Identification of Cell Types and Transcriptome Landscapes of Porcine Epidemic Diarrhea Virus–Infected Porcine Small Intestine Using Single-Cell RNA Sequencing

Coronaviruses are one of the most important zoonotic pathogens; they represent an ongoing public health threat because of their high variability and complex cross-species transmission ability (1). In the twenty-first century, emerging and reemerging coronavirus diseases have appeared suddenly and spread globally, demonstrating a critical need for strategies that identify high-risk zoonotic coronaviruses (2). Several swine coronaviruses cause diarrhea symptoms in piglet, including transmissible gastroenteritis virus, porcine epidemic diarrhea (PED) virus (PEDV), porcine deltacoronavirus (PDCoV), and swine acute diarrhea syndrome coronavirus (3–6). However, to date, the identification of cell types in the porcine intestinal tract has not been reported, and the biological responses of different cell types after swine coronavirus infection have also been poorly studied.

PEDV, first discovered in 1971, is an ancient virus. Infection with this virus causes PED, a highly infectious digestive tract disease that can cause vomiting and diarrhea in pigs and has an especially high mortality rate in piglets (2, 7, 8). Since the winter of 2010, a mutant PEDV strain with higher pathogenicity caused serious PED outbreaks worldwide (9–12), and it can have 100% mortality in piglets aged ≤7 d (8). Several studies applied high-throughput technologies to analyze the cellular responses to PEDV infection at the population level, whether by using the cell lines in vitro or the intestinal tissues in vivo (13–16). However, the mixture of multiple cell types was also a major short board of these studies. The heterogeneities of different cell types result in different responses to viral infection. Therefore, under the premise of defining cell types by using single-cell RNA sequencing (scRNA-seq), the systematic analyses of the responses made by different cells would greatly improve the understanding of the pathogenicity of the virus.

Porcine enterocytes have been confirmed as the target cells of PEDV, and viral infection alters a variety of cellular functions, such as apoptosis and intercellular tight junctions (17–21). Jung et al. (22) found that goblet cells have a limited infection capacity for PEDV. Goblet cells, which secrete mucin 2 (MUC2) and multiple mucous components, play an important role in maintaining the integrity of the intestinal mucous layer (23). In addition, antimicrobial peptides (AMPs) and secretory IgA are two other important components that make up the mucous layer of the mucosal barrier, which protects the mucosa of the gastrointestinal tract and plays a key role in maintaining the host homeostasis (24). However, the studies of PEDV infection on mucous layer are still not comprehensive.

In this study, we performed scRNA-seq of the jejunum of piglets and analyzed the transcriptome landscapes of normal and PEDV-infected intestinal and immune cells. We identified different cell types according to representative markers in the jejunum and confirmed the susceptibilities of goblet and tuft cells to PEDV. The alteration of function of different intestinal and immune cell types by PEDV infection was analyzed. Downregulation of multiple mucins by PEDV-infected goblet cells and antiviral capacity of REG3G were determined for the first time, to our knowledge. This study provides a more comprehensive analysis of the alteration of intestinal mucosal barrier in piglets by PEDV infection and furthers the understanding of its pathogenic mechanism.

A PEDV variant strain, AH2012/12 (GenBank accession number KU646831), which belongs to the G2 gene group, was isolated from a pig farm in China (25). Eight 3-d-old piglets selected from one sow, which was PEDV RNA and specific Ab negative, were further confirmed negative for transmissible gastroenteritis virus, porcine rotavirus, and PDCoV by virus-specific RT-PCRs and then were divided into two groups of four piglets each and housed in separate rooms. The piglets were fed a mixture of liquid milk replacement and yogurt and given free access to water. Those in the case group were challenged orally with 2 ml of AH2012/12 virus solution with 10⁶ 50% tissue culture infective dose (TCID₅₀)/ml. The control group piglets were inoculated with an equal amount of cell culture medium. All animals were monitored daily for clinical signs of disease, including diarrhea and vomiting. Rectal swabs were collected for scoring fecal denseness (scored as normal, 0; pasty stool, 1; semiliquid diarrhea, 2; and liquid diarrhea, 3) and for enumerating fecal viral RNA shedding by quantitative RT-PCR (25, 26). At necropsy, the intestinal tissues and contents were evaluated grossly. Various intestinal segments were frozen for viral load determination or fixed in 10% neutral-buffered formalin for histopathology. All animal experiments were performed with the approval of the Jiangsu Academy of Agricultural Sciences Experimental Animal Ethics Committee (NKYVET 2015-0127) and in accordance with relevant guidelines and regulations. All efforts were made to minimize animal suffering and to reduce the number of animals used.

Jejunum segments from piglets at 24 h postinfection (hpi) in the case and control groups were isolated and processed to obtain single-cell sequencing samples, then transported rapidly to the research facility. Each sample was subsequently minced on ice into pieces <1 mm³ in size, followed by enzymatic digestion using TrypLE Express (Life Technologies) with manual shaking every 5 min. Samples were then centrifuged at 300 × g for 30 s at room temperature, and the resulting supernatant was removed without disturbing the cell pellet. The cell pellet was resuspended with 1× PBS (calcium and magnesium free) containing 0.04% w/v BSA (400 µg/ml), and this mixture was centrifuged at 300 relative centrifugal force for 5 min. The resulting cell pellet was resuspended in 1 ml of RBC lysis buffer and then incubated for 10 min at 4°C. After the RBCs were lysed, the samples were resuspended in 1 ml of PBS containing 0.04% BSA. Next, the samples were filtered over Scienceware Flowmi 40-µm cell strainers (VWR). After sample dissociation, the cell concentration and cell viability were determined by using a hemocytometer and trypan blue staining.

Single-cell suspensions were loaded onto the Chromium Controller (10× Genomics) for droplet formation. Single-cell sequencing libraries were prepared using the Chromium Single Cell 3′ Reagent Kit (10× Genomics) in accordance with the manufacturer’s protocol. Samples were sequenced on the NovaSeq6000 with 150 paired reads to generate a gene-expression library.

The Sus scrofa genome version Sscrofa11.1 was used to align the scRNA-seq data. The Cell Ranger software pipeline (version 3.1.0) provided by 10× Genomics was used to detect multiplex cellular barcodes, map reads to the genome and transcriptome using the STAR aligner, and down-sample reads as required to generate normalized aggregate data across samples, producing a matrix of gene counts versus cells. We processed the unique molecular identifier count matrix using the R package Seurat (version 3.1.1) (27) to remove low-quality cells and likely multiple captures. A total of 12,774 single cells were included in our downstream analyses. Library size normalization was performed with the NormalizeData function in Seurat to obtain the normalized count. Specifically, the global-scaling normalization method “LogNormalize” was used to normalize the gene expression measurements for each cell by the total expression, multiplied by a scaling factor (10,000 by default), and the results were log transformed.

Top variable genes across single cells were identified using the method described by Macosko et al. (28). The most variable genes were selected using the Find Variable Genes function (mean. function = ExpMean, dispersion. function = LogVMR) in Seurat. To remove batch effects from the scRNA-seq data, we performed the mutual nearest neighbors method, as described by Haghverdi et al. (29), with the R package batchelor. Graph-based clustering was performed to cluster cells according to their gene expression profile using the FindClusters function in Seurat. Cells were visualized by using a two-dimensional t-distributed stochastic neighbor embedding (t-SNE) algorithm with the RunTSNE function in Seurat. We used the FindAllMarkers function (test.use = bimod) in Seurat to identify marker genes for each cluster. For each cluster, FindAllMarkers identified positive markers compared with all other cells.

Differentially expressed genes were identified using the Find Markers function (test.use = MAST) in Seurat. The threshold for significantly differential expression was set at p < 0.05 and |log2foldchange| > 1. Gene Ontology enrichment and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analyses of differentially expressed genes were each performed using R based on the hypergeometric distribution.

The fixed jejunum tissues were placed at 4°C overnight. After being washed three times with PBS, the tissues were dehydrated in 20% sucrose and embedded in Tissue-Tek OCT. The slices were then permeabilized with 0.5% PBS-Triton for 30 min and blocked with FBS (Visteck) in 0.5% PBS-Triton for 1.5 h at room temperature. The slices were incubated with primary Abs diluted in blocking solution overnight at 4°C and then incubated with secondary Abs diluted in blocking solution for 2 h at room temperature. DAPI (blue) was used to stain the cell nuclei. Finally, tissue slices were mounted for use in confocal laser scanning microscopy.

To determine the effects of PEDV infection on host protein genes, we obtained the total RNA of jejunum tissues from case and control piglets and used them in real-time PCR assays for analyzing the mRNA levels of different host genes. In addition, IPEC-J2 or human colon cancer goblet cell line L174T was cultured to a single layer in the wells of a 24-well cell plate and then infected with PEDV AH2012/12 at a multiplicity of infection (MOI) of 0.1 in the presence of trypsin (2 µg/ml in IPEC-J2 or 4 µg/ml in L174T), which is needed for virus entry into cells (25). At different time points, total RNA of the cells was extracted for analyzing the expression levels of different host protein genes. The total RNA and protein lysate of the cells obtained at different time points were analyzed to determine the change in host protein genes after PEDV infection.

Recombinant vector to induce the expression of swine-origin REG3G was constructed using pcDNA3.1 with a Flag tag at the C terminus of the target gene. Different amounts of expression vector for REG3G were transfected into IPEC-J2 cells by using Lipofectamine 3000 transfection reagent in accordance with the manufacturer’s recommendations (Thermo). Other IPEC-J2 cells were transfected with 50 and 100 nM small interfering RNAs (siRNAs) for REG3G or negative control siRNA (Ribobio, Guangzhou, China). After 24 h, all cells were infected with AH2012/12 at 0.1 MOI in the presence of 2 µg/ml trypsin. At 24 hpi, the total RNA and protein lysate of the cells were analyzed to determine the PEDV yields and protein levels of PEDV N protein (PEDV-N). Sequences of all primers used for the recombinant vectors and siRNA are listed in Supplemental Table I.

Total RNA was extracted using a total RNA kit (Omega, Guangzhou, China), and cDNA was prepared using the HiScriptII Q RT SuperMix (+gDNA wiper) (Vazyme), both in accordance with the manufacturer’s instructions. Real-time PCR was performed using the ChamQ Universal SYBR qPCR Master Mix (Vazyme) and AceQ qPCR Probe Master Mix (Vazyme) for the analyses of target gene relative expressions and viral gene copy numbers. Viral gene copy numbers were measured as previously reported (25, 26). Sequences of all primers used in these assays are listed in Supplemental Table I.

All data are presented as means ± SD. Student t tests and one-way ANOVA tests were used to compare the data from differently treated groups. Differences with p values <0.05 were considered significant. All statistical analyses and calculations were performed using GraphPad PRISM software (version 7.0 for Windows; GraphPad Software).

Four 3-d-old piglets in the case group were challenged orally with 2 ml PEDV variant strain AH2012/12 (10⁶ TCID₅₀/ml). At 24 hpi, all piglets in the case group showed severe diarrhea symptoms, and their fecal diarrhea scores were ∼2 (Fig. 1A, 1B). The fecal samples from all challenged piglets had a viral copy number of >1 × 10⁶ copies/ml (Fig. 1C), which indicated that PEDV effectively replicated in the intestinal tract of piglets. After the piglets were euthanized, clinical necropsies revealed that the small intestinal walls of the PEDV-challenged piglets were thin, transparent, and filled with watery liquid (Fig. 1A). Detections of different intestinal segments showed that the jejunum contained the highest amount of virus (>1 × 10⁷ copies/g) (Fig. 1D). Ag immunofluorescence staining of the jejunum revealed that the intestinal villi were abundantly colonized with PEDV (Fig. 1E). Finally, we selected jejunum segments from four piglets in each case and control group to pool together for the scRNA-seq analysis.

Single-cell suspensions were obtained via trypsin digestion of jejunum segments from piglets in the case or control group. Single-cell Gel Beads in-emulsion were then obtained by using 10× Genomics microfluidic technology. Reverse transcriptions were performed to construct a cDNA library; the Illumina platform was used for sequencing, and further analysis was performed (Supplemental Fig. 1A). In this study, 6000 and 6774 high-quality cells were obtained from the case and control piglets, respectively. For the case and control piglets, respectively, the average number of unique molecular identifiers in each cell was 4031.06 and 6603.90, the average number of genes in each cell was 1224.65 and 1916.45, and the average proportion of mitochondrial genes in each cell was 0.1354 and 0.1649 (Supplemental Fig. 1B). The scRNA-seq dataset has been uploaded to GEO under accession number GSE175411 (https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE175411).

All the cells in case and control groups were used to define the cell types. As shown in (Fig. 2A and 2B, after the first-dimension reduction clustering, cells were divided into 18 clusters. According to known representative markers of human and mouse cell types, cells in clusters 2, 4, and 8, which highly expressed CD3D, CD3G, CD8A, and CD8B, were defined as CD8⁺ T cells. Th17 cells, differentiated from CD4⁺ T cells, specifically secrete IL17 and IL22 (30). As shown in (Fig. 2C, excepting CD3E, the specific intracellular markers IL17A, IL17F, and IL22 and extracellular markers IL1R1 and CCR6 are mainly expressed in clusters 3 and 9, so these two clusters tended to be determined as Th17 cells. Cluster 11, which highly expressed SLAMF8 and IGSF6, was defined as myeloid cells (mainly dendritic cells and macrophages). The cells in clusters 15 and 16 highly expressed the B cell markers FCRLA and MS4A1; meanwhile, cells in cluster 15 expressed the B cell maturation Ag TNFRSF17, so cluster 15 represented plasma cells and cluster 16 represented B cells. MFAP4, LHFP, and GPX7 were mainly expressed in cluster 17, which was defined as PROCR⁺ progenitor cells (Fig. 2A, 2B).

Cells in the remaining clusters (1, 5–7, 10, 12–14) highly expressed the intestinal cell markers KRT20 and CDX2 (Fig. 2D), so a second-dimension reduction cluster analysis was performed to define intestinal cell types. As shown in (Fig. 2E and 2F, the cells in clusters 1, 2, 3, 4, 14, and 15 were classified as enterocytes, while cells in clusters 5, 6, 8, 11, and 12 were identified as enterocyte progenitor cells because they highly expressed the enterocyte cell markers LCT and ANPEP and the stem cell markers SOX9 and MKI67. Cells in clusters 7 and 9 expressed the specific markers MUC2 and SPDEF of goblet cells, while cluster 7 cells also expressed the stem cell marker PROM1. Therefore, cluster 7 and cluster 9 cells were identified as goblet progenitor cells and goblet cells, respectively. The cells in cluster 10 specifically and highly expressed the enteroendocrine cell markers TRPA1 and APLP1, and cluster 13 cells highly expressed the tuft cell markers TRPM5 and RGS13, so these two clusters were identified as enteroendocrine and tuft cells, respectively.

We found that tuft cells specifically expressed DNAH11 (Fig. 2G). DNAH11 has not been previously identified as a marker of tuft cells, so we confirmed this possibility by conducting indirect immunofluorescence of intestinal tissues. As shown in (Fig. 2H, DNAH11 was specifically expressed in tuft cells with the identified cell-type marker TRPM5 (Fig. 2H). Further immunofluorescence detections of PEDV Ag revealed that, excepting porcine intestinal enterocytes, goblet and tuft cells also had the ability to infect PEDV (Fig. 2H, 2I). Meanwhile, the process of viral infection of goblet cells might be shown in (Fig. 2I, the virus first gathered at the microvilli of the cell and subsequently propagated at the base of the cell, whereas the virus was always absent at the expansion part that was rich in mucin particles. Moreover, (Fig. 2I also indicated that PEDV infected goblet cells earlier than enterocytes. All identified cell types and their specific markers are listed in Table I.

Enrichment analysis of KEGG pathways associated with the DEGs can be used to determine the biochemical metabolic and signal transduction pathways under the conditions of virus infection. The identified top 20 KEGG pathways were significantly enriched in DEGs in four intestinal cell types: enterocyte, goblet, tuft, and enteroendocrine cells were shown in (Fig. 3. The significantly up-enriched pathways in all four intestinal cell types mainly included the apoptosis and MAPK signaling pathways. The tight and adherens junction pathways were significantly increased in goblet, tuft, and enteroendocrine cells. In addition, the related genes of the two junctions were detected in the PEDV-infected IPEC-J2 cells in vitro. The results showed that tight junction–related genes claudin and ZO-1 were significantly reduced in PEDV-infected cells. However, the adherens junction–related gene cadherin was significantly increased with PEDV infection at the transcript and protein levels (Supplemental Fig. 2). What is more, the focal adhesion pathways in goblet, tuft, and enteroendocrine cells, with TNF and IL-17 signaling pathways in enterocyte and goblet cells, were all significantly increased. In the top 20 down-enriched pathways, the oxidative phosphorylation was decreased in all four intestinal cell types.

The identified KEGG pathways that were significantly enriched in immune cell types were also shown in (Fig. 4. In myeloid cells, the upregulated TNF and MAPK signaling pathways and the downregulated TCR, mRNA surveillance, and oxidative phosphorylation signaling pathways were found in the case group. In T and Th17 cells, the TCR signaling pathway and Th1 and Th2 cell differentiation were both decreased in these two cell types in the case group. What is more, the BCR signaling pathway in B cells and the intestinal immune network for IgA production in both B and plasma cells were significantly decreased with PEDV infection.

The keggmaps of the selected MAPK signaling pathway and oxidative phosphorylation of enterocyte cells, Th1 and Th2 cell differentiation in T cells, and intestinal immune network for IgA production in plasma cells were shown to exhibit the significantly changed genes. As shown in Supplemental Fig. 3A, NF-κB, ERK, MEK1 and MEK2, MEKK1, and p38, among others, were significantly upregulated in the case group compared with the control group. The oxidative phosphorylation metabolic pathway generates ATP by transport of electrons to a series of transmembrane protein complexes in the mitochondrial inner membrane, known as the electron transport chain. As shown in Supplemental Fig. 3B, the multiple ND, NDUFA, and NDUFB genes in NADH dehydrogenase, COX genes in cytochrome c oxidase, and ISP and QCR genes in cytochrome c reductase were significantly downregulated in the case of viral infection. Ag processing and presentation and the TCR signaling pathway play important roles in evoking humoral and cellular immune responses against pathogens (31). As shown in Supplemental Fig. 3C and 3D, the reduction of MHC II and TCR genes were both found in T and plasma cells, which restrict the Th1 and Th2 cell differentiation and IgA production.

Partially important chemokines were analyzed as shown in (Fig. 5A; CCL20 and CCL25 were upregulated after viral infection in all intestinal cell types, with the rise being most pronounced in goblet progenitor cells. However, CXCL8 and CXCL16 were significantly decreased in tuft cells and enteroendocrine cells with viral infection (Fig. 5B). Analyses of TNF-related genes revealed that viral infection rendered the TNFSF10 in a downregulated state in most cell types. In PROCR⁺ progenitor cells, TNFSF11 expression decreased significantly after viral infection (Fig. 5C). IL22 is predominantly expressed by Th17 cells, and its expression is markedly upregulated on viral infection. However, IL34 downregulation after viral infection was evident in tuft cells, and IL33 was upregulated in several immune cell types after viral infection (Fig. 5D).

In previous results, we identified goblet cells as another target cell for PEDV. Goblet cells secrete MUC2, the most important secretory mucin, and along with other typical mucous components, FCGBP, CLCA1, ZG16, and AGR2, play key roles in maintaining the integrity of the intestinal mucous layer. A heatmap display of the 25 most differentially expressed genes in goblet cells was generated, and MUC2, FCGBP, CLCA1, and AGR2 were significantly reduced in case piglets (Fig. 6A). The assay of fold change revealed that all four genes had a reduction of more than two times (Fig. 6B).

There is no porcine intestinal goblet cell line at present. Therefore, human colon cancer goblet cell line LS174T was selected for PEDV infection test, and the results showed that the LS174T cells could be infected by PEDV and exhibited the cell fusion phenomenon. In addition, the infection efficiency was also trypsin dependent (Fig. 6C). The growth curve showed that the virus titer reached ∼10⁵ TCID₅₀/ml at 48 hpi (Fig. 6D). These above results revealed that LS174T cells were susceptible to PEDV.

After LS174T cells were infected with PEDV, cell and supernatant samples were collected at different time points to detect MUC2 and other typical mucous components. The results showed that the content of PEDV increased with the extension of infection time (Fig. 6E). The mRNA levels of MUC2 and other mucous components FCGBP, CLCA1, and AGR2 were significantly decreased in the PEDV-infected group compared with the control group (Fig. 6F–I). Western blot analysis also showed that the level of MUC2 protein in LS174T cells decreased significantly with viral infection (Fig. 6J). Meanwhile, the MUC2 protein content in the supernatant was determined by ELISA, and the degree of decline increased significantly with the extension of virus infection time (Fig. 6K). These results indicated that PEDV infection had a significant downregulation effect on MUC2 in goblet cells.

Members of the AMP family play important roles in maintaining the homeostasis of the intestinal environment and resisting infections of pathogenic bacteria (32). The transcripts of multiple important AMPs were analyzed, and only DEFB115 and REG3G were more abundantly present in the intestinal cells of jejunum of piglets. In the control piglets, the scRNA-seq showed REG3G with high expression in enterocyte cells. Moreover, REG3G was upregulated after PEDV infection in enterocyte, enterocyte progenitor, and enteroendocrine cells (Fig. 7A). By using IPEC-J2 cells, we measured the transcript levels of AMPs after PEDV infection in vitro. As shown in (Fig. 7B, REG3G was significantly upregulated in PEDV-infected IPEC-J2 cells, which was consistent with the transcription analysis of scRNA-seq.

Stimulation of REG3G expression in intestinal epithelial cells involves multiple host factors, such as IL22, MyD88, and IL33, and the activation of mTOR and STAT3. The effects of PEDV infection on these factors were also detected. As shown in (Fig. 7C, the transcript levels of IL33 were significantly increased in PEDV infection. Western blot assays exhibited that the levels of MyD88 expression and mTOR phosphorylation were not altered. However, the phosphorylation of STAT3 was activated by virus infection, with the upregulation of REG3G expression (Fig. 7D). These results revealed high expression of REG3G in PEDV infection mostly caused by activated STAT3 by IL33, which is consistent with previous findings in mice (33).

The antiviral effect of REG3G was also detected. As shown in (Fig. 7E–G, the virus copies and titers in supernatants and the amount of PEDV N protein were all significantly decreased with REG3G overexpression in a dose-dependent manner. In addition, compared with control siRNA-transfected groups, special siRNA of REG3G-transfected groups significantly reduced the expression levels of REG3G, and the virus copies and titers in supernatants and the amount of PEDV N protein in cells were all significantly increased (Fig. 7H–J). The earlier results indicated that REG3G significantly inhibited PEDV replication. The inhibition of the PEDV replication phase by REG3G was further assayed, and the results showed that the amount of N protein expressed during the intracellular replication stage of PEDV was significantly decreased in the REG3G overexpression group compared with the control group (Fig. 7K, 7L).

Coronaviruses endanger human public health and safety, so it is important to study the cross-species transmission of these viruses. In particular, the severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) pandemic and the infectivity of coronaviruses in animals that are closely related to humans (such as cats, dogs, and swine) prompted us to analyze coronavirus-related receptors in the small intestine of pigs (34). The identified coronavirus receptors ANPEP, SIAE, TMPRSS2, NANS, ACE2, and DPP4 (35, 36), as well as OCLN, BSG, CTSL, CTSB, PCBP2, and NPM1, which play key roles in coronavirus replication, were analyzed (21, 37–40). As shown in (Fig. 8A, among intestinal cell types, enterocyte and enterocyte progenitor cells expressed almost all the known coronavirus receptors and replication-related genes, with the highest expression levels of ANPEP and ACE2 in the enterocyte cells. In immune cell types, only plasma cells highly expressed the coronavirus receptor NANS, but all immune cell types highly expressed four viral replication–related genes, BSG, CTSB, PCBP2, and NPM1. After PEDV infection, the transcript levels of coronavirus receptor and replication-related genes were changed in different cell types (Fig. 8B). In enterocytes, the expression level of ANPEP did not change after PEDV infection, whereas levels of TMPRSS2 and ACE2 increased. In enterocyte progenitor cells, the expression levels of TMPRSS2 and DPP4 decreased and that of ANPEP increased after PEDV infection. In multiple immune cell types, the replication-related genes increased significantly, such as PCBP2 and NPM1 in T and Th17 cells and CTSB and PCBP2 in B cells. Subsequently, we validated the expression of the earlier-mentioned genes in intestinal tissues and found that the expression level of ACE2 increased significantly in the PEDV-challenged piglets (Fig. 8C). In vitro assays similarly found that the expression level of ACE2 increased significantly over the course of PEDV infection by using IPEC-J2 cells (Fig. 8D).

Pig, as host of a variety of coronaviruses, plays an extremely important role in public health safety. Diarrhea caused by several different swine coronaviruses reveals that the porcine intestinal tract has its particularity, which makes it susceptible to related coronaviruses (4–6). Therefore, it is of great significance to analyze the cellular composition of the porcine intestine and the characteristics of these cell types. We carried out this study to identify the cell types in the jejunum segment of the pig small intestine. By applying identification markers for human intestinal cell types, we found that most of those specific markers were also suitable for cell-type identification in the pig small intestine, providing a convenient method for identifying the cell types. In further analyses, we screened several unique markers of different pig jejunum cell types and identified them. Finally, we confirmed that DNAH11 was one unique marker for pig tuft cells.

DNAH11 encodes a ciliary outer dynein arm protein that is a member of the dynein H chain family and localizes exclusively to the proximal region of respiratory cilia (41, 42). DNAH11 mutations have been found to result in abnormal ciliary ultrastructure and hyperkinetic ciliary beating (41). An overwhelming majority of previous studies on DNAH11 have focused on primary ciliary dyskinesia and situs inversus totalis (42). To date, no studies have reported the localization of DNAH11 protein in the intestinal tract and tuft cells, so further study on the function of DNAH11 in tuft cells is needed. In addition, in the identification of T cells, we found high expression levels of Th17-specific molecules IL17A, IL7F, and IL22 in clusters 3 and 9, but not a high expression level of T cell representative molecule CD3. This may be related to the differentiation of pig T cells. However, considering the high expression levels of specific secretory cytokine IL17, we still tended to determine clusters 3 and 9 as Th17 cells.

Intestinal goblet cells, whose main function is to secrete mucins, play an important role in the innate barrier of the gut (23). Studies have shown that another important function of goblet cells is to present Ags from the small intestinal lumen to small intestinal lamina propria CD103⁺ dendritic cells, thereby acting as an immune regulator (43–45). In this article, we identified goblet cells as target cells of PEDV. Moreover, we found that the infection initiation of virus was from the microvilli of the cell, and PEDV infected goblet cells earlier than enterocytes. All the phenomena suggested that goblet cells might play an important role in the initiation of PEDV infection, which might rely on the presentation function of this cell. Considering the important role of dendritic cells in multiple coronavirus infections (46–48), the role played by goblet cells in different enterovirus infections needs to be investigated in the future. Tuft cells, a rare type of intestinal epithelial cell, have been identified as a murine norovirus target cell in the mouse intestine (49). In this study, we also confirmed susceptibility of tuft cells to PEDV, exhibiting the tropism of coronavirus PEDV for a variety of different cell types.

Mucus secreted by goblet cells continually replenishes the mucous layer that overlies the gut epithelium as a first barrier against commensal bacteria and invading pathogens. IgA and AMPs are secreted into the mucus as a defense against pathogens and potentially harmful commensal bacteria. MUC2 glycoproteins are a major constituent of small intestinal and colonic mucus (24). The studies have confirmed that PEDV infection results in time-dependent changes in intestinal barrier integrity and function, thereby further causing piglets to develop severe secondary infection, dehydration, and death. In this study, an important finding was that infection of goblet cells with PEDV caused significant downregulation of multiple mucin components, especially MUC2. This inhibition would result in an inability of mucins to be replenished in a sustained, sufficient amount, causing the intestinal mucous layer to lose its proper integrity and function. Such findings also enrich the understanding of PEDV pathogenicity.

AMPs are a class of small peptides that distributed across all kingdoms of life, and they are an important part of the innate immune system. AMPs have a wide range of inhibitory effects against bacteria, fungi, parasites, and viruses (50). In this study, the transcripts of multiple important AMPs were analyzed, and only DEFB115 and REG3G were more abundantly present in the intestinal cells of jejunum segments of the piglets. Importantly, PEDV infection causes a significant upregulation of REG3G, which, in part, complements the anti-infection capacity of the mucous layer. In addition, the expression of REG3G involves several host cytokines and proteins, such as IL22, IL33, MyD88, STAT3, and mTOR. In this study, the significant increase of IL33 and phosphorylation of STAT3 were found in PEDV-infected IPEC-J2 cells, which indicates that the reported IL33-STAT3 signaling pathway also plays an important role in inducing REG3G expression in PEDV infection (33). What is more, the REG3G was confirmed to exhibit the antiviral effect against PEDV. REG3G has been served as a molecular switch to activate multifarious oncogenic signaling pathways, such as MAPK/p38, TGF-β/Smad, and Wnt/β-catenin (51). Several studies have reported that these pathways are involved in the replication of multiple viruses with inconsistent roles (52–55). As to which pathway REG3G inhibits PEDV replication requires further clarification. Integrity of the intestinal epithelium is maintained mainly by tight junction and adherens junction (17). Previous studies found that several components of the two junctions showed a downward trend in PEDV-infected groups both in vitro and in vivo (17, 56). In this study, the scRNA-seq also revealed the significant downregulation of tight and adherens junctions in goblet, tuft, and enteroendocrine cells. Moreover, the lower expression of tight junction components was found in PEDV-infected IPEC-J2 cells in vitro. However, the cadherin of adherens junction was upregulated in IPEC-J2 cells at the later stage of viral infection. This may be because of the different responses of different intestinal cell types to PEDV infection. Mucosal Abs play an important role in resistance to PEDV infection. However, what makes us equally impotent is that a reduction in IgA secretion by immune cells may occur in PEDV infection (Supplemental Fig. 3), which is also very likely to be one reason for the poor immunogenicity of PEDV. Collectively, PEDV exerts its infectious, pathogenic, and immunomodulatory effects through multiple ways.

CXCL16 is classified as an α subfamily chemokine. It is expressed in small intestine and several lymphoid organs, including Peyer’s patches. CXCL16 is important in immune cell accumulation at an inflammation site, mediates the adhesion of Gram-negative and Gram-positive bacteria, and leads to bacterial phagocytosis by cells expressing CXCL16, such as macrophages and dendritic cells (57). IL-34 drives macrophage differentiation, proliferation, maintenance, migration, and adhesion. In pathological situations, such as bacterial or viral infection or inflammation, IL-34 can act as a proinflammatory or antiviral/anti-inflammatory agent (58). In this study, these two cytokines were both reduced in the tuft cells in the PEDV-infected piglets. These reductions would also lead to a loss of functional integrity of macrophages and the elimination of pathogenic bacteria, thus reducing the defective intestinal immune function.

Through an analysis of coronavirus receptors in different porcine intestinal tract cell types, we found that porcine small intestinal enterocytes highly express a variety of different coronavirus receptors. This finding supports the susceptibility of pigs to coronaviruses and the intestinal symptoms associated with infection. Despite the spread of PEDV in pigs for decades and pigs being an important food source, no cases of human infection with PEDV have been reported to date. However, in agreement with previous work, we found that human Huh7 cells are susceptible to PEDV infection (data not shown) (59, 60). In addition, the newly discovered swine acute diarrhea syndrome coronavirus can replicate efficiently in several different primary human lung cell types, as well as in primary human intestinal cells (61). Moreover, the transient expression of human APN renders cells susceptible to PDCoV infection (62). Therefore, there is potential for the spread of new higher-risk zoonotic viruses to humans, which would negatively impact the global economy and human health.

In our scRNA-seq database, we did not detect the virus transcripts. In the experimental process of single-cell transcription and sequencing, compared with the amount of RNA from tissue cells, the amount of RNA from virus is very low. Coupled with the low transcriptional efficiency, we believe that these may be the possible reasons why the virus expression is not detected in this project. Nevertheless, the detection of PEDV Ag in enterocyte cells can help us judge the characteristics of virus infection. Meanwhile, for the number of repetitions, generally, compared with clinical samples, the background of animal model samples (including pigs) is relatively single, and the results will be relatively stable. Further, we used three pig samples in each group to pool together for the experiment, which can reduce individual data error. In addition, because of the detection of thousands of cells in each sample by single-cell sequencing, the p value of the difference between groups can be calculated.

In summary, this study identified different cell types in pig jejunum and analyzed the transcriptome landscapes of different types of cell in normal and PEDV-infected jejunums. The effect of PEDV infection on intestinal mucous layer was comprehensively analyzed for the first time, to our knowledge. The findings provide a deeper understanding of the pathogenic mechanism of PEDV.

We thank Song Shi (Shanghai OE Biotech. Co., Ltd) for support and help in the analysis of scRNA-seq data.

The authors have no financial conflicts of interest.