Genome-wide CRISPR screen reveals key role of sialic acids in PEDV and porcine coronavirus infections

To identify key host factors essential in PEDV replication, we screened a lentiviral-mediated genome-wide CRISPR screen in Cas9-expressing Huh7 cells that were mutagenized using a lentiviral gRNA library. For this screen, we performed three consecutive rounds of infection with a highly virulent PEDV (GDU strain, GIIb type) at an MOI of 0.1. Surviving cells were collected and single-guide RNA (sgRNA) sequences were PCR amplified and subjected to next-generation sequencing (NGS) (Fig. 1A). The top 10 ranked sgRNAs were identified by calculating the log2 fold change between control and surviving mutant cells. The top 10 most enriched genes after the third round of PEDV infection included SLC35B2, DHCR7, ST3GAL4, NDST1, DIPK2A, FOXA2, and PATZ1 (Fig. 1B). To identify the function of candidate PEDV-resistance genes, we conducted KEGG pathway enrichment analysis for top 50 ranked sgRNA targets. The analysis showed significant enrichment of genes related to sialic acid and heparan sulfate biosynthesis, as well as ferroptosis, following PEDV selection (Fig. 1C). Based on the results of our screen, we selected four genes for further validation, prioritizing them according to their rankings in the analysis. Porcine kidney proximal tubular epithelial cells (LLC-PK1) stably expressing Cas9 protein were used to construct KO cell lines. Polyclonal KO cells were then infected with PEDV at an MOI of 0.1 on day 9 post-sgRNA transduction. The results showed that KO of ST3GAL4 greatly inhibited PEDV infection (Fig. 1D). Moreover, we found that ST3GAL4 KO reduced both virus titers and viral RNA copies (Fig. 1E and F).

Since ST3GAL4 KO cells exhibited the strongest inhibition of PEDV replication, we constructed a clonal ST3GAL4 KO LLC-PK1 cell line. A four-base pair deletion was identified at the sgRNA binding region compared to the wild-type (WT) sequence, verifying the KO of the ST3GAL4 gene in the cell line (Fig. 2A). To determine whether ST3GAL4 is required for PEDV replication, we constructed a ST3GAL4-KO-complement LLC-PK1 cell line, which stably expressed the ST3GAL4 gene (Fig. S1). Western blot analysis using a Flag-tag antibody confirmed stable expression of ST3GAL4 in the complemented cells (Fig. 2B). No significant difference was observed in cell viability among ST3GAL4 KO, complement, and WT cells (Fig. 2C). Knocking out ST3GAL4 significantly reduced virus titers after infection, while PEDV replication was fully restored upon ST3GAL4 complementation (Fig. 2D through F). These results demonstrate that ST3GAL4 is an important host factor for PEDV infection.

Among the sialyltransferase proteins, ST3GAL4 catalyzes the transfer of sialic acid to sugar chains through α2,3-linked linkages. To assess which subtypes of sialic acid have high affinity for PEDV infection, we constructed KO cells of sialic acid synthesis-related genes (SLC35A1 and ST6GAL1) and ST3GAL4/ST6GAL1 double KO LLC-PK1 cells (ST3GAL4/ST6GAL1^DKO) (Fig. 3A). Maackia amurensis lectin II (MAL II) specifically binds to α2,3-linked sialic acid, and Sambucus nigra lectin (SNA) preferentially binds α2,6-linked sialic acid. Lectin binding analysis revealed a significant decrease in the synthesis of α2,3-linked sialic acid between SLC35A1, ST3GAL4, and ST3GAL4/ST6GAL1^DKO cells, and KO SLC35A1, ST6GAL1, and ST3GAL4/ST6GAL1^DKO had a strong effect on the synthesis of α2,6-linked sialic acid (Fig. 3B). To evaluate PEDV infection, we performed immunofluorescence assays targeting the PEDV-S1 and TCID₅₀ assays for both WT and KO cells. Compared to control cells, all KO cell lines (ST3GAL4, SLC31A1, ST6GAL1, and ST3GAL4/ST6GAL1^DKO) showed significantly reduced PEDV infection. Notably, the SLC35A1 KO and ST3GAL4/ST6GAL1^DKO cells demonstrated the strongest inhibitory effects (Fig. 3C and D).

To further examine the role of α2,6-linked sialic acid, we constructed SLC35A1 and ST6GAL1 complementation cell lines, which stably expressed the SLC35A1-Flag and ST6GAL1-Flag gene (Fig. S1B and C). Western blot analysis using the Flag-tag antibody confirmed successful expression (Fig. 3E). CCK-8 kit assay determined that neither KO nor complementation of SLC35A1 or ST6GAL1 significantly affected cell viability (Fig. S2A and B). Consistent with results from ST3GAL4 complementation, both SLC35A1 and ST6GAL1 complementation restored sensitivity to PEDV infection (Fig. 3F).

To assess whether combined inhibition enhances the antiviral effect, WT cells were pretreated with 3Fax-Neu5Ac, a specific sialyltransferase inhibitor. Cell viability was unaffected by treatment with 100 µM 3Fax-Neu5Ac (Fig. S3A). Compared to the control group, pretreatment with 3Fax-Neu5Ac significantly reduced PEDV replication in a dose-dependent manner (Fig. S3B). Furthermore, WT and KO-SLC35A1 cells were pretreated with 3FAx-Neu5Ac. In WT cells, 100 µM 3FAx-Neu5Ac significantly inhibited the PEDV replication compared to DMSO controls. Interestingly, PEDV replication was further decreased in KO-SLC35A1 cells pretreated by 3FAx-Neu5Ac compared to DMSO-treated KO-SLC35A1 cells, suggesting an enhanced antiviral effect from the combination of genetic KO and chemical inhibition (Fig. 3G and H). Collectively, these results indicated that PEDV relies on both α2,6-linked and α2,3-linked sialic acid for efficient cell entry and infection.

Our studies confirmed the importance of sialic acid in PEDV infection. To evaluate its role in the viral infection cycle, we performed virus attachment and internalization assays. LLC-PK1 (WT), KO-SLC35A1, KO-ST3GAL4, KO-ST6GAL1, and ST3GAL4/ST6GAL1^DKO cells were incubated with PEDV for 1 h at 4°C to allow virus absorption. The attachment efficiency was detected by real-time fluorescence quantitative PCR (qRT‒PCR). Compared to WT cells, KO cell lines showed a significant reduction in virus attachment (Fig. 4A). In the internalization assay, the cells were washed three times with acidic PBS (pH = 1.3) at 4°C to remove noninternalized virus particles. The assay showed that viral internalization was also reduced in cells lacking sialic acid synthesis-related genes (Fig. 4B). These findings were corroborated using PEDV-S-pseudovirus. KO of α2,3-linked or α2,6-linked sialic acid synthesis in cells also impaired PEDV-S-pseudovirus attachment and internalization (Fig. 4C and D).

To assess the sialoglycan specificity of PEDV, we performed a glycan microarray screening using the PEDV S1 protein. Among all types of glycans screened, we find that α2,3-linked and α2,8-linked sialosides were most favored by PEDV S1. In particular, the S1 protein preferred binding to sialylated glycans containing terminal α2,8-Neu5Ac-α2,3-Neu5Ac epitope with or without internal α2,3-Neu5Ac or α2,8-Neu5Ac-α2,3-Neu5Ac motifs (Fig. 5). However, terminal α2,3-linked sialosides exhibited relatively weaker or negligible binding in the glycan array, contrasting with our cell-based assays where they serve as key attachment factors for PEDV. This discrepancy may be due to the ability of low-affinity glycans to engage in multivalent interactions on the cell surface, thereby enhancing virus binding and leading to biologically relevant interactions not captured in the array format. Notably, unlike other forms of sialylation, α2,8-linkage is more protein specific and only found on a few glycoproteins in mammalian intestinal cells (30, 31). Hence, α2,3-linked sialic acid serves as preferred attachment factors for PEDV in vivo.

Sialic acid has been shown to enhance the infection of a variety of viruses, leading us to speculate that sialic acid may also play a significant role in other porcine enteric CoVs. Therefore, we investigated the necessity of sialic acid for infection of major epidemic PEDV strains (genotype GII-b and GII-c) as well as PDCoV, transmissible gastroenteritis virus (TGEV), and swine acute diarrhea syndrome coronavirus (SADS-CoV). Immunofluorescence assay (IFA) and viral titration were performed to detect CoV infection in both WT cells and KO cells. Compared to WT cells, infection of these viruses was significantly reduced in cells lacking ST3GAL4 and ST6GAL1 (Fig. 6). These results demonstrate that sialic acid, particularly in α2,3-linked and α2,6-linked forms, is essential for the efficient infection of cells by multiple porcine enteric coronaviruses.

Human embryonic kidney 293 cells stably expressing the SV40 large T antigen (HEK-293T), African green monkey kidney epithelial cells (Vero), Porcine kidney proximal tubular epithelial cells (LLC-PK1), and swine testicular (ST) cells were purchased from ATCC. All cells were maintained in Dulbecco's modified Eagle medium (Gibco, Shanghai, China) supplemented with 10% heat-inactivated fetal bovine serum and 1% penicillin/streptomycin (Cat#15140122, Gibco, Shanghai, China) were incubated at 37°C with 5% CO₂.

The following viruses were used: the virulent PEDV strain GDU (GenBank accession number KU985230↗), the PDCoV strain (GenBank accession number MF095123.1↗), the TGEV strain WH-1 (GenBank accession number HQ462571.1↗), and SADS-GFP was kindly provided by Dr. Yao-Wei Huang (42). PEDV strain (GII-b: PP472643.1 and GII-c: PP472642.1) were isolated from our laboratory. The PEDV and SADS-GFP strain was propagated in Vero cells in the presence of 10 µg/mL trypsin, and the PDCoV strain was propagated in LLC-PK1 cells in the presence of 5 µg/mL trypsin.

Mouse anti-PEDV-S1, Flag tag, PDCoV-S1, and TGEV-S1 monoclonal antibodies were prepared in our laboratory. Antibodies and dyes were purchased, including Alexa Fluor 488 Donkey anti-Mouse IgG (antGene, Wuhan, China), DAPI and Hoechst (Beyotime, Shanghai, China), FITC-labeled SNA and FITC-labeled MAL II lectins (Vector Lab, Burlingame, USA).

Genome-wide sgRNA library was generated by ±260,000 unique sgRNA in Huh7-Cas9 cells as described before (43). The coverage of genome-wide sgRNA library was >400-fold (44). For the genome-wide CRISPR screening assays, ~100 million Huh7-Cas9 library cells were infected with PEDV at an MOI of 0.1. At 21 days after infection, the surviving cells were expanded for NGS analysis and the next round of infection.

Lentiviral vector, pMD2.G,, and psPAX2 plasmid were co-transfected in HEK 293T cells via jetPRIME transfection reagent (Polyplus, Paris, France) according to the manufacturer's instructions. After 48 h postinfection (hpi) of transfection, the supernatant was collected and stored at −80°C.

All the sgRNAs used in this experiment were designed according to the dedicated website (http://crispor.tefor.net/crispor.py↗). The sgRNA sequence targeting the target genes was cloned and inserted into the LentiGuide-Puro vector to produce the recombinant lentivirus. LLC-PK1-Cas9 cells were infected with LentiGuide-Puro lentivirus, and puromycin (2.5 µg/mL) was added to select the positive cells at 24 hpi. We used the serial dilution method to select monoclonal cells. All primers used are listed in Table S1.

A colorimetric-based cell counting kit-8 (CCK-8) assay (Dojindo Molecular Technologies, Rockville, MD, USA) was used to determine cell viability. In brief, control and KO cells were seeded in 96-well plates, and then 10 µL of CCK-8 reagent was added to each well at 12, 24, and 36 h. The plates were incubated in a 37°C constant temperature incubator for 4 h, and the absorbance at 450 nm was measured with a microplate reader.

The stock virus was serially diluted to inoculate confluent Vero cell monolayers and grown in 96-well plates. The cell monolayer was washed three times with PBS before infection with 100 µL of the virus. Eight replicate wells were inoculated for each dilution, and the plates were incubated at 37°C with 5% CO₂ for 2 days. The viral titers were calculated by using the Reed-Muench method and are expressed as TCID₅₀ per milliliter.

The cells were gently washed in PBS and then fixed with 4% paraformaldehyde for 10 min at room temperature. This was followed by a permeabilization step using ice-cold 0.1% Triton X-100 was added for 15 min at room temperature. Then, the samples were washed three times with PBS, and the cells were then blocked with 3% bovine serum albumin (BSA) in PBS for 30 min at room temperature. Afterward, the cells were incubated with an anti-PEDV spike antibody in PBS for 1 hpi at 37°C, followed by three washes with PBS. Finally, the cells were incubated with Alexa 488-labeled anti-human antibody for 45 min at 37°C, washed in PBS, and treated with DAPI at room temperature for 5 min to stain the nuclei.

Cells were lysed by RIPA and NP40 lysis buffer (Beyotime, Shanghai, China) containing 1 mM phenylmethanesulfonyl fluoride (Beyotime, Shanghai, China), then denaturation at 95°C for 5 min. The protein samples were resolved by 10% sodium dodecyl sulfate-polyacrylamide gel electrophoresis and transferred to polyvinylidene difluoride membranes (Millipore, USA) that were blocked with PBS containing 5% skim milk at room temperature for 2 h. Membranes were incubated in TBST containing primary antibodies overnight at 4°C. On the following day, membranes were washed and incubated with horseradish peroxidase-conjugated anti-mouse secondary antibodies for 1 h at room temperature. Immunoreactive bands were visualized using the ChemiDoc MP Imaging System (Bio-Rad, USA).

Viral RNA from cell suspensions was extracted using the Viral RNA Extraction kit (TaKaRa Bio, Japan) following the manufacturer's instructions. The cDNA was obtained by RT‒PCR using HiScript III 1st Strand cDNA Synthesis Kit (Vazyme #R312, China). The absolute qRT‒PCR assay for quantifying the PEDV genome was performed with RealUniversal SYBR Green Premix (Tiangen, Beijing, China) following the manufacturer's instructions. Following the program: one cycle of 30 s at 95°C, followed by 45 cycles of 5 s at 95°C and 30 s at 60°C. PEDV N protein coding sequence was cloned and inserted into the pMD19-T vector as a standard reference for the quantification of PEDV copy numbers. Table S1 lists primers utilized in qRT‒PCR (45).

The cells were washed in PBS, then fixed with 4% formaldehyde for 10 min, washed with PBS two times and incubated with 20 µg/mL FITC-labeled lectins for 1 hpi on ice, washed in PBS and treated with Hoechst at room temperature for 5 min to stain the nuclei. A microscope (EVOS FL, Thermo Fisher Scientific, USA) was used to obtain images.

Virus attachment and internalization were assayed by qRT‒PCR analysis. For the attachment assay, plated cells were washed three times with ice-cold PBS, incubated with PEDV-GDU (MOI = 10) at 4°C for 1 hpi, and the cells were washed with ice-cold PBS to remove unbound virus. For the internalization assay, plated cells were washed three times with ice-cold PBS, incubated with PEDV-GDU (MOI = 10) at 4°C for 1 hpi, and then supernatant was removed. The cells were transferred to 37 ℃ for 30 min that allowed virus entry, then washed three times with acidic PBS (pH = 1.3) at 4°C to remove noninternalized particles.

The ST-tagged pA-LS protein was expressed in HEK293F cells. Briefly, the plasmids were transfected using polyethylenimine reagent and the medium was replaced with 293 SFM II-based expression medium (Gibco Life Technologies) 6 h posttransfection. After 5–6 days, the culture supernatants were collected and the expressed proteins were purified via StrepTactin Sepharose beads (IBA) according to the manufacturer's instruction. The Fc-tagged PEDV-S1 protein was expressed in GnTI (N-acetylglucosaminyltransferase I)-KO HEK-293T cell lines, which were generated as described previously (46). The fusion proteins were purified using pA affinity chromatography (17-0780-01; GE Healthcare) following the manufacturer's instructions.

The synthetic ganglioside glycans 1–24 (100 µM) were dissolved in sodium phosphate buffer (pH 8.5, 250 mM) and printed in replicates of six with spot volume ~400 pl at 55% humidity (47). The slides were incubated overnight in a saturated NaCl chamber (an incubation chamber with 75% relative humidity, maintained by saturated NaCl-soaked cotton), followed by quenching with 50 mM ethanolamine in a Tris buffer (pH 9.0, 100 mM). The blocked slides were dried by centrifugation and kept in a desiccator at room temperature prior to use. Microarray slides were blocked with TSM blocking buffer (20 mM Tris Cl, pH 7.4, 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, 0.1% Tween-20, and 5% skimmed milk). Purified pA-LS and PEDV-S1-Fc were uniformly mixed according to a molar ratio of 1:1 and incubated for 30 min on ice, and then empty pA-LS nanoparticles or nanoparticles complexed with PEDV-S1-Fc were incubated for 120 min at 4°C in TSM binding buffer (20 mM Tris Cl, pH 7.4, 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, 0.05% Tween, and 1% BSA) on the microarray slides. After two repeated washes with TSM washing buffer (20 mM Tris-Cl, 150 mM NaCl, 2 mM CaCl₂, 2 mM MgCl₂, and 0.05% Tween) and TSM buffer (20 mM Tris-Cl, pH 7.4, 150 mM NaCl, 2 mM CaCl₂, and 2 mM MgCl₂) respectively, the subarrays were incubated with 100 µL StrepMAB-Classic DY-649 (diluted 1:200 in TSM binding buffer) and washed with TSM washing buffer, TSM buffer, and deionized water. The resulting array slides were scanned for fluorescence on an InnoScan 710 Microarray Scanner. The data were processed with Mapix software and further analyzed using IBM SPSS 19 (SPSS Inc.). Relative fluorescence values (RFUs) were calculated by averaging four independent replicates (corrected for mean background) on the glycan array after removal of the highest and lowest signals.

All the statistical analyses, except for the genetic screening and single-cell sequencing data, were performed using GraphPad Prism software. Two-tailed unpaired t-tests were used to compare two unpaired groups. The data were analyzed for normality and log (normal distribution) using the Kolmogorov–Smirnov test before analysis of variance was applied. All analyses were performed at a threshold α level of 0.05. The specific statistical tests used for each data set are mentioned in the respective figure captions, as well as any data transformation that was applied. All measurements were taken from distinct samples.