Construction of a Recombinant Porcine Epidemic Diarrhea Virus Encoding Nanoluciferase for High-Throughput Screening of Natural Antiviral Products

Porcine epidemic diarrhea virus (PEDV) is the predominant cause of an acute, highly contagious enteric disease in neonatal piglets. There are currently no approved drugs against PEDV infection. Here, we report the development of a nanoluciferase (NLuc)-based high-throughput screening (HTS) platform to identify novel anti-PEDV compounds. We constructed a full-length cDNA clone for a cell-adapted PEDV strain YN150. Using reverse genetics, we replaced the open reading frame 3 (ORF3) in the viral genome with an NLuc gene to engineer a recombinant PEDV expressing NLuc (rPEDV-NLuc). rPEDV-NLuc produced similar plaque morphology and showed similar growth kinetics compared with the wild-type PEDV in vitro. Remarkably, the level of luciferase activity could be stably detected in rPEDV-NLuc-infected cells and exhibited a strong positive correlation with the viral titers. Given that NLuc expression represents a direct readout of PEDV replication, anti-PEDV compounds could be easily identified by quantifying the NLuc activity. Using this platform, we screened for the anti-PEDV compounds from a library of 803 natural products and identified 25 compounds that could significantly inhibit PEDV replication. Interestingly, 7 of the 25 identified compounds were natural antioxidants, including Betulonic acid, Ursonic acid, esculetin, lithocholic acid, nordihydroguaiaretic acid, caffeic acid phenethyl ester, and grape seed extract. As expected, all of the antioxidants could potently reduce PEDV-induced oxygen species production, which, in turn, inhibit PEDV replication in a dose-dependent manner. Collectively, our findings provide a powerful platform for the rapid screening of promising therapeutic compounds against PEDV infection.


Introduction
Porcine epidemic diarrhea virus (PEDV), a member of swine enteric coronaviruses (CoVs), causes diarrhea, vomiting, anorexia, and dehydration in neonatal piglets [1]. After PEDV classical strains of the genogroup I (GI) were first reported in the UK in 1971 [2], it has spread throughout most European and Asian countries [3,4]. In 2010, highly pathogenic

Construction of a Full-Length cDNA Clone
The full-length cDNA clone of the PEDV YN150 strain was constructed using a unidirectional molecular clone strategy, which was previously described for the PC22A strain of PEDV [26]. Briefly, the full-length PEDV cDNA clone was divided into six contiguous fragments that could be systematically linked by five BsmBI restriction sites at nucleotide positions 6055, 9880, 14880, 19216, and 23,497. A T7 promoter sequence and 22 A residues were added to the viral genome's immediate 5 end and 3 end. Five naturally occurring BsmBI sites at nucleotide positions 376, 787, 2842, 11564/11566, and 23,882 were removed by introducing synonymous mutations to ensure the appropriate assembly of the full-length infectious clone. Six cDNA fragments (F1-6) were synthesized and cloned into the pJET1.2/blunt cloning vector (Thermo Scientific, Waltham, MA, USA). After propagation in Escherichia coli, all PEDV fragments were sequenced to ensure their sequence fidelity. These fragments were digested using restriction enzymes, separated through 0.8% agarose gels, excised, and purified using a QIAquick gel extraction kit (Qiagen, Hilden, Germany). The digested fragments were ligated overnight at 4 • C using T4 DNA ligase (Thermo Scientific), extracted with phenol/chloroform, and precipitated with isopropanol. Full-length PEDV RNA was generated by in vitro transcription using mMESSAGE mMACHINE T7 Transcription Kit (Ambion, Austin, TX, USA) according to the manufacturer's instruction. SP6 PEDV N gene transcripts were generated from the PCR-purified PEDV N gene sample using a 4:1 ratio of cap to GTP (Ambion).
To construct the PEDV-NLuc with the ORF3 deletion, ORF3 in the PEDV-F6 fragment was replaced with synthesized NLuc gene using fusion PCR. PEDV-F6-ORF3-NLuc was amplified in Escherichia coli and sequenced to guarantee seamless substitution of ORF3 with NLuc and the retainment of the transcription regulatory sequence (TRS) of ORF3.

In Vitro Transfection
The 30 µL of the full-length PEDV RNA and 10 µL of N RNA transcripts were together transfected into Vero cells (8.0 × 10 6 cells/mL) in phosphate-buffered saline (PBS), and three electrical pulses of 450 V at 50 µF were given by a Gene Pulser X cell electroporation system (Bio-Rad, Hercules, CA, USA). The transfected cells were seeded into a 75 cm 2 flask and incubated at 37 • C. After 12 h, the cells were washed with PBS and incubated in DMEM containing 10 µg/mL trypsin. Rescued viruses were harvested and further purified by plaque assay.

Virus Genome Sequencing
Viral RNAs were extracted from the supernatant of PEDV-infected Vero cells using TRIzol Reagent (Invitrogen, Waltham, MA, USA). Viral genome RNA was converted to cDNA by Transcriptor First Strand cDNA Synthesis Kit (Roche, Basel, Switzerland). The full-length genome of rescued PEDVs was sequenced by conventional PCR with 28 pairs of primers, listed in Table S1.

Luciferase Assay
Vero cells seeded in 96-well plates were infected with rPEDV-NLuc at the indicated viral amount. At 20 h post-infection, infected cells were lysed by Passive Lysis Buffer (Promega, Madison, WI, USA) and assayed for luciferase activity using the Nano-Glo Luciferase Assay System (Promega) according to the manufacturer's instructions.

HTS of a Natural Product Library
A library of 803 natural products was purchased from Selleck Chemicals (Houston, TX, USA). Compounds were stored in 10 mM stock solution in DMSO at −80 • C until use. The primary HTS assay was performed in Vero cells. Briefly, Vero cells were seeded into 96-well plates in DMEM containing 10% FBS. After 12 h, 1 µL of each compound was added to the cell cultures in plates at a final concentration of 10 µM (one compound per well) in triplicate; 1 µL of DMSO was used for the controls. After 1 h incubation, the cells were washed three times and inoculated with 0.01 MOI rPEDV-NLuc diluted in 100 µL/well maintenance medium containing 10 µM compound and 10 µg/mL trypsin. After 20 h incubation, the luciferase activity in infected cells was determined using the Nano-Glo Luciferase Assay System (Promega). The primary candidates were identified using the criteria of no apparent cytotoxicity and an average >90% inhibition in triplicate wells. The criteria of dose-dependent inhibition and cell viability of >80% were used to reconfirm the candidates. Moreover, the compound-specific toxicity (CC 50 ) was calculated using GraphPad Prism 6, and compounds with an SI > 10 were considered hits in this study. Cytotoxicity was assessed by the 3-(4,5-dimethyl-2-thiazolyl)-2,5-diphenyl-2H-tetrazolium bromide (MTT) assay.

Detection of ROS Production
The intracellular ROS production was examined using the fluorescent probe 6-carboxy-2 , 7 -dichlorodihydrofluorescein diacetate (DCFH-DA, Beyotime Biotechnology, Shanghai, China). Following PEDV infection, Vero cells were washed with PBS and then treated with 10 µM DCFH-DA for 20 min at 37 • C. After treatment, cells were washed twice in PBS and subjected to fluorescence microscopy observation and FACS analysis.

Construction of the Full-Length cDNA Clone of PEDV
Recently, the reverse genetics system for the PEDV PC22A strain has been successfully developed based on the unidirectional assembly of a full-length genome cDNA from a set of consecutive shorter cDNA fragments [26]. Here, we employed a similar strategy to construct the full-length cDNA clone of PEDV YN150. The complete viral genome was divided into  Figure 1A). Five naturally occurring BsmBI sites at nucleotide positions 379, 790, 2834, 11,557, and 23,886 located at F1, F3, and F6 were removed by silent mutations but not changing protein products. Notably, BsmBI is a type-IIS restriction endonuclease that cleaves outside their recognition site and leaves unique 4-nucleotide overhangs, allowing for the systematic, efficient, and directional assembly of the six smaller cDNA fragments into the full-length PEDV genomic cDNA by in vitro ligation. A T7 promoter sequence and 22 A residues were added to the immediate 5 end and 3 end of the viral genome ( Figure 1A) for in vitro transcription and polyadenylation of RNA transcripts.

Construction of the Full-Length cDNA Clone of PEDV
Recently, the reverse genetics system for the PEDV PC22A strain has been successfully developed based on the unidirectional assembly of a full-length genome cDNA from a set of consecutive shorter cDNA fragments [26]. Here, we employed a similar strategy to construct the full-length cDNA clone of PEDV YN150. The complete viral genome was divided into six contiguous fragments (F1-6) that could be systematically joined by five BsmBI restriction sites at nucleotide positions 6055, 9880, 14,880, 19,216, and 23,497 ( Figure  1A). Five naturally occurring BsmBI sites at nucleotide positions 379, 790, 2834, 11,557, and 23,886 located at F1, F3, and F6 were removed by silent mutations but not changing protein products. Notably, BsmBI is a type-IIS restriction endonuclease that cleaves outside their recognition site and leaves unique 4-nucleotide overhangs, allowing for the systematic, efficient, and directional assembly of the six smaller cDNA fragments into the full-length PEDV genomic cDNA by in vitro ligation. A T7 promoter sequence and 22 A residues were added to the immediate 5′ end and 3′ end of the viral genome ( Figure 1A) for in vitro transcription and polyadenylation of RNA transcripts.

Recovery, Identification, and Characterization of rPEDV
Each PEDV fragment was digested, purified, and ligated to generate a full-length viral cDNA genome. PEDV transcripts were then synthesized with the T7 RNA polymerase using the full-length cDNA as a template. Since the supplement of N gene transcript can facilitate the replication efficiency of several CoVs, such as MHV, TGEV, SARS-CoV, and MERS-CoV [27][28][29][30], capped PEDV-N gene transcripts were mixed with the full-length transcripts, and co-electroporated into Vero cells. Apparent cytopathic effects (CPE) were observed within 30 h post-transfection, and the expression of the S protein could be detected by IFA using anti-PEDV S mAb ( Figure 1B). After plaque purification, the complete genome of rPEDV was sequenced. We observed that the whole genome sequence of rPEDV was identical to the cDNA clone with no unwanted mutation, including the five genetic markers, as shown in Figure 1C. These results indicated the successful rescue of rPEDV in Vero cells.

Construction and Rescue of rPEDV-NLuc
To express NLuc from the PEDV genome, the ORF3 gene in the F6 fragment was replaced with that of NLuc ( Figure 2A). Transcription regulatory sequences (TRS) of ORF3 and E were preserved for initiating subgenomic RNA expression ( Figure 2A). After in vitro ligation and transcription, the full-length transcripts were electroporated into Vero cells for rPEDV-NLuc recovery. Similar to the wild-type PEDV, detectable CPE caused by rPEDV-NLuc was readily observed within 30 h post-transfection ( Figure 2B). The expression of S and N proteins was also detected in Vero cells infected with the wild-type PEDV and rPEDV-NLuc by IFA and Western blotting ( Figure 2B,C). After plaque purification, we sequenced the whole genome of rPEDV-NLuc and confirmed that the ORF3 was successfully replaced by NLuc in rPEDV-NLuc without unwanted nucleotide change compared to its cDNA clone ( Figure 2D). Notably, we observed that rPEDV-NLuc retained the inserted NLuc gene up to five passages in Vero cells (data not shown). To investigate whether the presence of NLuc affects viral replication, we determined the viral plaque size and growth kinetics of rPEDV-NLuc. As shown in Figure 2E, rPEDV-NLuc produced similar plaque morphology with the wild-type PEDV and rPEDV. Titers of rPEDV-NLuc at the different time points post-infection were also similar to those of the wild-type PEDV and rPEDV ( Figure 2F). These results suggested that the replacement of ORF3 with exogenous NLuc in PEDV YN150 did not affect viral proliferation in Vero cells.

Assessment of rPEDV-NLuc Replication via Luciferase Assay
To further confirm the expression of NLuc, we examined the bioluminescent signal in cells infected with rPEDV-NLuc. As shown in Figure 3A, plaques produced by rPEDV-NLuc could be easily visualized as luminescent signals using the In Vivo Imaging System (IVIS). Bioluminescence signals of Vero cells infected with rPEDV-NLuc were also enhanced with the increased MOI ( Figure 3B). To investigate whether the level of NLuc activity could accurately reflect the extent of viral replication, we detected the NLuc activity in cell lysates infected with rPEDV-NLuc at different times post-infection. The NLuc activity could be measured as early as 4 h post-infection and increased until 28 h post-infection ( Figure 3C), which has similar dynamics of rPEDV-NLuc growth determined by TCID 50 as presented in Figure 2F. Additionally, an increased amount of rPEDV-NLuc infection could result in elevated NLuc activity in a dose-dependent manner. A linear log RLU-log PFU correlation was observed in the cells infected with rPEDV-NLuc ( Figure 3D). These data indicated that the level of viral replication could be effectively monitored by measuring the NLuc activity in the Vero cells infected with rPEDV-NLuc.

Assessment of rPEDV-NLuc Replication via Luciferase Assay
To further confirm the expression of NLuc, we examined the bioluminescent signal in cells infected with rPEDV-NLuc. As shown in Figure 3A, plaques produced by rPEDV-NLuc could be easily visualized as luminescent signals using the In Vivo Imaging System tivity could be measured as early as 4 h post-infection and increased until 28 h post-infection ( Figure 3C), which has similar dynamics of rPEDV-NLuc growth determined by TCID50 as presented in Figure 2F. Additionally, an increased amount of rPEDV-NLuc infection could result in elevated NLuc activity in a dose-dependent manner. A linear log RLU-log PFU correlation was observed in the cells infected with rPEDV-NLuc ( Figure  3D). These data indicated that the level of viral replication could be effectively monitored by measuring the NLuc activity in the Vero cells infected with rPEDV-NLuc.

Screening of a Natural Product Library for Inhibitors of PEDV Infection
Using the rPEDV-NLuc reporter virus, we optimized the HTS assay conditions to be 10,000 Vero cells/well in 96-well plates and rPEDV-NLuc infection at an MOI of 0.01. Under this condition, the coefficient of variation (CV) and Z factor were 2.5% and 0.94, respectively, demonstrating that the assay was robust and suitable for HTS.

Screening of a Natural Product Library for Inhibitors of PEDV Infection
Using the rPEDV-NLuc reporter virus, we optimized the HTS assay conditions to be 10,000 Vero cells/well in 96-well plates and rPEDV-NLuc infection at an MOI of 0.01. Under this condition, the coefficient of variation (CV) and Z factor were 2.5% and 0.94, respectively, demonstrating that the assay was robust and suitable for HTS.
A flowchart of the HTS is depicted in Figure 4A. In the primary screening from the 803 compounds library at a concentration of 10 µM, 63 hits were found to significantly inhibit rPEDV-NLuc replication with 90% reduced NLuc activity and no apparent cytotoxicity ( Figure 4B). Among these prime candidates, 25 compounds were selected based on their dose-dependent inhibition and a selective index (SI) > 10 ( Figure S1, Table 1). To verify the results obtained by the luciferase reporter assays, we confirmed the antiviral effect of the 25 hits in Vero cells using wild-type PEDV. Consistent with the HTS results, all 25 compounds significantly reduced PEDV replication with at least a 1-log-unit decrease in the viral titers at the concentration of 10 µM ( Figure 4C). toxicity ( Figure 4B). Among these prime candidates, 25 compounds were selected based on their dose-dependent inhibition and a selective index (SI) > 10 ( Figure S1, Table 1). To verify the results obtained by the luciferase reporter assays, we confirmed the antiviral effect of the 25 hits in Vero cells using wild-type PEDV. Consistent with the HTS results, all 25 compounds significantly reduced PEDV replication with at least a 1-log-unit decrease in the viral titers at the concentration of 10 μM ( Figure 4C).   Of note, we observed that 7 of the 25 identified compounds had been reported to possess antioxidant activity, including Betulonic acid (BA) [31], Ursonic acid (UA) [32], esculetin [33], lithocholic acid (LCA) [34], nordihydroguaiaretic acid (NDGA) [35], caffeic acid phenethyl ester (CAPE) [36] and grape seed extract (GSE) [37]. To further verify the antiviral activities of the natural antioxidants, we examined whether the natural antioxidants could inhibit PEDV replication in a dose-dependent manner via TCID50 and IFA. Compared with the DMSO group, all seven antioxidants significantly inhibited PEDV proliferation at 2.5, 5, and 10 µM concentrations, and the dose-dependent inhibition of viral replication were also observed in the cells treated with all antioxidants (Figure 5A,B). Notably, all seven antioxidants at 2.5 µM are efficient in reducing PEDV replication with at least a 60% decline in viral load, suggesting that these antioxidants exhibited the potent inhibition of PEDV proliferation.

Seven Natural Products Inhibit PEDV-Induced Reaction Oxygen Species (ROS)
Recent studies showed that PEDV infection induces ROS accumulation in Vero cells [38]. We further explored whether the seven natural antioxidants can suppress the PEDV-induced ROS production. Using fluorescence microscopy, we observed that PEDV infection dramatically increased DCFH-DA fluorescence as expected. In contrast, only a few fluorescence signals were detected in PEDV-infected cells when treated with BA, UA, esculetin, LCA, NDGA, CA, and GSE, respectively ( Figure 6A). To further investigate the effect of the natural antioxidants on PEDV-induced ROS, we employed flow cytometry to detect the ROS levels in PEDV-infected cells after the treatment of antioxidants. As shown in Figure 6B, ROS production induced by PEDV was significantly decreased in the cells treated with these natural antioxidants.  tion dramatically increased DCFH-DA fluorescence as expected. In contrast, only a few fluorescence signals were detected in PEDV-infected cells when treated with BA, UA, esculetin, LCA, NDGA, CA, and GSE, respectively ( Figure 6A). To further investigate the effect of the natural antioxidants on PEDV-induced ROS, we employed flow cytometry to detect the ROS levels in PEDV-infected cells after the treatment of antioxidants. As shown in Figure 6B, ROS production induced by PEDV was significantly decreased in the cells treated with these natural antioxidants.

Discussion
PEDV has been considered one of the most devastating porcine viruses that have emerged or re-emerged in recent decades, causing severe economic losses in the global pork industry. Even though both inactivated and live-attenuated vaccines have been widely used to control the prevalence of PEDV in China and other counties, the PEDV outbreak is still ongoing, and novel PEDV strains were continually detected in vaccinated pig herds [3,8]. The development of effective antiviral agents is another strategy to control PEDV infection. Salinomycin [39] and remdesivir [40] have been reported as potential antiviral drugs against PEDV infection in vitro. Several natural products extracted from plants, such as glycyrrhizin [16], quercetin 7-rhamnoside [15], and epigallocatechin-3gallate [41], have been demonstrated to reduce PEDV infection in Vero cells. To search for new anti-PEDV agents, Deejai et al. screened the FDA-approved drugs that could associate with the PEDV N protein using virtual screening and found that trichlormethiazide, D-(+) biotin and acetazolamide could inhibit PEDV replication in vitro [42]. Recently, based on the assessment of CPE and viral protein expression by IFA, Wang et al. screened a natural product library and demonstrated that tomatidine could reduce PEDV proliferation in Vero cells [4].
NLuc is a novel engineered luciferase that is smaller and brighter than firefly or Renilla luciferase (about a 150-fold increase in luminescence). It utilizes the synthetic substrate furimazine to produce high-intensity, glow-type luminescence but low background activity [25]. Recently, Xie et al., constructed a recombinant SARS-CoV-2 expressing NLuc that can rapidly detect neutralization antibodies and screen antiviral drugs [43]. In this study, we employed the reverse genetics system to engineer a recombinant PEDV expressing the NLuc gene ( Figure 2). PEDV ORF3 is the only accessory ORF that encodes a putative ion channel protein [44]; however, this is dispensable for viral replication in vitro and in vivo [24]. During serial passaging, PEDV ORF3 usually becomes abortive due to deletion mutations [45][46][47]. In the cell-culture adapted YN150 strain, a nucleotide deletion within the ORF3 gene leads to early termination of ORF3 at 144 aa. Therefore, the ORF3 of PEDV YN150 is suitable for replacement with the NLuc gene. As expected, we found that the NLuc gene was stably incorporated into the genome of PEDV YN150 with no apparent effects on viral kinetics or plaque phenotype ( Figure 2). Bioluminescence signal produced during rPEDV-NLuc infection could accurately reflect the extent of viral replication ( Figure 3). The luciferase levels during rPEDV-NLuc infection showed an excellent linear relationship with the viral titers (Figure 3), indicating that bioluminescence signal produced during rPEDV-NLuc infection can be used to monitor viral replication in Vero cells.
Using the engineered rPEDV-NLuc, we performed HTS of an 803-compound library of natural products and identified 25 compounds exhibiting anti-PEDV activity ( Figure 4). Intriguingly, we discovered that 7 of the 25 identified inhibitors are antioxidants, including Betulonic acid, Ursonic acid, esculetin, lithocholic acid, nordihydroguaiaretic acid, caffeic acid, and grape seed extract ( Figure 4). All seven antioxidants were confirmed to inhibit wild-type PEDV replication in a dose-dependent manner in Vero cells ( Figure 5). Even at 2.5 µM concentration, all selected antioxidants showed a level of inhibition of at least 60% ( Figure 5). Previous studies demonstrated that infection of Vero cells with PEDV causes ROS accumulation in a time-dependent manner [38]. Here, we showed that the seven identified antioxidants could dramatically reduce PEDV-trigger ROS accumulation, and their inhibitory effect of ROS accumulation was more substantial than that of a wellknown ROS inhibitor N-acetyl-L-cysteine (NAC) ( Figure 6). NAC was found to limit the replication of influenza virus (IV) and reduce IV-induced apoptosis [48][49][50]. In our study, NAC also exhibited anti-PEDV activity in a dose-dependent manner ( Figure 5), suggesting that ROS production was involved in PEDV replication. More recently, Sun et al. reported that ROS accumulation played a critical role in ER stress and autophagy during PEDV infection [51]. In our previous study, PEDV-mediated autophagy facilitated viral replication [52]. We speculated that the seven hit antioxidants reduced PEDV-induced ROS accumulation, probably leading to the cells being defective in autophagy, thus impairing viral replication. Nevertheless, the precise mechanism by which antioxidants inhibit PEDV replication remains to be determined.
Besides the antioxidants, two hit compounds, protoporphyrin IX and 7-Ethylcamptothecin, exhibited inhibitory activities on the replication of PEDV with an SI of >80.7-Ethylcamptothecin was reported as an anti-cancer chemical with potent activity against various murine tumors via topoisomerase inhibition [53]. To our knowledge, there has been no report on the antiviral activity of 7-Ethylcamptothecin; however, it showed anti-PEDV activity with IC 50 of <1.25 µM. Notably, protoporphyrin IX can insert into the lipid vesicles of the vesicular stomatitis virus and impair the viral membrane organization, leading to the inhibition of viral replication [54]. Recently, Gu et al., reported that protoporphyrin IX could completely inhibit the cytopathic effect caused by SARS-CoV-2. They speculated protoporphyrin IX probably binds with the receptor ACE2 and blocks the S protein-mediated entry [55,56]. Whether protoporphyrin IX inhibits PEDV entry needs further exploration.
In conclusion, our study has established a powerful reverse genetics platform for the PEDV YN150 strain and engineered a reporter PEDV expressing the NLuc (rPEDV-NLuc). We found that the recombinant virus showed similar viral plaque size and growth kinetics to the parental virus in vitro. The bioluminescence signal produced during rPEDV-NLuc infection could accurately reflect the level of viral replication in Vero cells. Using the engineered virus, we performed the HTS of anti-PEDV compounds from a library of 803 natural products and successfully identified 25 compounds inhibiting the replication of PEDV. Our findings provide a powerful tool for the rapid screening of antiviral drugs and neutralization antibody detection and offer new and promising therapeutic strategies against PEDV infection.
Supplementary Materials: The following are available online at https://www.mdpi.com/article/10 .3390/v13091866/s1, Figure S1: Dose-response curves for 25 inhibitors of PEDV in vitro. Table S1: Primers used for viral genome sequencing in this study.