTSB) (Supplementary Figure 3).
We next analysed the mRNA levels of 27 genes involved in human ESCRT of viruses and 30 host genes involved in SARS-CoV replication in ACE2-and TMPRSS2-positive cells in STBs. All the ACE2+TMPRSS2+ STBs uniformly expressed most of these genes ( Figure 3D); however, the other cells had a very heterogeneous expression across the different subtypes (Supplementary Figure 4). We also analysed the mRNA levels of 332 host proteins that are known to interact with SARS-CoV-2 and found that there was minimal heterogeneity in expression of these genes in the first trimester STBs ( Figure 3D) as compared to other cell types (Supplementary Figure 4).

Second trimester ACE2+TMPRSS2+ cells are invasive extravillous trophoblasts and express markers of endovascular trophoblasts
Amongst the second trimester EVTs, 15% of cells were ACE2+TMPRSS2+ while 33% did not express either of the transcripts (Supplementary Table 3). Two populations of second trimester EVTs are reported and characterized by the expression of TAC3. Type 1 EVTs are TAC3-high and express genes involved in migration and invasion; type 2 TAC3-low cells express genes involved in cell proliferation (Liu et al., 2018). In addition to TAC3, the type 1 EVTs also express JAM2, SERPENIN1 and PRG2 (Liu et al., 2018). We observed that the levels of TAC3 were marginally but not significantly higher in ACE2+TMPRSS2+ EVTs (Supplementary Figure 5A), and the mRNA levels of other genes were identical in ACE2+TMPRSS2+ EVTs as compared to cells not expressing either of the two genes (ACE2-TMPRSS2-) (Supplementary Figure 5A). Principal component analysis did not reveal major differences in the transcriptome of the ACE2+TMPRSS2+ and ACE2-TMPRSS2-cells (Supplementary Figure 5B). We compared the expression profiles of classical EVT genes between the ACE2+TMPRSS2+ and ACE2-TMPRSS2-cells and observed that both the cell types abundantly expressed the transcripts for HLA-G and ITGB1 (Supplementary Figure  5C). Both these subsets of EVTs also abundantly expressed other SARS-CoV-2 S protein primer proteins FURIN and CTSB of which the levels of FURIN were significantly higher (p-value<0.05) in ACE2+TMPRSS2+ EVTs as compared to ACE2-TMPRSS2-EVTs (Supplementary Figure 5C).
To characterize if there are any specific classes of genes differentially abundant between the ACE2+TMPRSS2+ versus the ACE2-TMPRSS2-EVT cells, pseudo-bulk analysis was carried out. There were 983 differentially abundant genes (including ACE2 and TMPRSS2) between these two cell types (Supplementary Table 5) of which 931 were overrepresented and 52 were underrepresented in the ACE2+TMPRSS2+ cells as compared to ACE2-TMPRSS2-cells. Further, these genes were very heterogeneously expressed in the ACE2-TMPRSS2-EVTs while most ACE2+TMPRSS2+ cells uniformly expressed these genes ( Figure 4A). Most of these differentially abundant genes enriched GO biological processes including viral entry, release and intracellular transport ( Figure 4B), and other enriched GO biological processes were nucleic acid replication, epithelial morphogenesis and cell migration ( Figure 4B).
Of note, the ACE2+TMPRSS2+ cells significantly (p-value<0.05) overexpressed the markers of endovascular trophoblasts CDH5, VCAM, CCR1 and CD59 ( Figure 4C). These cells also overexpressed OVOL2 and ICAM, the markers of terminally differentiated EVTs, as well as the invasion-related marker AKT1 ( Figure 4C). The increase in the expression of OVOL2 and AKT1 was statistically significant (p-value<0.05).
Analysis of the mRNA levels of 27 genes involved in human ESCRT of viruses and 30 host genes involved in SARS-CoV replication in ACE2-and TMPRSS2-positive cells at single-cell resolution revealed that all the ACE2+TMPRSS2+ EVTs uniformly expressed most of these genes ( Figure 4D), while the first trimester EVTs that had no ACE2+TMPRSS2+ cells had a very heterogeneous expression of these genes (Supplementary Figure 4). We also analysed the mRNA levels of 332 host proteins that interact with SARS-CoV-2 and observed that almost all these genes were expressed in most ACE2+ second trimester EVTs (Figure 4D), while the first trimester EVTs had a very heterogeneous expression (Supplementary Figure 4).

Unique signatures of genes involved in viral response in first trimester STB and second trimester EVTs
We studied the baseline expression of 487 genes involved in viral response in both first trimester STBs and second trimester EVTs (Supplementary Table 1). Only a subset of viral response genes was expressed in both EVTs and STBs irrespective of ACE2 and TMPRSS2 status ( Figure 5A). The heatmaps showed minimal heterogeneity across cells but high variability in expression across genes involved in the viral response. To characterize these genes in EVT and STB cells, genes were clustered by a hierarchical clustering method using the "hclust" function available in R. Genes were grouped based on the most optimum threshold, which resulted in four different gene clusters ( Figure 5B). Further, the mean expression of each gene cluster was estimated and checked across EVT and STB cells of which the mean expression of genes in Clusters 1 and 2 were identical in both EVTs and STBs. However, genes in Clusters 3 and 4 were more abundant in EVTs as compared to STBs ( Figure 5C).
Next, GO analysis was performed using the PANTHER database for all the four gene clusters. For each cluster, an over-representation test was performed using reference genes of PANTHER pathways, and a pathway with the highest fold-enrichment value was selected as the enriched pathway for a given gene cluster. GO classification of these revealed that most of the genes in Cluster 1 had a role in the Toll-like receptor (TLR) signalling response and the genes in Cluster 2 had a role in apoptosis. Additionally, genes of the JAK-STAT pathway and axon guidance mediated by semaphorins were enriched in EVTs as compared to STBs (p-value<0.05) ( Figure 5D).

Figure 5: Viral response genes in first trimester syncytiotrophoblasts and second trimester extravillous trophoblasts A)
Heat map showing the expression of genes involved in viral response in ACE2-and TMPRSS2-positive (ACE2+TMPRSS2+) cells of first trimester STB (n=9) and second trimester EVT (n=29). Each row is a gene and each column represents data for a single ACE2 and TMPRSS2 co-expressing cell. B) Cluster dendrogram of first trimester STB and second trimester EVT cells. The horizontal axis of the dendrogram represents the distance or dissimilarity between clusters. The vertical axis represents the clusters. C) Mean of four clusters of upregulated genes in first trimester STB and second trimester EVT cells. Y axis represents cluster mean expression and X axis represents different cell types. D) Pathways enriched in the in four clusters of first trimester STBs and second trimester EVTs, with their enrichment score and p-values. EVT =, STB = Syncytiotrophoblast. Data was extracted from single cell RNA-seq of human placenta (Liu et al 2018) [Accession number GSE89497].

Discussion
Herein, we utilized scRNA-seq to identify and characterize the potential cellular targets of SARS-CoV-2 infection in human placenta. To review the data presented: 1) The SARS-CoV-2 binding receptor ACE2 and the S protein priming protease TMPRSS2 are co-expressed by a subset of syncytiotrophoblasts (STB) in the first trimester and extravillous trophoblasts (EVT) in the second trimester human placenta 2) These ACE2+TMPRSS2+ subsets are highly differentiated STBs and endovascular EVTs 3) The ACE2+TMPRSS2+ placental subsets readily express mRNA for proteins involved in ESCRT of viruses and replication as well as transcripts for proteins that are known to interact with SARS-CoV-2 structural and nonstructural proteins 4) STBs and EVTs differentially express genes involved in the host response to viral infection Using a scRNA-seq dataset of first trimester human feto-maternal interface (Vento-Tormo et al., 2018), a recent study reported the presence of ACE2-positive villous cytotrophoblasts and syncytiotrophoblasts but not the extravillous trophoblasts of first trimester human placenta (Li et al., 2020a). Herein using a different dataset (Liu et al., 2018), we corroborated these findings and show the presence of ACE2 in cytotrophoblast and syncytiotrophoblast cells of first trimester placenta. However, a low-abundance subset of ACE2+ extravillous trophoblasts was additionally identified in the present study. Differences in methods of tissue sampling, cell isolation and inefficiencies in detection of low-abundance transcripts in scRNA-seq can underestimate the actual frequencies of ACE2+ cells in a given tissue. In addition to ACE2 (the SARS CoV and SARS-COV-2 receptor) we also detected expression of DPP4 (the receptor for MERS-CoV) and ANPEP (the receptor for CoV-229E) in the cells of the placenta. Like ACE2, DPP4 was detected in all the cell types of the first trimester placenta and also in EVTs of second trimester, very few STBs expressed ANPEP suggesting different cell types could be targets of different corona viruses in the developing placenta.
While ACE2 is the primary receptor for SARS-CoV-2 entry, the S protein of SARS-CoV-2 undergoes cleavage by the cell surface protease TMPRSS2 (Hoffmann et al., 2020). Whether ACE2 and TMPRSS2 are required on the same cell to activate SARS-CoV-2 S-protein to invade ACE2 single-positive cells is a matter of investigation. However, as active S protein has a finite lifetime (Shulla et al., 2011), its processing at the plasma membrane will make it most effective for viral entry. Thus, we assumed that for SARS-CoV-2, the ACE2 and TMPRSS2 co-expressing cells would have the highest infectivity. Our analysis revealed a proportion of STBs (14%) in the first trimester and a subset of EVTs (15%) in second trimester human placenta co-express ACE2 and TMPRSS2. While these numbers may appear insignificant considering the total placental volume, it must be borne in mind that only 3-6% of lung airway epithelial cell subtypes (the primary site of SARS-CoV-2 action) express both ACE2 and TMPRSS2 (Ziegler et al., 2020). Therefore, we propose that only the first trimester STBs and second trimester EVTs are likely to be potential common entry routes for SARS-CoV-2 in the placenta.
In addition to ACE2-expressing STBs and EVTs, our study revealed that BSG/CD147, the alternate receptor for SARS-CoV-2 , is expressed by almost all the placental cells. Presently, the mechanism by which BSG/CD147 mediates viral entry in host cells is unknown. In other cells, BSG/CD147 promotes entry of viruses by endocytosis (Pushkarsky et al., 2001). It is possible that the same mechanism may be operative in the case of SARS-CoV-2. In this context, it is especially interesting that all the BSG/CD147-positive cells co-express the endosomal protease CTSL, indicating that both ACE2-and BSG/CD147-mediated viral entry may be operative in placental cells. Further, we observed that almost all the ACE2+ STBs and EVTs co-expressed BSG/CD147, suggesting that more than one mechanism may operate for viral entry in these cells of the human placenta. Beyond TMPRSS2, studies have identified that SARS-CoV-2 may have a FURIN cleavage site, leading to a broader set of host proteases that could mediate S-protein priming (Coutard et al., 2020). The ACE2+TMPRSS2+ STBs and EVTs abundantly express FURIN as well as another endosomal protease, CTSB. Together our data conclusively show that multiple cells of human placenta are targets for SARS-CoV-2 binding and entry by S protein priming.
We next characterized the placental cells that are potential targets for SARS-CoV-2 infection. As the ACE2-mediated viral entry is a well-established mechanism, we focused only on ACE2+TMPRSS2+ cells in the placenta. In the developing placenta, the trophoblast stem cells differentiate into cytotrophoblasts, which undergo further differentiation to form the non-selfrenewing cytotrophoblasts, extravillous trophoblasts and syncytiotrophoblasts (Turco and Moffett, 2019;Hemberger et al., 2020), The syncytiotrophoblasts covering the villi are major hormone secreting cells and function as a protective immunological barrier (Maltepe and Fisher, 2015;Gupta et al., 2016;Liu et al., 2018;Vento-Tormo et al., 2018;Turco and Moffett, 2019). We observed that the STBs that co-express both ACE2 and TMPRSS2 also express the mRNA for the peptide hormones and enzymes for steroid hormone biosynthesis. However, their levels are not significantly different from the ACE2-TMPRSS2-counterparts, suggesting that both these cell types retain the basic functions of STBs. However, pseudo-bulk analysis revealed that the ACE2+TMPRSS2+ cells are enriched for genes involved in cell cycle checkpoints, actin filament remodelling, mitochondrial functions, hexose transport and type I interferon signalling. Indeed, the terminally differentiated STBs have replicative senescence and require extensive cytoskeletal remodelling for syncytialization; the mitochondria of STBs play a key role in progesterone synthesis by providing cholesterol (Martinez et al., 2015). Additionally, these cells are enriched in OVOL1, the transcription factor required for STB specification, as well as proteins involved in glucose transport across the feto-maternal barrier, a key function of well differentiated STBs (Jansson and Ylve, 2002;Renaud et al., 2015;Vento-Tormo et al., 2018;Turco and Moffett, 2019). These results imply that the ACE2+TMPRSS2+ cells are a subset of highly differentiated STBs and these cells are potential target for viral entry. What could be the consequence of viral infection on STBs is yet to be demonstrated, increased syncytiotrophoblastic knots are observed in placenta from pregnant women with COVID-19 , which is suggestive of injury to the STBs in the placenta.
We next probed the second trimester EVTs, 15% of which co-express ACE2 and TMPRSS2. The EVTs differentiate from cytotrophoblast stem cells and populate the tips of the placental villi to form the anchoring villi, thus defining the boundary between mother and fetus. The EVTs are central to placentation as they invade into maternal decidua and are involved in remodelling of maternal spiral arteries, veins and lymphatics (Sharma et al., 2016;Pollheimer et al., 2018). We observed that the ACE2+TMPRSS2+ cells abundantly express the classical EVT marker ITGB1 and also HLA-G that induces tolerogenic immune responses leading to acceptance of the semi-allogeneic. Two kinds of EVTs are reported in the second trimester human placenta: the proliferative EVTs in the cell columns and the invasive endovascular or interstitial EVTs (Pollheimer et al., 2018;Turco and Moffett, 2019) and both these have a unique transcript signature (Liu et al., 2018). Herein, we observed that while the levels of TAC3 and other molecules associated with columnar versus invasive EVTs are not significantly different between ACE2+TMPRSS2+ and ACE2-TMPRSS2-second trimester EVTs, the double-positive cells overexpressed key invasive EVT markers OVOL2, GJA5, ICAM and AKT1 (Sharma et al., 2016;Bai et al., 2018;Liu et al., 2018;Jeyarajah et al., 2020), suggesting that these cells are invasive trophoblasts. We further observed that many of the ACE2+TMPRSS2+ cells were enriched in genes having a role in cell migration. The EVTs can either invade in to the decidua (called as interstitial EVTs) or remodel the spiral arteries (termed as endovascular EVTs). While both these EVTs are invasive in nature, they have differential expression of certain marker genes. For e.g. the endovascular EVTs overexpress the vascular endothelial cadherin (CDH5) and vascular cell adhesion molecule-1 (VCAM-1), they also have higher expression of CCR1 and CD59 (Bulla et al., 2005;Cartwright and Balarajah, 2005;Liu et al., 2018;Ueda et al., 2019;Sato, 2020). Intriguingly, the ACE2+TMPRSS2+ cells also overexpressed several of the key endovascular EVT markers indicating that the ACE2+TMPRSS2+ population of second trimester EVTs are potentially endovascular trophoblasts and are targets of SARS-CoV-2 infection. Indeed, pseudo-bulk analysis of the ACE2+TMPRSS2+ cells and ACE2-TMPRSS2-cells revealed significant enrichment of genes with GO terms involving regulation of viral release from host cells. Most of the ACE2+TMPRSS2+ endovascular EVTs abundantly expressed most genes whose protein products in the host are known to be involved in human endocytosis and budding of viruses and replication. Together, this data shows that SARS-CoV-2 may affect the invading EVTs at the feto-maternal interface in the second trimester and can result in damaged vasculature. In this context, it is important to note that the maternal endothelial cells in the decidua also express ACE2 and BSG, making the maternal endothelium another entry point of SARS-CoV-2 infection at the feto-maternal interface; any damage to these cells can cause placental damage and vertical transmission of the virus. Indeed, increased fibrin deposition, chorionic haemangioma and massive placental infarction indicative of vasculopathy are observed in term placenta of women infected with SARS-CoV. An increase in intervillous and subchorionic fibrin and fetal thrombotic vasculopathy with zones of avascular fibrotic villi are also reported in placenta of women infected with coronaviruses including SARS CoV-2 (Ng et al., 2006;Baud et al., 2020;Mulvey et al., 2020). Together, these results indicate that the integrity of endovascular trophoblasts and the endothelial compartment of the feto-maternal interface may be compromised in women with SARS-CoV-2 infection.
Once the virus binds to their receptors on host cells and gains an entry it utilizes a plethora of host genes for its replication. Post replication, the enveloped viruses complete their cycle by forming vesicles that bud from plasma membrane via the cellular ESCRT (endosomal sorting complexes required for transport) machinery. Interestingly, we observed that the ACE2+TMPRSS2+ STBs and EVTs were enriched for the key genes that encode for proteins involved in ESCRT and viral replication. Using affinity-purification mass spectrometry 332 human proteins that interact with SARS-CoV-2 have been identified and many of these play a role in ESCRT and viral replication (Gordon et al., 2020). We observed that most ACE2+TMPRSS2+ STBs and EVTs abundantly expressed most of these genes. Thus SARS-CoV-2 would hijack proteins in the EVTs and STBs thereby interfering with normal placental functions. In this context it is important to highlight that a significant proportion of the human-SARS-CoV-2 interacting proteins also interact with proteins of other viruses including Zika and HCV which replicate in the trophoblast cells (Giugliano et al., 2015;Tabata et al., 2016Tabata et al., , 2018Gordon et al., 2020) Together our data strongly imply the first trimester STBs and second trimester EVTs are not just targets for SARS-CoV-2 entry but the virus would also be potentially pathogenic to these cells.
A proportion of SARS-CoV-2 proteins target the members of the innate immune signalling pathway including the NF-kappa-B (Gordon et al., 2020). We decided to probe this in further details by profiling the STBs and EVTs for 487 genes whose protein products are involved in viral response in host cells. We observe that only a proportion of the genes are expressed in most EVTs and STBs. Based on their expression levels in the host cells, they could be classified in four clusters and interestingly, the first trimester STBs and the second trimester EVTs expressed genes in the TLR signalling pathway, the primary response to viral infection. Previous studies in SARS-CoV have identified involvement of TLR pathways in protection against viral response (Dosch et al., 2009;Totura et al., 2015). However, we found that the genes in the JAK-STAT pathway were overexpressed in in the EVTs but not STBs. This is however not surprising as the JAK-STAT pathway is required for physiological function of EVTs mainly invasion (Fitzgerald et al., 2010;Suman et al., 2013;Sharma et al., 2016;Godbole et al., 2017). However, the JAK-STAT pathway is also central for mounting a host response to viral infection and treatment with interferon gamma induces the expression of interferon-stimulated genes in EVT cells (Verma et al., 2020). Thus, EVTs are not just the entry sites for SARS-CoV-2 infection, but they also have the cellular machinery to mount an inflammatory response towards an infection. However, with regards to coronaviruses including SARS-CoV-2 there is an overexuberant inflammatory response even at lower viral loads which contributes to the viral pathogenicity in the lungs (Liao et al., 2020). Further, the ACE2 receptors are induced by the interferon signalling at least in the lung cells (Ziegler et al., 2020) thereby amplifying the infection cycle in the host tissues. Whether a similar mechanism is operative in the placental cells is under investigation, the heightened baseline expression of the JAK-STAT pathway genes in the EVTs itself could readily lead to placental inflammation which may be detrimental to pregnancy. Infiltration of leucocytes and chorioamnionitis is observed in placenta from women with COVID-19 (Ng et al., 2006;Baud et al., 2020). Since inflammation of the feto-maternal interface causes preterm births (Silasi et al., 2015;Surve et al., 2016), it is plausible that the increased incidence of preterm delivery in women with COVID-19 could be linked to this process.
Beyond preterm births, the demonstration that the ACE+TMPRSS2+ subpopulation of EVTs consists of invasive endovascular trophoblasts is clinically relevant in conditions like preeclampsia. The invasion of the trophoblast cells and remodelling of the spiral arteries deep into the myometrium is essential for normal fetal growth and development (Norwitz, 2006;Soares et al., 2015). If the arteries are not sufficiently remodelled, there is disordered perfusion of blood and an inadequate supply of nutrients and oxygen, resulting in fetal growth restriction, stillbirth, preeclampsia, placental abruption and preterm labor (Brosens et al., 2019). Since SARS-CoV-2 and other coronaviruses may target the endovascular trophoblasts, it is plausible that the infection could lead to other adverse pregnancy outcomes. Indeed, preeclampsia, preterm labor, fetal distress and premature rupture of membranes are reported in pregnant women infected with SARS-CoV-2 in the third trimester (Gajbhiye et al., 2020). For other human coronavirus infections also high rates of miscarriages, preterm birth and premature rupture of membranes have been reported (Alfaraj et al., 2019;Mullins et al., 2020) . These observations imply that SARS-CoV-2 infection is detrimental to pregnancy due the possible infection of placenta cells such as EVTs.
Although our studies demonstrate the presence of SARS-CoV-2 receptors in the placenta, viral mRNA is not reported in placenta of most babies born to mothers with COVID-19 in the third trimester (Gajbhiye et al., 2020;Mulvey et al., 2020). While inadequate sampling of the cells during sample collection, site of sampling and sensitivity of the methods used can underestimate the presence of virus in tissues, it must also be borne in mind that most of the infections have occurred in the late third trimester, often just before delivery and when the viral loads are unknown. Nevertheless, vertical transmission of the virus from mother to child is observed in 8% of cases (Gajbhiye et al., 2020), presence of viral RNA in amniotic fluid and placenta of a mother infected with SARS-CoV-2 are reported (Baud et al., 2020;Zamaniyan et al., 2020). These observations point towards the possibility of placental damage resulting in the breakthrough of SARS-CoV-2 infections. However, more cases should be investigated before reaching a final conclusion.
To summarise, herein we exploited the power of scRNA-seq and identified the cellular basis of SARS-CoV-2 infection in the human placenta. This data provides crucial insight regarding where the genes required for viral endocytosis, replication, budding and response were expressed amongst specific subsets of placental cells. It will now be essential to determine how SARS-CoV-2 infection alters the temporal dynamics of host responses at the single-cell resolution. Given the complexities of host-pathogen interactions in SARS-CoV-2 infections, a deep understanding of the full spectrum of mechanisms operative in different tissues will be required to combat this pandemic. Our results provide a basic framework in understanding of the paraphernalia involved in SARS-CoV-2 infections in pregnancy. We believe that this work will also aid in developing rational strategies for management of COVID-19 and other coronavirus infections in pregnancy.